def getxy(event, x, y, flags, param):
if event == cv2.EVENT_LBUTTONDOWN:
contourList.append((y, x))
# cv2.drawContours( imageContour, contour, -1, ( 128,128,128 ) )
print "(row, col) = ", (y, x)
cv2.drawMarker(imageColor, (x, y), (0, 0, 0))
cv2.imshow('image', imageColor)
else:
if event == cv2.EVENT_RBUTTONDOWN:
contourPoints = np.array(contourList)
startSnake( contourPoints )
x, y, w, h = cv2.boundingRect(cnt)
# creates a rectangle around contour
cv2.rectangle(image, (x, y), (x + w, y + h),
(255, 0, 0), 2)
# Prints centroid text in order to double check later on
cv2.putText(image,
str(cx) + "," + str(cy),
(cx + 10, cy + 10),
cv2.FONT_HERSHEY_SIMPLEX, .3, (0, 0, 255),
1)
cv2.drawMarker(image, (cx, cy), (0, 0, 255),
cv2.MARKER_STAR,
markerSize=5,
thickness=1,
line_type=cv2.LINE_AA)
# adds centroids that passed previous criteria to centroid list
cxx[i] = cx
cyy[i] = cy
# eliminates zero entries (centroids that were not added)
cxx = cxx[cxx != 0]
cyy = cyy[cyy != 0]
# empty list to later check which centroid indices were added to dataframe
minx_index2 = []
miny_index2 = []
def gen():
"""Video streaming generator function."""
cap = cv2.VideoCapture('768x576.avi')
# Read until video is completed
while (cap.isOpened()):
ret, frame = cap.read() # import image
if not ret: #if vid finish repeat
frame = cv2.VideoCapture("768x576.avi")
continue
if ret: # if there is a frame continue with code
image = cv2.resize(frame, (0, 0), None, 1, 1) # resize image
gray = cv2.cvtColor(image,
cv2.COLOR_BGR2GRAY) # converts image to gray
fgmask = sub.apply(gray) # uses the background subtraction
kernel = cv2.getStructuringElement(
cv2.MORPH_ELLIPSE, (5, 5)) # kernel to apply to the morphology
closing = cv2.morphologyEx(fgmask, cv2.MORPH_CLOSE, kernel)
opening = cv2.morphologyEx(closing, cv2.MORPH_OPEN, kernel)
dilation = cv2.dilate(opening, kernel)
retvalbin, bins = cv2.threshold(
dilation, 220, 255, cv2.THRESH_BINARY) # removes the shadows
contours, hierarchy = cv2.findContours(dilation, cv2.RETR_EXTERNAL,
cv2.CHAIN_APPROX_SIMPLE)
minarea = 400
maxarea = 50000
for i in range(len(
contours)): # cycles through all contours in current frame
if hierarchy[
0, i,
3] == -1: # using hierarchy to only count parent contours (contours not within others)
area = cv2.contourArea(contours[i]) # area of contour
if minarea < area < maxarea: # area threshold for contour
# calculating centroids of contours
cnt = contours[i]
M = cv2.moments(cnt)
cx = int(M['m10'] / M['m00'])
cy = int(M['m01'] / M['m00'])
# gets bounding points of contour to create rectangle
# x,y is top left corner and w,h is width and height
x, y, w, h = cv2.boundingRect(cnt)
# creates a rectangle around contour
cv2.rectangle(image, (x, y), (x + w, y + h),
(0, 255, 0), 2)
# Prints centroid text in order to double check later on
cv2.putText(image,
str(cx) + "," + str(cy),
(cx + 10, cy + 10),
cv2.FONT_HERSHEY_SIMPLEX, .3, (0, 0, 255),
1)
cv2.drawMarker(image, (cx, cy), (0, 255, 255),
cv2.MARKER_CROSS,
markerSize=8,
thickness=3,
line_type=cv2.LINE_8)
#cv2.imshow("countours", image)
frame = cv2.imencode('.jpg', image)[1].tobytes()
yield (b'--frame\r\n'
b'Content-Type: image/jpeg\r\n\r\n' + frame + b'\r\n')
#time.sleep(0.1)
key = cv2.waitKey(20)
if key == 27:
break
def draw_vector(self, event, x, y, flags, params):
# self.show_img = self.img_copy()
values = None
x_list = []
y_list = []
z_list = []
if event == 10:
#sign of the flag shows direction of mousewheel
if flags > 0:
self.vector_depth += 10
else:
self.vector_depth -= 10
front_point = np.float32([[x, 0, y]])
front, jac = cv2.projectPoints(front_point, self.rotation_vector,
self.translation_vector,
self.camera_matrix,
self.distioriton_matrix)
back_point = np.float32([[x, self.room_length, y]])
back, jac = cv2.projectPoints(back_point, self.rotation_vector,
self.translation_vector,
self.camera_matrix,
self.distioriton_matrix)
floater_point = np.float32([[x, self.vector_depth, y]])
floater, jac = cv2.projectPoints(floater_point, self.rotation_vector,
self.translation_vector,
self.camera_matrix,
self.distioriton_matrix)
# print(x,y)
front_point = (front[0][0][0], front[0][0][1])
back_point = (back[0][0][0], back[0][0][1])
floater_point = (floater[0][0][0], floater[0][0][1])
self.show_img = cv2.circle(self.show_img, front_point, 8,
(125, 0, 125), 2, 0)
self.show_img = cv2.drawMarker(self.show_img,
back_point,
color=(255, 255, 0),
markerType=1)
self.show_img = cv2.line(self.show_img, front_point, back_point,
(100, 100, 0), 2)
self.show_img = cv2.circle(self.show_img, front_point, 8,
(255, 255, 0), 2, 0)
self.show_img = cv2.circle(self.show_img, floater_point, 8,
(255, 0, 0), 2, 0)
ifp = (int(front_point[0]), int(front_point[1]))
if (x, y) in self.mapped_dictionary.keys():
values = self.mapped_dictionary[(x, y)]
for v in values:
self.show_img = cv2.drawMarker(self.show_img, (x, y),
color=(0, 0, 255),
markerType=3)
x_list = []
y_list = []
z_list = []
for v in values:
x_list.append(v[0])
y_list.append(v[2])
z_list.append(v[1])
cv2.imshow(self.window_name, self.show_img)
cv2.waitKey(1)
if self.show_3D_plot == True:
plt.ion
if self.fig is None:
self.fig = plt.figure(1)
self.ax = ax = plt.axes(projection='3d')
self.front_plt = ax.scatter3D(front_point[0], front_point[1],
0)
self.back_plt = ax.scatter3D(back_point[0], back_point[1],
self.vector_depth)
self.float_plt = ax.scatter3D(floater_point[0],
floater_point[1],
self.vector_depth)
self.line_plt = ax.plot3D([front_point[0], back_point[0]],
[front_point[1], back_point[1]],
[0, self.room_length],
color='teal')
# plt.quiver(*origin, V[:,0], V[:,1], color=['r','b','g'], scale=21)
self.ax.set_xlim(-self.room_width, self.room_width)
self.ax.set_ylim(-self.room_height, self.room_width)
self.ax.set_zlim(-self.room_length, self.room_width)
self.fig.canvas.draw()
self.fig.canvas.flush_events()
self.fig.show()
else:
self.ax.cla()
if values:
self.points_plt = self.ax.scatter3D(x_list,
y_list,
z_list,
marker="D",
color="red")
# plt.quiver((0,0,0), (0,1,0), , color=['r','b','g'], scale=21)
self.ax.set_xlim(-self.room_width, self.room_width)
self.ax.set_ylim(-self.room_height, self.room_width)
self.ax.set_zlim(-self.room_length, self.room_width)
self.ax.set_xlabel("X")
self.ax.set_ylabel("Y")
self.ax.set_zlabel("Z")
self.float_plt = self.ax.scatter3D(floater_point[0],
floater_point[1],
self.vector_depth,
marker="o",
color="blue")
self.front_plt = self.ax.scatter3D(front_point[0],
front_point[1],
0,
marker=("o"),
color="cyan")
self.back_plt = self.ax.scatter3D(back_point[0],
back_point[1],
self.room_length,
marker="x",
color="cyan")
self.line_plt = self.ax.plot3D([front_point[0], back_point[0]],
[front_point[1], back_point[1]],
[0, self.room_length],
color='teal')
self.quiver = self.ax.quiver([0, 0, 0], [0, 0, 0], [0, 0, 0],
[self.room_width, 0, 0],
[0, self.room_height, 0],
[0, 0, self.room_length],
length=0.1,
normalize=False)
self.fig.canvas.draw()
self.fig.canvas.flush_events()
return self.show_img
def find_cube_side(self):
cam_idx = [1, 2]
# tracking_points_all = np.array([self.views[j].find_right_top_bottom_tracking_points() for j in cam_idx])
tracking_points_all = np.array([[[137, 55], [137, 100], [102, 127],
[71, 104], [60, 58], [96, 71],
[103, 47], [137, 79], [98, 99]],
[[134, 61], [133, 95], [125, 113],
[86, 109], [81, 68], [125, 70],
[101, 58], [134, 79], [125, 91]]])
i = 0
colors = [(255, 0, 0), (0, 255, 0), (0, 0, 255), (0, 0, 0),
(255, 255, 255), (255, 255, 0), (255, 0, 255), (0, 255, 255),
(128, 128, 255)]
ims = []
for j in cam_idx:
im = self.views[j].im.copy()
for k in range(tracking_points_all[i].shape[0]):
x, y = tracking_points_all[i][k].astype(np.int)
cv2.drawMarker(im, (int(x), int(y)), colors[k], markerSize=5)
cv2.putText(im,
str(k), (int(x), int(y)),
cv2.FONT_HERSHEY_SIMPLEX,
.5,
colors[k],
thickness=1)
i += 1
ims.append(im)
# cv2.imwrite(str(j) + '.png', im)
mapped_xyzs = []
mapped_enus = []
errs = []
for i in range(tracking_points_all.shape[1]):
xyz, enu, err = self.extract_3d_points(
tracking_points_all[:,
i], [self.views[j].cam for j in cam_idx])
mapped_xyzs.append(xyz)
mapped_enus.append(enu)
errs.append(err)
for k in range(len(cam_idx)):
im_ = self.views[cam_idx[k]].cam.viz_plane(200, 1)
im = (.5 * ims[k] + .5 * im_).astype(np.uint8)
cv2.imwrite(str(cam_idx[k]) + '-en_overlay.png', im)
mapped_xyzs = np.array(mapped_xyzs)
mapped_enus = np.array(mapped_enus)
side_lens = np.linalg.norm(mapped_xyzs[1:] - mapped_xyzs[:-1], axis=1)
side_len = np.mean(side_lens[[0, 1, 2, 3, 4, 7]])
side_len_uncertainity = side_lens.std()
top_center_enu_estimates = np.vstack([
mapped_enus[[4, 0]].mean(axis=0), mapped_enus[[5, 6]].mean(axis=0)
])
top_center_enu = top_center_enu_estimates.mean(axis=0)
bottom_center_enu = mapped_enus[[1, 3]].mean(axis=0)
top_bottom_centroid_enu = np.mean([top_center_enu, bottom_center_enu],
axis=0)
corner_centroid_enu_estimates = np.array([
np.mean(mapped_enus[[0, 3]], axis=0),
np.mean(mapped_enus[[2, 6]], axis=0),
np.mean(mapped_enus[[1, 4]], axis=0)
])
centroid_enu = np.vstack(
[top_bottom_centroid_enu,
corner_centroid_enu_estimates]).mean(axis=0)
corner_vertices = mapped_enus[:7]
centroid_corner_dists = np.linalg.norm(corner_vertices - centroid_enu,
axis=1)
centroid_estimation_uncertainity = centroid_corner_dists.std()
return side_len, centroid_enu, corner_vertices, mapped_enus,\
centroid_estimation_uncertainity, side_len_uncertainity
def _state_follow(self):
"""
Spot will position its body to follow the light
"""
from simple_pid import PID
try:
# Get an image from the robot
img = self._robot.get_image()
# Find the center of the image
img_height, img_width, _ = img.shape
img_center_x = img_width // 2
img_center_y = img_height // 2
# The ROI is at the center of the image with half the width and height
# The following crops the corner where is the plastic reflects the cmd light
x1 = 0
x2 = int(0.8 * img_width)
y1 = 0
y2 = int(0.8 * img_height)
# Crop it to the ROI
roi = img[y1:y2, x1:x2]
# Detect a blob
light_found, img_with_keypoints, pt = self._detect_blob(roi, 3500)
# Check to see if we see a blob
if not light_found:
# If we have an initial point that mean we were following
# and now the light is gone, so we'll stop
if self._init_pt is not None and self._light_count > 50:
self._state_idx = self._state_idx + 1
self._init_pt = None
self._light_count = 0
self._light_count = self._light_count + 1
cv2.imshow('', roi)
cv2.waitKey(5)
return
# Save the initial point if we don't have one yet
if self._init_pt is None:
self._init_pt = pt
# Mark the 2 points for visualization
cv2.drawMarker(img_with_keypoints, tuple(map(int, pt)),
(255, 0, 0))
cv2.drawMarker(img_with_keypoints, tuple(map(int, self._init_pt)),
(0, 255, 0))
# Maximum angle range is +/- 0.5 [rad]
MAX_ANGLE = 0.50
self._pitch_pid.output_limits = (-MAX_ANGLE, MAX_ANGLE)
# Figure out how much to yaw and pitch
adj = numpy.divide(numpy.subtract(self._init_pt, pt),
self._init_pt) * MAX_ANGLE
yaw_cmd = self._yaw_pid(-adj[0])
pitch_cmd = self._pitch_pid(adj[1])
# Add text to show adjustment value
font = cv2.FONT_HERSHEY_SIMPLEX
bottom_left = (25, 25)
scale = 0.5
color = (0, 255, 0)
lineType = 1
disp_text = 'Yaw: {} Pitch: {}'.format(adj[0], pitch_cmd)
cv2.putText(img_with_keypoints, disp_text, bottom_left, font,
scale, color, lineType)
cv2.imshow('', img_with_keypoints)
cv2.waitKey(5)
self._robot.orient(yaw_cmd, pitch_cmd, 0)
test = 0
except TypeError as err:
print(err)
def FAST(argv=sys.argv[1:]):
img = cv2.imread(argv[1])
height, width, channels = img.shape
height = int(height / 2)
width = int(width / 2)
dim = (width, height)
img = cv2.resize(img, dim, interpolation=cv2.INTER_LINEAR)
gray = cv2.cvtColor(img, cv2.COLOR_RGB2GRAY)
cv2.GaussianBlur(gray, (3, 3), 0, gray)
result = img.copy()
keypoints = {}
filtered_keypoints = []
keypoints_score = np.zeros(shape=[0, width], dtype=np.int)
for i in range(radius, height - radius):
score_line = np.zeros(shape=[1, width], dtype=np.int)
keypoints_line = np.empty(shape=[0, 2], dtype=np.int)
for j in range(radius, width - radius):
center = gray[i, j]
a = 0
pixel_diff = np.int16(gray[i + fast_index[0][0],
j + fast_index[0][1]]) - center
a += 1 if math.fabs(pixel_diff) > pixel_t else 0
pixel_diff = np.int16(gray[i + fast_index[8][0],
j + fast_index[8][1]]) - center
a += 1 if math.fabs(pixel_diff) > pixel_t else 0
if a < 2:
continue
pixel_diff = np.int16(gray[i + fast_index[4][0],
j + fast_index[4][1]]) - center
a += 1 if math.fabs(pixel_diff) > pixel_t else 0
pixel_diff = np.int16(gray[i + fast_index[11][0],
j + fast_index[11][1]]) - center
a += 1 if math.fabs(pixel_diff) > pixel_t else 0
if a < 3:
continue
same_pixel_index = []
for k in range(len(fast_index)):
pixel_diff = np.int16(gray[i + fast_index[k][0],
j + fast_index[k][1]]) - center
if math.fabs(pixel_diff) <= pixel_t:
same_pixel_index.append(k)
if len(same_pixel_index) > 1:
for l in range(len(same_pixel_index)):
dist = 0
if l >= 1:
dist = same_pixel_index[l] - same_pixel_index[l - 1]
else:
dist = 16 - (same_pixel_index[len(same_pixel_index) -
1] - same_pixel_index[l])
if math.fabs(dist) >= threshold_num:
# print("Found key point")
# cv2.drawMarker(result, (j,i), (255, 0, 0),
# markerType=cv2.MARKER_SQUARE,
# markerSize=6, thickness=1)
score = corner_score((i, j), gray)
score_line[0, j] = score
keypoints_line = np.append(keypoints_line, [[i, j]],
axis=0)
# keypoints = np.sort(keypoints, axis=0, kind='quicksort')
break
else:
score = corner_score((i, j), gray)
score_line[0, j] = score
keypoints_line = np.append(keypoints_line, [[i, j]], axis=0)
keypoints_score = np.append(keypoints_score, score_line, axis=0)
keypoints[i] = keypoints_line
# 不够non-maximal suppression计算
if np.size(keypoints_score, axis=0) < 3:
continue
prev_idx = i - 2
curr_idx = i - 1
next_idx = i
try:
for k in range(len(keypoints[curr_idx])):
index = keypoints[curr_idx][k]
score = keypoints_score[curr_idx - 3, index[1]]
if score > keypoints_score[prev_idx-3, index[1] - 1] \
and score > keypoints_score[prev_idx-3, index[1]] \
and score > keypoints_score[prev_idx-3, index[1] + 1] \
and score > keypoints_score[curr_idx-3, index[1] - 1] \
and score > keypoints_score[curr_idx-3, index[1] + 1] \
and score > keypoints_score[next_idx-3, index[1] - 1] \
and score > keypoints_score[next_idx-3, index[1]] \
and score > keypoints_score[next_idx-3, index[1] + 1]:
filtered_keypoints.append((index[1], index[0]))
except KeyError:
continue
for i in range(len(filtered_keypoints)):
cv2.drawMarker(result,
filtered_keypoints[i], (255, 0, 0),
markerType=cv2.MARKER_SQUARE,
markerSize=6,
thickness=1)
cv2.imshow("FAST result", result)
# Initiate FAST object with default values
fast = cv2.FastFeatureDetector_create()
# find and draw the keypoints
kp = fast.detect(img, None)
img2 = cv2.drawKeypoints(img, kp, None, color=(255, 0, 0))
cv2.imshow("FAST opencv result", img2)
while (1):
if cv2.waitKey(20) & 0xFF == 27:
break
cv2.destroyAllWindows()
"""
特征提取之关键点检测(GFTTDetector)
"""
import cv2 as cv
image = cv.imread("images/test4.jpg")
cv.imshow("input", image)
# 创建GFTT特征检测器
gftt = cv.GFTTDetector_create(100, 0.01, 1, 3, False, 0.04)
kp1 = gftt.detect(image, None)
for marker in kp1:
result = cv.drawMarker(image,
tuple(int(i) for i in marker.pt),
color=(0, 255, 0))
cv.imshow("GFTT-Keypoint-Detect", result)
cv.waitKey(0)
cv.destroyAllWindows()
def update(self):
ret, img = self.cap.read()
preprocessed = None
if color_keying:
hsv = cv2.cvtColor(img, cv2.COLOR_BGR2HSV)
mask = cv2.inRange(hsv, (4, 15, 100), (70, 170, 255))
hsv = cv2.cvtColor(img, cv2.COLOR_BGR2HSV)
mask2 = cv2.inRange(hsv, (140, 0, 50), (165, 160, 255))
if show_detailed:
cv2.imshow("1", mask)
cv2.imshow("2", mask2)
imask = mask + mask2 > 0
skin = np.zeros_like(img, np.uint8)
skin[imask] = img[imask]
preprocessed = skin
else:
preprocessed = img
fg_mask = self.fgbg.apply(preprocessed)
whites_count = cv2.countNonZero(fg_mask)
blur = cv2.GaussianBlur(fg_mask, (11, 11), 0)
_, gaussian = cv2.threshold(blur, 70, 255, cv2.THRESH_BINARY + cv2.THRESH_OTSU)
if show_detailed:
cv2.imshow("gaussian", gaussian)
contours, hierarchy = cv2.findContours(fg_mask, cv2.RETR_TREE, cv2.CHAIN_APPROX_SIMPLE)
max_area = 0
ci = None
for i in range(len(contours)):
cnt = contours[i]
area = cv2.contourArea(cnt)
if area > max_area:
max_area = area
ci = i
if ci is not None:
cnt = contours[ci]
hull = cv2.convexHull(cnt)
x, y, w, h = cv2.boundingRect(cnt)
cropped_gaussian = gaussian[y:y + h, x:x + w]
if show_detailed:
cv2.imshow("crop", cropped_gaussian)
drawing = np.zeros(img.shape, np.uint8)
if whites_count > minimal_count:
cv2.drawContours(drawing, [cnt], 0, (0, 255, 0), 2)
cv2.drawContours(drawing, [hull], 0, (0, 0, 255), 2)
cv2.drawContours(img, [cnt], 0, (0, 255, 0), 2)
cv2.drawContours(img, [hull], 0, (0, 0, 255), 2)
if DEBUG:
cv2.drawContours(img, [surfer_thing_shape_hull.shape[0]], 0, (255, 0, 255), 2)
x_previous = self.x_center
y_previous = self.y_center
self.x_center = 0
self.y_center = 0
for i in hull:
self.x_center += i[0][0]
self.y_center += i[0][1]
self.x_center /= len(hull)
self.y_center /= len(hull)
if x_previous is not None and y_previous is not None and whites_count > minimal_count:
x_delta = (self.x_center - x_previous)
y_delta = (self.y_center - y_previous)
else:
for s in shapes:
s.reset_count()
x_delta = 0
y_delta = 0
if whites_count < minimal_count:
for s in shapes:
s.reset_count()
self.x_counter = 0
self.y_counter = 0
else:
self.x_counter += x_delta
self.y_counter += y_delta
hull = cv2.convexHull(cnt)
if whites_count > minimal_count and x_previous is not None and y_previous is not None:
cv2.drawMarker(drawing, (int(self.x_center), int(self.y_center)), (255, 0, 255))
cv2.drawMarker(drawing, (int(x_previous), int(y_previous)), (255, 255, 255))
cv2.drawMarker(img, (int(self.x_center), int(self.y_center)), (255, 0, 255))
cv2.drawMarker(img, (int(x_previous), int(y_previous)), (255, 255, 255))
contour_gaussian_cropped = np.matrix(list(map(lambda a: [a[0][0] - x, a[0][1] - y], cnt)))
self.last_contour = cv2.convexHull(contour_gaussian_cropped)
if use_hull:
contour_gaussian_cropped = cv2.convexHull(contour_gaussian_cropped)
m = max(list(map(lambda x: x[0][1], contour_gaussian_cropped)))
c = list(filter(lambda y: abs(y[0][1]-m) > 40, contour_gaussian_cropped))
c = list(map(lambda x: list(x[0]), c))
c = np.array(c)
contour_gaussian_cropped = c
if DEBUG and c.shape[0] != 0:
cv2.drawContours(img, [contour_gaussian_cropped], 0, (255, 0, 255), 2)
rets = [x.bestMatch(contour_gaussian_cropped) for x in shapes]
m_ret = 1
m_sh = None
message = None
for i, sh in zip(rets, shapes):
if m_ret > i:
m_ret = i
m_sh = sh
if whites_count > minimal_count and m_ret < 0.15:
m_sh.inc_count()
if self.x_counter > 400:
tmp_m_sh = None
tmp_m_count = -1
for s in shapes:
if tmp_m_count < s.count:
tmp_m_count = s.count
tmp_m_sh = s
if tmp_m_sh.count > 5:
message = tmp_m_sh.msg + ' left ' + str(tmp_m_sh.count)
if self.x_counter < -400:
tmp_m_sh = None
tmp_m_count = -1
for s in shapes:
if tmp_m_count < s.count:
tmp_m_count = s.count
tmp_m_sh = s
if tmp_m_sh.count > 5:
message = tmp_m_sh.msg + ' right ' + str(tmp_m_sh.count)
if color_keying:
cv2.imshow('skin', skin)
if show_detailed:
cv2.imshow('fg_mask', fg_mask)
cv2.imshow('drawing', drawing)
return cv2.cvtColor(img, cv2.COLOR_BGR2RGB), message
def draw_corners(img, corners, ids):
for corner, id in zip(corners, ids):
x, y = corner
cv2.drawMarker(img, (x, y), (255, 0, 0), cv2.MARKER_CROSS, 10, 2)
cv2.putText(img, str(id), (x, y), cv2.FONT_HERSHEY_PLAIN, 2,
(255, 0, 0), 2)
def process_video(self):
global j
j = 1
cap = cv2.VideoCapture(self.link.text())
kernel = cv2.getStructuringElement(cv2.MORPH_ELLIPSE, (3, 3))
kernel2 = cv2.getStructuringElement(cv2.MORPH_ELLIPSE, (3, 3))
kernel3 = cv2.getStructuringElement(cv2.MORPH_ELLIPSE, (10, 10))
substractor = cv2.createBackgroundSubtractorMOG2(varThreshold=150,
detectShadows=False)
self.sayac = 0
frame_width = int(cap.get(3))
frame_height = int(cap.get(4))
# save as .264
fourcc = cv2.VideoWriter_fourcc('H', '2', '6', '4')
kayit = cv2.VideoWriter('kayit.264', fourcc, 10,
(frame_width, frame_height))
while True:
_, frame = cap.read()
self.exit = 1
mask = substractor.apply(frame)
mask = cv2.morphologyEx(mask, cv2.MORPH_CLOSE, kernel)
mask = cv2.erode(mask, kernel2, iterations=1)
mask = cv2.dilate(mask, kernel3, iterations=1)
contours, _ = cv2.findContours(mask, cv2.RETR_EXTERNAL,
cv2.CHAIN_APPROX_NONE)
# cv2.drawContours(frame, contours, -1, (0,255,0), 3)
cv2.line(frame, (self.yatay.value(), self.dikey.value()),
(self.yatay2.value(), self.dikey2.value()), (0, 255, 0),
2)
m = self.egimofline()
if 1 >= m >= 0:
self.detectlinedikey = int(
(self.dikey2.value() + self.dikey.value()) / 2)
for contour, in zip(contours):
(x, y, w, h) = cv2.boundingRect(contour)
a = int(w / 2)
b = int(h / 2)
if w > 20 and h > 20:
cv2.rectangle(frame, (x, y), (x + w, y + h),
(255, 0, 0), 2)
centroid = (x + a, y + b)
cv2.drawMarker(frame,
centroid, (0, 255, 0),
markerType=50,
markerSize=10)
if self.yatay.value() < x + a<self.yatay2.value() and self.detectlinedikey-2 < y + b < \
self.detectlinedikey + 2:
self.sayac += 1
self.saveJson()
print(self.sayac)
elif m > 1:
self.detectlineyatay = int(
(self.yatay2.value() + self.yatay.value()) / 2)
for contour, in zip(contours):
(x, y, w, h) = cv2.boundingRect(contour)
a = int(w / 2)
b = int(h / 2)
if w > 20 and h > 20:
cv2.rectangle(frame, (x, y), (x + w, y + h),
(255, 0, 0), 2)
centroid = (x + a, y + b)
cv2.drawMarker(frame,
centroid, (0, 255, 0),
markerType=50,
markerSize=10)
if self.detectlineyatay-2 < x + a < self.detectlineyatay + 2 and self.dikey2.value() < y + b < \
self.dikey.value():
self.sayac += 1
self.saveJson()
print(self.sayac)
else:
break
"""for contour, in zip(contours):
(x, y, w, h) = cv2.boundingRect(contour)
a = int(w / 2)
b = int(h / 2)
if w > 20 and h > 20 and y + b > 70:
cv2.rectangle(frame, (x, y), (x + w, y + h), (255, 0, 0), 2)
centroid = (x + a, y + b)
cv2.drawMarker(frame, centroid, (0, 255, 0), markerType=50, markerSize=10)
if self.yatay.value() < x + a and self.dikey.value() < y + b:
self.sayac += 1
self.saveJson()
print(self.sayac)"""
cv2.putText(frame, "Vehicle: " + str(self.sayac), (50, 45),
cv2.FONT_HERSHEY_SIMPLEX, 0.5, (0, 250, 0), 2)
self.t = int(self.nmbrofcars.text())
if (self.sayac == self.t) & (j == 1):
cv2.imwrite('resim3.jpg', frame)
self.sendMail()
j = 0
kayit.write(frame)
# write as .bin
# with open('file1.bin', 'ab') as f:
# f.write(frame)
# self.save264()
cv2.imshow("Frame", frame)
cv2.imshow("mask", mask)
key = cv2.waitKey(25)
if key == 27:
break
cap.release()
kayit.release()
cv2.destroyAllWindows()
def main():
im = cv.imread("flag.png")
gray = cv.cvtColor(im, cv.COLOR_RGB2GRAY)
_, thresh = cv.threshold(gray, 127, 255, 0)
# Now find contours using the chain approx none parameter
_, contours, _h = cv.findContours(thresh, cv.RETR_TREE,
cv.CHAIN_APPROX_NONE)
cv.drawContours(im, contours[0], -1, (0, 255, 0), 2)
print("Contours: ", contours)
print("Contours count: ", len(contours))
# [Explanation]
# So the find contorus function works so that the number of areas we have
# that are contoured seperately, the points for that particular box/shape
# is stored as a different element on the contours list
# [Working Example] This example works for the flag.png image
# print(contours[0][0])
# print(contours[1][0])
# print(contours[2])
# Lets determine the moments of the given image using the given contours
# [Working Example] This example works for the flag.png
# I have arbitrarily chosen the 0th contour and will be working with that for this example
star = contours[0]
moments = cv.moments(
star
) #its important to note that moments are a dictionary that store the value for m00 and m11 and so on till a third degree
# [Explanation] I would also like to mention the use of moments and the intuition behind there use here
# So a moment of a 2d function would be, intuitively speaking
# For m10,we can image that as we go across a particular row of the image
# the value for our x will keep increase i.e. the idx of our row
# As we find increments there the weight assigned to the function impulse at
# that point is greater hence the moment, considering the pivot at the 0th idx
# is also much greater as we move vertically across the image.
# What this does for m10 for example is it finds a summation of
# the complete row weighted in a linear fashion with farther impulses getting weighted a linearly larger value
# We also get a larger value exactly for this reason
# Im also wondering if the contours have to be closed in any way. For that i should
# probably research how that chain_approximation works
# Calculating centroids and displaying area for all contours
# Calculating the moment for all contours
all_moments = []
for c in contours:
all_moments.append(cv.moments(c))
print(all_moments)
mom1 = all_moments[0]
centroidx = mom1['m10'] / mom1['m00']
centroidy = mom1['m01'] / mom1['m00']
cv.drawMarker(im, (int(centroidx), int(centroidy)), (255, 100, 0),
cv.MARKER_STAR, 3)
# Lets also try to bound the star in a bounding box using its contours
x, y, w, h = cv.boundingRect(contours[0])
cv.rectangle(im, (x, y), (x + w, y + h), (0, 0, 255), 5)
box = cv.minAreaRect(contours[0])
box_p = cv.boxPoints(box)
box_p = np.int0(box_p)
cv.drawContours(im, [box_p], 0, (0, 255, 255), 2)
# Lets also label the smallest contour area
# The area is simply m00
cv.namedWindow('original', cv.WINDOW_AUTOSIZE)
cv.moveWindow('original', 0, 0)
cv.imshow('original', im)
cv.waitKey(0)
cv.destroyAllWindows()
return
def image_scanning(imageObject, imageScene, detectionBoxes, heightImageObject,
widthImageObject, heightOverlap, widthOverlap):
heightImageScene, widthImageScene = imageScene.shape[:2]
heightSteps = int(
math.floor(heightImageScene / (heightImageObject - heightOverlap)))
widthSteps = int(
math.floor(widthImageScene / (widthImageObject - widthOverlap)))
print('Horizontal Steps: %d' % (widthSteps))
print('Vertical Steps: %d' % (heightSteps))
colorMap = np.ones((heightImageScene, widthImageScene), np.uint8) * 255
colorMap = cv2.cvtColor(colorMap, cv2.COLOR_GRAY2BGR)
for heightStep in range(0, heightSteps - 1):
for widthStep in range(0, widthSteps - 1):
iniHeightPosition = int(
round((heightImageObject - heightOverlap)) * heightStep)
iniWidthPosition = int(
round((widthImageObject - widthOverlap)) * widthStep)
endHeightPosition = iniHeightPosition + heightImageObject
endWidthPosition = iniWidthPosition + widthImageObject
if iniHeightPosition < endHeightPosition - 1 and iniWidthPosition < endWidthPosition - 1:
subImageFromScene = imageScene[
iniHeightPosition:endHeightPosition,
iniWidthPosition:endWidthPosition]
rfactor = 1
keyPointsImageObject, descriptorsImageObject = detector.detectAndCompute(
imageObject, None)
keyPointsImageScene, descriptorsImageScene = detector.detectAndCompute(
subImageFromScene, None)
MIN_KEYPOINTS = 12
if len(keyPointsImageScene) > MIN_KEYPOINTS:
# Show object image key points
vis = np.zeros((heightImageObject, widthImageObject),
np.uint8)
vis[:heightImageObject, :widthImageObject] = imageObject
kp_color = (250, 0, 0)
for marker in keyPointsImageObject:
vis = cv2.drawMarker(vis,
tuple(int(i) for i in marker.pt),
color=kp_color)
cv2.imshow('key points', vis)
# Do matching and draw resuts for this image
match_and_draw(colorMap, detectionBoxes,
keyPointsImageObject, keyPointsImageScene,
descriptorsImageObject,
descriptorsImageScene, iniHeightPosition,
iniWidthPosition)
cv2.waitKey(10)
return colorMap
idx = np.argmax(probs)
prob = probs[idx]
bbox = candRects[
idx] #take the region with highest probability to be a phone
else:
bbox = None
if len(candImgs) == 0 or bbox is None:
print(basefile + ' Not found!')
cv2.imshow(basefile, tempImg)
else:
cx = bbox[0] + bbox[2] / 2.0
cy = bbox[1] + bbox[3] / 2.0
cv2.drawMarker(tempImg, (int(cx), int(cy)), (255, 0, 0), 1, 10,
2)
#normalized to the shape
cx = cx / w
cy = cy / h
distance = np.sqrt((cx - x)**2 + (cy - y)**2)
if distance > 0.05:
cv2.imshow(basefile, tempImg)
print(basefile, ' too far by ', distance)
elif isinstance(check_type, list):
for i in check_type:
fn = os.path.join(path, i)
basefile = i
x = test_set[test_set['imgID'] == i]['x'].values[0]
def draw_clusters(image, image_name, radius, boxes, labels_boxes):
colors = []
for i in range(len(boxes) + 100):
random_col = (random.randint(0, 255), random.randint(0, 255),
random.randint(0, 255))
colors.append(random_col)
red = (0, 0, 255)
black = (0, 0, 0)
count = 0
offset_x = -25
offset_y = -10
for idxx, b in enumerate(boxes):
idx = labels_boxes[idxx]
col = colors[idx + 1]
if not WEIGHTED_DISTANCE_CLUSTERING:
image = cv2.circle(image,
tuple(b.x0),
radius,
color=col,
thickness=3)
image = cv2.circle(image,
tuple(b.x1),
radius,
color=col,
thickness=3)
image = cv2.circle(image,
tuple(b.x2),
radius,
color=col,
thickness=3)
image = cv2.circle(image,
tuple(b.x3),
radius,
color=col,
thickness=3)
else:
image = cv2.ellipse(image,
tuple(b.x0), (radius, int(radius / Y_WEIGHT)),
angle=0,
startAngle=0,
endAngle=360,
color=col,
thickness=6)
image = cv2.ellipse(image,
tuple(b.x1), (radius, int(radius / Y_WEIGHT)),
angle=0,
startAngle=0,
endAngle=360,
color=col,
thickness=6)
image = cv2.ellipse(image,
tuple(b.x2), (radius, int(radius / Y_WEIGHT)),
angle=0,
startAngle=0,
endAngle=360,
color=col,
thickness=6)
image = cv2.ellipse(image,
tuple(b.x3), (radius, int(radius / Y_WEIGHT)),
angle=0,
startAngle=0,
endAngle=360,
color=col,
thickness=6)
# image = cv2.putText(image, text=str(labels_boxes[idxx]), org=tuple(b.x0), fontFace=cv2.FONT_HERSHEY_SIMPLEX, fontScale=0.6, color=(36, 255, 12), thickness=2)
# image = cv2.putText(image, text=str(labels_boxes[idxx]), org=tuple(b.x1), fontFace=cv2.FONT_HERSHEY_SIMPLEX, fontScale=0.6, color=(36, 255, 12), thickness=2)
# image = cv2.putText(image, text=str(labels_boxes[idxx]), org=tuple(b.x2), fontFace=cv2.FONT_HERSHEY_SIMPLEX, fontScale=0.6, color=(36, 255, 12), thickness=2)
# image = cv2.putText(image, text=str(labels_boxes[idxx]), org=tuple(b.x3), fontFace=cv2.FONT_HERSHEY_SIMPLEX, fontScale=0.6, color=(36, 255, 12), thickness=2)
# image = cv2.putText(image, text=str(count), org=tuple([b.x0[0]+offset_x, b.x0[1]+offset_y]), fontFace=cv2.FONT_HERSHEY_SIMPLEX, fontScale=0.6, color=(0, 0, 0), thickness=2)
# count += 1
# image = cv2.putText(image, text=str(count), org=tuple([b.x1[0]+offset_x, b.x1[1]+offset_y]), fontFace=cv2.FONT_HERSHEY_SIMPLEX, fontScale=0.6,color=(0, 0, 0), thickness=2)
# count += 1
# image = cv2.putText(image, text=str(count), org=tuple([b.x2[0]+offset_x, b.x2[1]+offset_y]), fontFace=cv2.FONT_HERSHEY_SIMPLEX, fontScale=0.6,color=(0, 0, 0), thickness=2)
# count += 1
# image = cv2.putText(image, text=str(count), org=tuple([b.x3[0]+offset_x, b.x3[1]+offset_y]), fontFace=cv2.FONT_HERSHEY_SIMPLEX, fontScale=0.6,color=(0, 0, 0), thickness=2)
# count += 1
for id_1, box_1 in enumerate(boxes):
idx = labels_boxes[id_1]
col = colors[idx + 1]
pts = np.array([box_1.x0, box_1.x1, box_1.x2, box_1.x3])
pts = pts.reshape((-1, 1, 2))
image = cv2.polylines(image, [pts], True, color=col, thickness=7)
image = cv2.drawMarker(image,
tuple(box_1.x0),
color=red,
thickness=9,
markerType=cv2.MARKER_CROSS)
image = cv2.drawMarker(image,
tuple(box_1.x1),
color=red,
thickness=9,
markerType=cv2.MARKER_CROSS)
image = cv2.drawMarker(image,
tuple(box_1.x2),
color=red,
thickness=9,
markerType=cv2.MARKER_CROSS)
image = cv2.drawMarker(image,
tuple(box_1.x3),
color=red,
thickness=9,
markerType=cv2.MARKER_CROSS)
cv2.imwrite(CLUSTERED_PATH + image_name, image)
if sub_tracker:
sub_boundingbox, sub_loc = sub_tracker.update(frame)
sub_boundingbox = list(map(int, sub_boundingbox))
# 이탈 정도(0.25), motion_tracker.interval, waitKey(x) 조정 필요
if kcf_tracker.peak_value < 0.1: # 0.25:
print('[KCF] Disabled: peak value({:.02f}) is too low'.format(kcf_tracker.peak_value))
kcf_tracker.enable = False
zoom.zoom_to(1)
else:
kcf_tracker.update_location(boundingbox)
# str = "{}x{}({}x{})".format(kcf_tracker.mean_width, kcf_tracker.mean_height, selected_width, selected_height)
# util.draw_str(frame_draw, (20, 20), str)
cv2.rectangle(frame_draw,(kcf_tracker.x1,kcf_tracker.y1), (kcf_tracker.x2,kcf_tracker.y2), (0,255,0), 1)
cv2.drawMarker(frame_draw, tuple(kcf_tracker.center), (0,255,0))
request_to_track_flag = True
# if something happened...
# if test_flag is True:
# test_flag = False
if sub_tracker:
sub_tracker.update_location(sub_boundingbox)
cv2.rectangle(frame_draw,(sub_tracker.x1,sub_tracker.y1), (sub_tracker.x2,sub_tracker.y2), (255,255,0), 1)
cv2.drawMarker(frame_draw, tuple(sub_tracker.center), (255,255,0))
if (kcf_tracker.mean_area > sub_tracker.mean_area / 0.49) or (kcf_tracker.mean_area < sub_tracker.mean_area / 0.81):
resized_window = util.resize_selection({'x1': sub_tracker.x1, 'y1': sub_tracker.y1, 'x2': sub_tracker.x2, 'y2': sub_tracker.y2}, 1/0.8)
kcf_tracker.x1 = resized_window['x1']
kcf_tracker.y1 = resized_window['y1']
kcf_tracker.x2 = resized_window['x2']
kcf_tracker.y2 = resized_window['y2']
def draw_eyes(img_bgr,
eyes_landmarks,
radius,
offsets_list,
is_new=True,
is_show=False,
save_name=None):
"""
绘制图像
"""
import cv2
import copy
import numpy as np
if is_new:
img_bgr = copy.deepcopy(img_bgr)
th = 1
eye_upscale = 1
img_bgr = cv2.resize(img_bgr, (0, 0), fx=eye_upscale, fy=eye_upscale)
for el, r, offsets in zip(eyes_landmarks, radius, offsets_list):
start_x, start_y, offset_scale = offsets
real_el = el * offset_scale + [start_x, start_y]
real_radius = r * offset_scale
# 眼睛
cv2.polylines(
img_bgr,
[
np.round(eye_upscale * real_el[0:8]).astype(np.int32).reshape(
-1, 1, 2)
],
isClosed=True,
color=(255, 255, 0),
thickness=th,
lineType=cv2.LINE_AA,
)
# 眼球
cv2.polylines(
img_bgr,
[
np.round(eye_upscale * real_el[8:16]).astype(np.int32).reshape(
-1, 1, 2)
],
isClosed=True,
color=(0, 255, 255),
thickness=th,
lineType=cv2.LINE_AA,
)
iris_center = real_el[16]
eye_center = real_el[17]
eye_center = (real_el[0] + real_el[4]) / 2
# 虹膜中心
cv2.drawMarker(
img_bgr,
tuple(np.round(eye_upscale * iris_center).astype(np.int32)),
color=(255, 0, 255),
markerType=cv2.MARKER_CROSS,
markerSize=4,
thickness=th + 1,
line_type=cv2.LINE_AA,
)
# 眼睑中心
cv2.drawMarker(
img_bgr,
tuple(np.round(eye_upscale * eye_center).astype(np.int32)),
color=(0, 0, 255),
markerType=cv2.MARKER_CROSS,
markerSize=4,
thickness=th + 1,
line_type=cv2.LINE_AA,
)
cv2.circle(img_bgr,
center=tuple(eye_center),
radius=real_radius,
color=(0, 0, 255))
if is_show:
show_img_bgr(img_bgr, save_name=save_name) # 显示眼睛
return img_bgr
import cv2
import numpy as np
img = cv2.imread("lenna_full.jpg")
center = (img.shape[1] // 2, img.shape[0] // 2)
img = cv2.drawMarker(img, center, color=(255, 0, 255))
cv2.imshow("lenna_full.jpg", img)
cv2.waitKey(0)
cv2.destroyAllWindows()
def _monitor(self):
tic = time.time()
gray = cv2.cvtColor(self.frame, cv2.COLOR_BGR2GRAY)
self._updateThreshold()
_, thresholded = cv2.threshold(gray, self.THRESH, 255,
cv2.THRESH_TOZERO)
# _, contours, _ = cv2.findContours(thresholded, cv2.RETR_EXTERNAL, cv2.CHAIN_APPROX_SIMPLE)
contours, _ = cv2.findContours(thresholded, cv2.RETR_EXTERNAL,
cv2.CHAIN_APPROX_NONE)
contours = contours[:2]
# if contours.__len__() > 2:
# QMessageBox.warning(self, "Thresholding error", "More than 2 projections were found. " + \
# "Please increase threshold manually or select a better noise area.")
cv2.drawContours(self.draw_only_frame, contours, -1, (0, 255, 0), 2)
self._draw_noiseArea()
try:
moments_star_1 = cv2.moments(contours[0])
moments_star_2 = cv2.moments(contours[1])
except IndexError:
print("Only {} were found ! (Must be at least 2)".format(
len(contours)))
else:
try:
cX_star1 = int(moments_star_1["m10"] / moments_star_1["m00"])
cY_star1 = int(moments_star_1["m01"] / moments_star_1["m00"])
cX_star2 = int(moments_star_2["m10"] / moments_star_2["m00"])
cY_star2 = int(moments_star_2["m01"] / moments_star_2["m00"])
except ZeroDivisionError:
return
if self.enable_seeing.isChecked():
delta_x = abs(cX_star2 - cX_star1)
delta_y = abs(cY_star2 - cY_star1)
self.arr_delta_x.append(delta_x)
self.arr_delta_y.append(delta_y)
# self._calcSeeing()
self._calcSeeing_arcsec()
cv2.drawMarker(self.draw_only_frame, (cX_star1, cY_star1),
color=(0, 0, 255),
markerSize=30,
thickness=1)
cv2.drawMarker(self.draw_only_frame, (cX_star2, cY_star2),
color=(0, 0, 255),
markerSize=30,
thickness=1)
finally:
self._displayImage()
threading.Thread(target=self._writeVideoFile, args=(),
daemon=True).start()
toc = time.time()
elapsed = toc - tic
try:
print("FPS max = {}".format(int(1.0 / elapsed)))
except ZeroDivisionError:
pass
focal = 1.0
E, mask = cv2.findEssentialMat(coords2, coords1, focal=1.0, pp=(0., 0.), method=cv2.RANSAC, prob=0.999, threshold=1.0)
# R, t = cv2.recoverPose(E, pcoords2, coords1, focal = 1.0, pp = (0.,0.), mask)
points, R, t, mask = cv2.recoverPose(E, coords2, coords1)
t_p = np.dot(t.T, R_pos)
t_pos = t_pos + t_p.T
R_pos = np.dot(R, R_pos)
# print(R_pos)
# print(t_pos)
# print("===========================")
# drawing matched points
for marker in coords1[:maxMatches]:
img4 = cv2.drawMarker(img3, tuple(int(i) for i in marker), color=(0, 255, 255))
th_x = math.asin(R[2][1])
th_y = math.atan(-R[2][0]/R[2][2])
th_z = math.atan(-R[0][1]/R[1][1])
print("\r{0} {1} {2}".format(th_x, th_y, th_z), end="")
theta = [0.0, th_z]
r = [0.0, 0.5]
plt.polar(theta, r)
def DebugMatches(filename,
width,
height,
img1,
points1,
H1,
img2,
points2,
H2,
TargetWidth=1024):
def transformPoints(points, H):
return cv2.perspectiveTransform(numpy.float32([points]), H)[0]
def Scale(sx, sy):
return numpy.array([[float(sx), 0.0, 0.0], [0.0, float(sy), 0.0],
[0.0, 0.0, 1.0]])
def Translate(tx, ty):
return numpy.array([[1.0, 0.0, float(tx)], [0.0, 1.0,
float(ty)],
[0.0, 0.0, 1.0]])
bounds1 = transformPoints([(0, 0), (width, 0), (width, height),
(0, height)], H1)
bounds2 = transformPoints([(0, 0), (width, 0), (width, height),
(0, height)], H2)
bounds = [*bounds1, *bounds2]
border = 10
x1 = int(min([x for x, y in bounds])) - border
x2 = int(max([x for x, y in bounds])) + border
y1 = int(min([y for x, y in bounds])) - border
y2 = int(max([y for x, y in bounds])) + border
W = x2 - x1
H = y2 - y1
vs = float(TargetWidth) / float(W)
W = int(W * vs)
H = int(H * vs)
H1 = numpy.matmul(numpy.matmul(Scale(vs, vs), Translate(-x1, -y1)), H1)
H2 = numpy.matmul(numpy.matmul(Scale(vs, vs), Translate(-x1, -y1)), H2)
# could be that img1 and img2 are at a lower resolution
Timg1 = numpy.matmul(
H1, Scale(width / float(img1.shape[1]), height / float(img1.shape[0])))
Timg2 = numpy.matmul(
H2, Scale(width / float(img2.shape[1]), height / float(img2.shape[0])))
# , cv2.INTER_LINEAR, cv2.BORDER_TRANSPARENT, (0, 0, 0, 0))
blended1 = cv2.warpPerspective(img1, Timg1, (W, H))
# , cv2.INTER_LINEAR, cv2.BORDER_TRANSPARENT, (0, 0, 0, 0))
blended2 = cv2.warpPerspective(img2, Timg2, (W, H))
img = numpy.zeros((H, W, 4), dtype='uint8')
img[:, :, 3].fill(255)
img[:, :, 0] = blended1
img[:, :, 1] = blended2
red = (0, 0, 255, 100)
green = (0, 255, 0, 100)
blue = (0, 0, 255, 100)
points1 = transformPoints(points1, H1)
points2 = transformPoints(points2, H2)
for point1, point2 in zip(points1, points2):
cv2.drawMarker(img, tuple(point1), red, cv2.MARKER_CROSS, 10, 2)
cv2.drawMarker(img, tuple(point2), green, cv2.MARKER_CROSS, 10, 2)
cv2.line(img, tuple(point1), tuple(point2), red, 3)
cv2.line(img, tuple(point1), tuple(point2), green, 3)
borders1 = transformPoints([(0, 0), (width, 0), (width, height),
(0, height)], H1)
borders2 = transformPoints([(0, 0), (width, 0), (width, height),
(0, height)], H2)
for I in range(4):
cv2.line(img, tuple(borders1[I]), tuple(borders1[(I + 1) % 4]), red, 2)
cv2.line(img, tuple(borders2[I]), tuple(borders2[(I + 1) % 4]), green,
2)
img = cv2.flip(img, 0)
SaveImage(filename, img)
def particle_tracker(v, file_name):
# Open output file
output_name = sys.argv[3] + file_name
output = open(output_name,"w")
frameCounter = 0
stepsize= 1
# read first frame
ret, frame = v.read()
if ret == False:
return
# detect face in first frame
c,r,w,h = detect_one_face(frame)
# Write track point for first frame
pt_x, pt_y=c + w/2.0,r + h/2.0
#channged
output.write("%d,%d,%d\n" % (frameCounter,pt_x,pt_y)) # Write as 0,pt_x,pt_y
frameCounter = frameCounter + 1
# set the initial tracking window
track_window = (c,r,w,h)
# initialize the tracker
# e.g. kf = cv2.KalmanFilter(4,2,0)
# or: particles = np.ones((n_particles, 2), int) * initial_pos
n_particles = 450
roi_hist = hsv_histogram_for_window(frame, (c,r,w,h)) # this is provided for you
init_pos = np.array([c + w / 2.0, r + h / 2.0], int) # Initial position
particles = np.ones((n_particles, 2), int) * init_pos # Init particles to init position
hsv = cv2.cvtColor(frame, cv2.COLOR_BGR2HSV)
prob = cv2.calcBackProject([hsv], [0], roi_hist, [0, 180], 1)
# print(init_pos.shape)
# print(particles.shape)
f0 = particleevaluator(prob, particles.T) * np.ones(n_particles) # Evaluate appearance model
weights = np.ones(n_particles) / n_particles # weights are uniform (at first)
# print(np.average(particles,axis=0))
pos=init_pos
while(1):
stepsize=30
ret ,frame = v.read() # read another frame
if ret == False:
break
# if you track particles - take the weighted average
# the Kalman filter already has the tracking point in the state vector
# hist_bp: obtain using cv2.calcBackProject and the HSV histogram
# c,r,w,h: obtain using detect_one_face()
# Particle motion model: uniform step (TODO: find a better motion model)
hsv = cv2.cvtColor(frame, cv2.COLOR_BGR2HSV)
prob = cv2.calcBackProject([hsv], [0], roi_hist, [0, 180], 1)
#moving in the direction of particle with maximum weight
# x=pos[0]+np.random.uniform(-5, 5)
# y=pos[1]+np.random.uniform(-5, 5)
# init_pos=pos
np.add(particles, np.random.uniform(-stepsize,stepsize, particles.shape), out=particles, casting="unsafe")
# print(particles.shape)
# print(np.argmax())
# Clip out-of-bounds particles,determine the width and height and clip to this rnge
particles = particles.clip(np.zeros(2), np.array((frame.shape[1], frame.shape[0])) - 1).astype(int)
f = particleevaluator(prob, particles.T) # Evaluate particles
weights = np.float32(f.clip(1)) # Weight ~ histogram response
weights /= np.sum(weights) # Normalize w
pos = np.sum(particles.T * weights, axis=1).astype(int) # expected position: weighted average
# print(f-f.clip(1))
if 1. / np.sum(weights ** 2) < n_particles / 2: # If particle cloud degenerate:
# print('resampled')
particles = particles[resample(weights), :] # Resample particles according to weights
# resample() function is provided for you
img2 = cv2.drawMarker(frame,(pos[0],pos[1]),(0,255,0),markerType=1,markerSize=10)
cv2.imshow('img',img2)
cv2.waitKey(60)
# plt.imshow(cv2.cvtColor(frame, cv2.COLOR_RGB2BGR))
# # plt.scatter(particles[1],particles[0], c='b', s=5)
# plt.pause(0.05)
# write the result to the output file
output.write("%d,%d,%d\n" % (frameCounter,pos[0],pos[1])) # Write as frame_index,pt_x,pt_y
frameCounter = frameCounter + 1
output.close()
def learn_image_maker(r1x, r1y, r1_theta, r2x, r2y, m11, m12, m21, m22, m31,
m32, m41, m42, mc1, mc2, mc3, mc4, rm1, rm2, rm3, erm1,
erm2, erm3): #r1 =robot r2 = enemy
#def learn_image_maker():
map = map_init()
#point state
if m11 == 1:
map = cv2.line(map, (125, 36), (111, 50), 170, thickness=2)
elif m11 == -1:
map = cv2.line(map, (125, 36), (111, 50), 85, thickness=2)
if m12 == 1:
map = cv2.line(map, (139, 50), (125, 64), 170, thickness=2)
elif m12 == -1:
map = cv2.line(map, (139, 50), (125, 64), 85, thickness=2)
if m21 == 1:
map = cv2.line(map, (200, 111), (186, 125), 170, thickness=2)
elif m21 == -1:
map = cv2.line(map, (200, 111), (186, 125), 85, thickness=2)
if m22 == 1:
map = cv2.line(map, (214, 125), (200, 139), 170, thickness=2)
elif m22 == -1:
map = cv2.line(map, (214, 125), (200, 139), 85, thickness=2)
if m31 == 1:
map = cv2.line(map, (125, 186), (111, 200), 170, thickness=2)
elif m31 == -1:
map = cv2.line(map, (125, 186), (111, 200), 85, thickness=2)
if m32 == 1:
map = cv2.line(map, (139, 200), (125, 214), 170, thickness=2)
elif m32 == -1:
map = cv2.line(map, (139, 200), (125, 214), 85, thickness=2)
if m41 == 1:
map = cv2.line(map, (50, 111), (36, 125), 170, thickness=2)
elif m41 == -1:
map = cv2.line(map, (50, 111), (36, 125), 85, thickness=2)
if m42 == 1:
map = cv2.line(map, (64, 125), (50, 139), 170, thickness=2)
elif m42 == -1:
map = cv2.line(map, (64, 125), (50, 139), 85, thickness=2)
if mc1 == 1:
map = cv2.line(map, (125, 100), (100, 125), 170, thickness=2)
elif mc1 == -1:
map = cv2.line(map, (125, 100), (100, 125), 85, thickness=2)
if mc2 == 1:
map = cv2.line(map, (125, 100), (150, 125), 170, thickness=2)
elif mc2 == -1:
map = cv2.line(map, (125, 100), (150, 125), 85, thickness=2)
if mc3 == 1:
map = cv2.line(map, (150, 125), (125, 150), 170, thickness=2)
elif mc3 == -1:
map = cv2.line(map, (150, 125), (125, 150), 85, thickness=2)
if mc4 == 1:
map = cv2.line(map, (125, 150), (100, 125), 170, thickness=2)
elif mc4 == -1:
map = cv2.line(map, (125, 150), (100, 125), 85, thickness=2)
#robot_state
#robot1
pts = np.array(
[[r1x, r1y - 13], [r1x + 13, r1y], [r1x, r1y + 13], [r1x - 13, r1y]],
np.int32)
pts = pts.reshape((-1, 1, 2))
map = cv2.fillPoly(map, [pts], 170)
#robot rotation
if r1_theta == 1:
map = cv2.drawMarker(map, (r1x - 7, r1y - 7),
255,
markerType=cv2.MARKER_TILTED_CROSS,
thickness=1)
#robot marker
if rm1 == 1:
map = cv2.line(map, (r1x, r1y - 13), (r1x + 13, r1y),
85,
thickness=2)
if rm2 == 1:
map = cv2.line(map, (r1x + 13, r1y), (r1x, r1y + 13),
85,
thickness=2)
if rm3 == 1:
map = cv2.line(map, (r1x, r1y + 13), (r1x - 13, r1y),
85,
thickness=2)
elif r1_theta == 2:
map = cv2.drawMarker(map, (r1x + 7, r1y - 7),
255,
markerType=cv2.MARKER_TILTED_CROSS,
thickness=1)
#robot marker
if rm1 == 1:
map = cv2.line(map, (r1x + 13, r1y), (r1x, r1y + 13),
85,
thickness=2)
if rm2 == 1:
map = cv2.line(map, (r1x, r1y + 13), (r1x - 13, r1y),
85,
thickness=2)
if rm3 == 1:
map = cv2.line(map, (r1x, r1y - 13), (r1x - 13, r1y),
85,
thickness=2)
elif r1_theta == 3:
map = cv2.drawMarker(map, (r1x + 7, r1y + 7),
255,
markerType=cv2.MARKER_TILTED_CROSS,
thickness=1)
#robot marker
if rm1 == 1:
map = cv2.line(map, (r1x, r1y + 13), (r1x - 13, r1y),
85,
thickness=2)
if rm2 == 1:
map = cv2.line(map, (r1x, r1y - 13), (r1x - 13, r1y),
85,
thickness=2)
if rm3 == 1:
map = cv2.line(map, (r1x + 13, r1y), (r1x, r1y - 13),
85,
thickness=2)
elif r1_theta == 4:
map = cv2.drawMarker(map, (r1x - 7, r1y + 7),
255,
markerType=cv2.MARKER_TILTED_CROSS,
thickness=1)
#robot marker
if rm1 == 1:
map = cv2.line(map, (r1x, r1y - 13), (r1x - 13, r1y),
85,
thickness=2)
if rm2 == 1:
map = cv2.line(map, (r1x + 13, r1y), (r1x, r1y - 13),
85,
thickness=2)
if rm3 == 1:
map = cv2.line(map, (r1x, r1y + 13), (r1x + 13, r1y),
85,
thickness=2)
#robot2
pts = np.array(
[[r2x, r2y - 13], [r2x + 13, r2y], [r2x, r2y + 13], [r2x - 13, r2y]],
np.int32)
pts = pts.reshape((-1, 1, 2))
map = cv2.fillPoly(map, [pts], 85)
map = cv2.drawMarker(map, (r2x + 7, r2y + 7),
255,
markerType=cv2.MARKER_TILTED_CROSS,
thickness=1)
#robot marker
if erm1 == 1:
map = cv2.line(map, (r2x, r2y + 13), (r2x - 13, r2y), 170, thickness=2)
if erm2 == 1:
map = cv2.line(map, (r2x, r2y - 13), (r2x - 13, r2y), 170, thickness=2)
if erm3 == 1:
map = cv2.line(map, (r2x + 13, r2y), (r2x, r2y - 13), 170, thickness=2)
#cv2.imwrite('test.png',map)
return map
def kf_tracker(v, file_name):
# Open output file
output_name = sys.argv[3] + file_name
output = open(output_name,"w")
frameCounter = 0
stepsize= 1
# read first frame
ret, frame = v.read()
if ret == False:
return
# detect face in first frame
c,r,w,h = detect_one_face(frame)
# Write track point for first frame
pt_x, pt_y=c + w/2.0,r + h/2.0
output.write("%d,%d,%d\n" % (frameCounter,pt_x,pt_y)) # Write as 0,pt_x,pt_y
frameCounter = frameCounter + 1
state = np.array([c + w / 2, r + h / 2, 0, 0], dtype=np.float32) # initial position
kalman = cv2.KalmanFilter(4, 2)
kalman.measurementMatrix = np.array([[1, 0, 0, 0], [0, 1, 0, 0]], np.float32)
kalman.transitionMatrix = np.array([[1, 0, 1, 0], [0, 1, 0, 1], [0, 0, 1, 0], [0, 0, 0, 1]], np.float32)
kalman.processNoiseCov = np.array([[1, 0, 0, 0], [0, 1, 0, 0], [0, 0, 1, 0], [0, 0, 0, 1]], np.float32) * 0.03
kalman.measurementNoiseCov = np.array([[1, 0], [0, 1]], np.float32) * 0.00003
kalman.errorCovPost = 1e-1 * np.eye(4, 4,dtype=np.float32)
kalman.statePost = state
# print(kalman.measurementMatrix)
while(1):
ret, frame = v.read() # read another frame
if ret == False:
break
c, r, w, h = detect_one_face(frame)
# if you track particles - take the weighted average
# the Kalman filter already has the tracking point in the state vector
# hist_bp: obtain using cv2.calcBackProject and the HSV histogram
# c,r,w,h: obtain using detect_one_face()
# Particle motion model: uniform step (TODO: find a better motion model)
# hsv = cv2.cvtColor(frame, cv2.COLOR_BGR2HSV)
# prob = cv2.calcBackProject([hsv], [0], roi_hist, [0, 180], 1)
# retval,track_window=cv2.CamShift(prob,track_window,term_crit)
# c, r, w, h = track_window
prediction = kalman.predict()
pos=prediction[:2]
# if not((c==0)and(r==0)and(w==0)and(h==0)): # e.g. face found
if (any([c,r,w,h])>0): # e.g. face found
posterior = kalman.correct(np.array([[np.float32(c+w/2)],[np.float32(r+h/2)]]))
# print(posterior[:2])
pos = posterior[:2]
else:
pass
# print('entered kalman',frameCounter)
# use prediction or posterior as your tracking result
img2 = cv2.drawMarker(frame,(pos[0],pos[1]),(0,255,0),markerType=1,markerSize=10)
cv2.imshow('img',img2)
cv2.waitKey(60)
# write the result to the output file
output.write("%d,%d,%d\n" % (frameCounter,pos[0],pos[1])) # Write as frame_index,pt_x,pt_y
frameCounter = frameCounter + 1
output.close()
def draw_point(self, img, front_pt, back_pt):
img = cv2.circle(img, front_pt, 8, (125, 0, 125), 2, 0)
img = cv2.drawMarker(img, back_pt, color=(255, 255, 0), markerType=1)
img = cv2.line(img, front_pt, back_pt, (100, 100, 0), 2)
img = cv2.circle(img, front_pt, 8, (255, 255, 0), 2, 0)
return img
def annotate(self):
"""Label all kinds of things in the image.
"""
if self.search_point is not None:
p1, p2 = self.get_search_window()
cv2.rectangle(self.img, p1, p2, color=(255, 255, 255), thickness=1)
else:
cv2.rectangle(self.img, (0, 0), (self.width - 1, self.height - 1),
color=(255, 255, 255),
thickness=1)
# draw all detected contours to see masking issues
if DRAW_MINOR_CONTOURS and self.contours is not None:
cv2.drawContours(self.img, self.contours, -1, (150, 150, 0),
THICKNESS_MINOR_CONTOUR)
# draw largest contour and contour label
if DRAW_MAJOR_CONTOURS and self.largest_contour is not None:
cv2.drawContours(self.img, [self.largest_contour], 0, (0, 0, 255),
THICKNESS_MAJOR_CONTOUR)
# Marker on centroid of largest contour
if self.largest_contour is not None:
cv2.drawMarker(img=self.img,
position=centroid(self.largest_contour),
color=(0, 255, 0))
# Label nodes, highlight node closest to largest contour centroid
for node_id, node in self.nodes.items():
color = (255, 0, 0) if node_id == self.active_node else (255, 255,
255)
cv2.circle(self.img, (node['x'], node['y']), MIN_DIST_TO_NODE // 2,
color)
# Draw the detection trail
points = self.results
if DRAW_TRAIL and len(points) > 1:
for p_idx in range(len(points) - 1):
try:
x1, y1 = map(int, points[p_idx])
x2, y2 = map(int, points[p_idx + 1])
except (ValueError, TypeError):
pass
else:
cv2.line(self.img, (x1, y1), (x2, y2),
color=(0, 255, 0),
thickness=2)
# Draw the Kalman filter predictions trail
cv2.drawMarker(self.img, position=self.last_kf_pos, color=(0, 0, 255))
points = self.kf_results
if DRAW_KF_TRAIL and len(points) > 1:
for p_idx in range(len(points) - 1):
try:
x1, y1 = map(int, points[p_idx])
x2, y2 = map(int, points[p_idx + 1])
except (ValueError, TypeError):
pass
else:
cv2.line(self.img, (x1, y1), (x2, y2),
color=(50, 50, 255),
thickness=1)
def get_frame(self):
font = cv2.FONT_HERSHEY_DUPLEX
#動画のフレーム数をカウント
f = 0
self.flg = 0
#図形の認識に用いる閾値
# self.threshold_time = 180
# self.threshold_new_block = 25
# self.threshold_block_move = 3
# self.threshold_area = 500
# self.detect_red_cx = [num*5 for num in range(self.threshold_time)]
# self.detect_red_cy = [num*5 for num in range(self.threshold_time)]
# self.detect_blue_cx = [num*5 for num in range(self.threshold_time)]
# self.detect_blue_cy = [num*5 for num in range(self.threshold_time)]
# self.detect_green_cx = [num*5 for num in range(self.threshold_time)]
# self.detect_green_cy = [num*5 for num in range(self.threshold_time)]
# self.detect_purple_cx = [num*5 for num in range(self.threshold_time)]
# self.detect_purple_cy = [num*5 for num in range(self.threshold_time)]
# self.detect_yellow_cx = [num*5 for num in range(self.threshold_time)]
# self.detect_yellow_cy = [num*5 for num in range(self.threshold_time)]
# # HSV色空間における"赤色"の値域を決定
# self.lower_red1 = np.array([0, 65, 0])
# self.upper_red1 = np.array([30, 255, 255])
# self.lower_red2 = np.array([122, 65, 0])
# self.upper_red2 = np.array([180, 255, 255])
# # HSV色空間における"青色"の値域を決定
# self.lower_blue = np.array([93, 90, 0])
# self.upper_blue = np.array([114, 255, 164])
# # HSV色空間における"緑色"の値域を決定
# _green = np.array([62, 90, 0])
# self.upper_green = np.array([91, 255, 164])
# # HSV色空間における"紫色"の値域を決定
# self.lower_purple = np.array([150, 70, 150])
# self.upper_purple = np.array([160, 176, 255])
# # HSV色空間における"黄色"の値域を決定
# self.lower_yellow = np.array([30, 120, 120])
# self.upper_yellow = np.array([60, 255, 255])
# self.flg = 0
f += 1
#ーーーーーーーーーーーーーーーーーーーーーーーーーーself.videoのカメラ映像取得ーーーーーーーーーーーーーーーーーーーーーーーーーーーーーーーー
ret, self.frame = self.video.read()
# resize the window
height = self.frame.shape[0]
width = self.frame.shape[1]
self.frame = cv2.resize(self.frame,
(int(width * 0.5), int(height * 0.5)))
# windowsize = (300, 300)
# self.frame = cv2.resize(self.frame, windowsize)
self.frame = cv2.flip(self.frame, 1)
#---------------------------------------------------------webカメラからの映像------------------------------------------------
self.frame_src = self.frame.copy()
hsv = cv2.cvtColor(self.frame, cv2.COLOR_BGR2HSV)
# ウェブカメラからの"赤色"の検出----------------------------------------------------
mask_red1 = cv2.inRange(hsv, self.lower_red1, self.upper_red1)
mask_red2 = cv2.inRange(hsv, self.lower_red2, self.upper_red2)
mask_red = mask_red1 + mask_red2
# マスク画像をブロブ解析
# 2値画像のラベリング処理
label = cv2.connectedComponentsWithStats(mask_red)
# ブロブ情報を項目別に抽出
n = label[0] - 1
#label[2]は各図形のバウンティンぐボックスとオブジェクトのサイズを保存
data = np.delete(label[2], 0, 0)
center = np.delete(label[3], 0, 0)
# 検出した図形の面積から一定の大きさのものだけ抽出
data2 = np.zeros((0, 5), dtype="int32")
center2 = np.zeros((0, 2), dtype="float64")
for i in range(n):
if data[i, 4] > self.threshold_area:
data2 = np.r_[data2, data[i:i + 1, :]]
center2 = np.r_[center2, center[i:i + 1, :]]
if data2.shape[0] != 0:
for i in range(data2.shape[0]):
# ブロブ面積最大のインデックス
#max_index = np.argmax(data[:, 4])
detected_blob = {}
# 各ブロブの各種情報を取得
detected_blob["upper_left"] = (data2[:, 0][i], data2[:, 1][i]
) # 左上座標
detected_blob["width"] = data2[:, 2][i] # 幅
detected_blob["height"] = data2[:, 3][i] # 高さ
detected_blob["area"] = data2[:, 4][i] # 面積
detected_blob["center"] = center2[i] # 中心座標
# ブロブの中心座標を取得
center_red_x = int(detected_blob["center"][0])
center_red_y = int(detected_blob["center"][1])
cv2.putText(self.frame, "bar", (center_red_x, center_red_y),
font, 1, (0), 2)
cv2.drawMarker(self.frame, (center_red_x, center_red_y),
(0, 0, 0),
markerSize=10,
thickness=10)
#過去に同じ座標で認識しているとき
for k in range(len(list_shape_cx)):
if (abs(list_shape_cx[k] - center_red_x) <
self.threshold_new_block) and (
abs(list_shape_cy[k] - center_red_y) <
self.threshold_new_block):
#print('already detected red'+str(center_red_x)+', '+str(center_red_y))
break
else:
continue
else:
self.detect_red_cx.append(center_red_x)
self.detect_red_cy.append(center_red_y)
for i_c in range(self.threshold_time):
#print("i_c:",i_c)
#print("self.detect_red_cx[-1]:",self.detect_red_cx[-1])
if (abs(self.detect_red_cx[-1] -
self.detect_red_cx[-(2 + i_c)]) <
self.threshold_block_move):
if (abs(self.detect_red_cy[-1] -
self.detect_red_cy[-(2 + i_c)])
) < self.threshold_block_move:
if i_c == (self.threshold_time - 1):
#print("bar")
list_shape.append(0)
list_shape_cx.append(center_red_x)
list_shape_cy.append(center_red_y)
#cv2.imwrite(path+"/new_object_"+str(f)+".jpg", self.frame)
self.flg = 1
else:
continue
else:
break
else:
break
else:
break
continue
# ウェブカメラからの"青色"の検出----------------------------------------------------
mask_blue = cv2.inRange(hsv, self.lower_blue, self.upper_blue)
# マスク画像を
# 2値画像のラベリング処理
label = cv2.connectedComponentsWithStats(mask_blue)
# ブロブ情報を項目別に抽出
n = label[0] - 1
#label[2]は各図形のバウンティンぐボックスとオブジェクトのサイズを保存
data = np.delete(label[2], 0, 0)
center = np.delete(label[3], 0, 0)
# 検出した図形の面積から一定の大きさのものだけ抽出
data2 = np.zeros((0, 5), dtype="int32")
center2 = np.zeros((0, 2), dtype="float64")
for i in range(n):
if data[i, 4] > self.threshold_area:
data2 = np.r_[data2, data[i:i + 1, :]]
center2 = np.r_[center2, center[i:i + 1, :]]
if data2.shape[0] != 0:
for i in range(data2.shape[0]):
# ブロブ面積最大のインデックス
#max_index = np.argmax(data[:, 4])
detected_blob = {}
# 各ブロブの各種情報を取得
detected_blob["upper_left"] = (data2[:, 0][i], data2[:, 1][i]
) # 左上座標
detected_blob["width"] = data2[:, 2][i] # 幅
detected_blob["height"] = data2[:, 3][i] # 高さ
detected_blob["area"] = data2[:, 4][i] # 面積
detected_blob["center"] = center2[i] # 中心座標
# ブロブの中心座標を取得
center_blue_x = int(detected_blob["center"][0])
center_blue_y = int(detected_blob["center"][1])
cv2.putText(self.frame, "box", (center_blue_x, center_blue_y),
font, 1, (0), 2)
cv2.drawMarker(self.frame, (center_blue_x, center_blue_y),
(0, 0, 0),
markerSize=10,
thickness=10)
#過去に同じ座標で認識しているとき
for k in range(len(list_shape_cx)):
if (abs(list_shape_cx[k] - center_blue_x) <
self.threshold_new_block) and (
abs(list_shape_cy[k] - center_blue_y) <
self.threshold_new_block):
#print('already detected')
break
else:
continue
else:
self.detect_blue_cx.append(center_blue_x)
self.detect_blue_cy.append(center_blue_y)
for i_c in range(self.threshold_time):
#print("i_c",i_c)
#print("self.detect_blue_cx[-1]:",self.detect_blue_cx[-1])
if (abs(self.detect_blue_cx[-1] -
self.detect_blue_cx[-(2 + i_c)]) <
self.threshold_block_move):
if (abs(self.detect_blue_cy[-1] -
self.detect_blue_cy[-(2 + i_c)]) <
self.threshold_block_move):
if i_c == (self.threshold_time - 1):
#print("box")
list_shape.append(5)
list_shape_cx.append(center_blue_x)
list_shape_cy.append(center_blue_y)
#cv2.imwrite(path+"/new_object_"+str(f)+".jpg", self.frame)
self.flg = 1
else:
continue
else:
break
else:
break
else:
break
continue
# ウェブカメラからの"緑色"の検出----------------------------------------------------
mask_green = cv2.inRange(hsv, self.lower_green, self.upper_green)
# マスク画像をブロブ解析
# 2値画像のラベリング処理
label = cv2.connectedComponentsWithStats(mask_green)
# ブロブ情報を項目別に抽出
n = label[0] - 1
#label[2]は各図形のバウンティンぐボックスとオブジェクトのサイズを保存
data = np.delete(label[2], 0, 0)
center = np.delete(label[3], 0, 0)
# 検出した図形の面積から一定の大きさのものだけ抽出
data2 = np.zeros((0, 5), dtype="int32")
center2 = np.zeros((0, 2), dtype="float64")
for i in range(n):
if data[i, 4] > self.threshold_area:
data2 = np.r_[data2, data[i:i + 1, :]]
center2 = np.r_[center2, center[i:i + 1, :]]
if data2.shape[0] != 0:
for i in range(data2.shape[0]):
# ブロブ面積最大のインデックス
#max_index = np.argmax(data[:, 4])
detected_blob = {}
# 各ブロブの各種情報を取得
detected_blob["upper_left"] = (data2[:, 0][i], data2[:, 1][i]
) # 左上座標
detected_blob["width"] = data2[:, 2][i] # 幅
detected_blob["height"] = data2[:, 3][i] # 高さ
detected_blob["area"] = data2[:, 4][i] # 面積
detected_blob["center"] = center2[i] # 中心座標
# ブロブの中心座標を取得
center_green_x = int(detected_blob["center"][0])
center_green_y = int(detected_blob["center"][1])
cv2.putText(self.frame, "triangle",
(center_green_x, center_green_y), font, 1, (0), 2)
cv2.drawMarker(self.frame, (center_green_x, center_green_y),
(0, 0, 0),
markerSize=10,
thickness=10)
#過去に同じ座標で認識しているとき
for k in range(len(list_shape_cx)):
if (abs(list_shape_cx[k] - center_green_x) <
self.threshold_new_block) and (
abs(list_shape_cy[k] - center_green_y) <
self.threshold_new_block):
#print('already detected green:'+str(center_green_x)+', '+str(center_green_y))
break
else:
continue
else:
self.detect_green_cx.append(center_green_x)
self.detect_green_cy.append(center_green_y)
for i_c in range(self.threshold_time):
#print("i_c:",i_c)
#print("self.detect_green_cx[-1]:",self.detect_green_cx[-1])
if (abs(self.detect_green_cx[-1] -
self.detect_green_cx[-(2 + i_c)]) <
self.threshold_block_move):
if (abs(self.detect_green_cy[-1] -
self.detect_green_cy[-(2 + i_c)]) <
self.threshold_block_move):
if i_c == (self.threshold_time - 1):
#print("triangle")
list_shape.append(3)
list_shape_cx.append(center_green_x)
list_shape_cy.append(center_green_y)
#cv2.imwrite(path+"/new_object_"+str(f)+".jpg", self.frame)
self.flg = 1
else:
continue
else:
break
else:
break
else:
break
continue
# ウェブカメラからの"紫色"の検出----------------------------------------------------
mask_purple = cv2.inRange(hsv, self.lower_purple, self.upper_purple)
# マスク画像をブロブ解析
# 2値画像のラベリング処理
label = cv2.connectedComponentsWithStats(mask_purple)
# ブロブ情報を項目別に抽出
n = label[0] - 1
#label[2]は各図形のバウンティンぐボックスとオブジェクトのサイズを保存
data = np.delete(label[2], 0, 0)
center = np.delete(label[3], 0, 0)
# 検出した図形の面積から一定の大きさのものだけ抽出
data2 = np.zeros((0, 5), dtype="int32")
center2 = np.zeros((0, 2), dtype="float64")
for i in range(n):
if data[i, 4] > self.threshold_area:
data2 = np.r_[data2, data[i:i + 1, :]]
center2 = np.r_[center2, center[i:i + 1, :]]
if data2.shape[0] != 0:
for i in range(data2.shape[0]):
# ブロブ面積最大のインデックス
#max_index = np.argmax(data[:, 4])
detected_blob = {}
# 各ブロブの各種情報を取得
detected_blob["upper_left"] = (data2[:, 0][i], data2[:, 1][i]
) # 左上座標
detected_blob["width"] = data2[:, 2][i] # 幅
detected_blob["height"] = data2[:, 3][i] # 高さ
detected_blob["area"] = data2[:, 4][i] # 面積
detected_blob["center"] = center2[i] # 中心座標
# ブロブの中心座標を取得
center_purple_x = int(detected_blob["center"][0])
center_purple_y = int(detected_blob["center"][1])
cv2.putText(self.frame, "half-circle",
(center_purple_x, center_purple_y), font, 1, (0),
2)
cv2.drawMarker(self.frame, (center_purple_x, center_purple_y),
(0, 0, 0),
markerSize=10,
thickness=10)
#過去に同じ座標で認識しているとき
for k in range(len(list_shape_cx)):
if (abs(list_shape_cx[k] - center_purple_x) <
self.threshold_new_block) and (
abs(list_shape_cy[k] - center_purple_y) <
self.threshold_new_block):
#print('already detected purple:'+str(center_purple_x)+', '+str(center_purple_y))
break
else:
continue
else:
self.detect_purple_cx.append(center_purple_x)
self.detect_purple_cy.append(center_purple_y)
for i_c in range(self.threshold_time):
#print("i_c:",i_c)
#print("self.detect_purple_cx[-1]:",self.detect_purple_cx[-1])
if (abs(self.detect_purple_cx[-1] -
self.detect_purple_cx[-(2 + i_c)]) <
self.threshold_block_move):
if (abs(self.detect_purple_cy[-1] -
self.detect_purple_cy[-(2 + i_c)]) <
self.threshold_block_move):
if i_c == (self.threshold_time - 1):
#print("half-circle")
list_shape.append(3)
list_shape_cx.append(center_purple_x)
list_shape_cy.append(center_purple_y)
#cv2.imwrite(path+"/new_object_"+str(f)+".jpg", self.frame)
self.flg = 1
else:
continue
else:
break
else:
break
else:
break
continue
# ウェブカメラからの"黄色"の検出----------------------------------------------------
mask_yellow = cv2.inRange(hsv, self.lower_yellow, self.upper_yellow)
# マスク画像をブロブ解析
# 2値画像のラベリング処理
label = cv2.connectedComponentsWithStats(mask_yellow)
# ブロブ情報を項目別に抽出
n = label[0] - 1
#label[2]は各図形のバウンティンぐボックスとオブジェクトのサイズを保存
data = np.delete(label[2], 0, 0)
center = np.delete(label[3], 0, 0)
# 検出した図形の面積から一定の大きさのものだけ抽出
data2 = np.zeros((0, 5), dtype="int32")
center2 = np.zeros((0, 2), dtype="float64")
for i in range(n):
if data[i, 4] > self.threshold_area:
data2 = np.r_[data2, data[i:i + 1, :]]
center2 = np.r_[center2, center[i:i + 1, :]]
if data2.shape[0] != 0:
for i in range(data2.shape[0]):
# ブロブ面積最大のインデックス
#max_index = np.argmax(data[:, 4])
detected_blob = {}
# 各ブロブの各種情報を取得
detected_blob["upper_left"] = (data2[:, 0][i], data2[:, 1][i]
) # 左上座標
detected_blob["width"] = data2[:, 2][i] # 幅
detected_blob["height"] = data2[:, 3][i] # 高さ
detected_blob["area"] = data2[:, 4][i] # 面積
detected_blob["center"] = center2[i] # 中心座標
# ブロブの中心座標を取得
center_yellow_x = int(detected_blob["center"][0])
center_yellow_y = int(detected_blob["center"][1])
cv2.putText(self.frame, "square",
(center_yellow_x, center_yellow_y), font, 1, (0),
2)
cv2.drawMarker(self.frame, (center_yellow_x, center_yellow_y),
(0, 0, 0),
markerSize=10,
thickness=10)
#過去に同じ座標で認識しているとき
for k in range(len(list_shape_cx)):
if (abs(list_shape_cx[k] - center_yellow_x) <
self.threshold_new_block) and (
abs(list_shape_cy[k] - center_yellow_y) <
self.threshold_new_block):
#print('already detected yellow:'+str(center_yellow_x)+', '+str(center_yellow_y))
break
else:
continue
else:
self.detect_yellow_cx.append(center_yellow_x)
self.detect_yellow_cy.append(center_yellow_y)
for i_c in range(self.threshold_time):
#print("i_c:",i_c)
#print("self.detect_yellow_cx[-1]:",self.detect_yellow_cx[-1])
if (abs(self.detect_yellow_cx[-1] -
self.detect_yellow_cx[-(2 + i_c)]) <
self.threshold_block_move):
#print("a")
if (abs(self.detect_yellow_cy[-1] -
self.detect_yellow_cy[-(2 + i_c)]) <
self.threshold_block_move):
#print("b")
if i_c == (self.threshold_time - 1):
#print("square")
list_shape.append(4)
list_shape_cx.append(center_yellow_x)
list_shape_cy.append(center_yellow_y)
#cv2.imwrite(path+"/new_object_"+str(f)+".jpg", self.frame)
self.flg = 1
else:
continue
else:
break
else:
break
else:
break
continue
# print('self.frame:'+str(f))
# cv2.imshow('mask_purple', mask_purple)
# cv2.imshow('mask_red', mask_red)
# cv2.imshow('mask_blue', mask_blue)
# cv2.imshow('mask_green', mask_green)
# cv2.imshow('mask_yellow', mask_yellow)
# cv2.imshow('main',self.frame)
# if cv2.waitKey(1) & 0xFF == ord('q'):
# break
# capture.release()
# cv2.destroyAllWindows()
#htmlに送る用のコード
ret, jpeg = cv2.imencode('.jpg', self.frame)
if self.flg == 0:
shape = 0
vec = [0, 0]
else:
shape = list_shape[-1]
vec = [list_shape_cx[-1], list_shape_cy[-1]]
x = np.random.randint(100, 400)
return jpeg.tobytes(), self.flg, shape, vec, x
def ImageProcessor(unprocessed_frames, processed_frames, proc_complete,
event_manager, n_calib):
processing = False
proc_complete.set()
# preallocate memory for all arrays used in processing
img_mean = np.zeros([h, w], dtype=np.float32)
img_std = np.ones([h, w], dtype=np.float32)
A = np.zeros([h, w], dtype=np.uint8)
B = np.zeros([h, w], dtype=np.uint8)
B_old = np.zeros([h, w], dtype=np.uint8)
C = np.zeros([h, w], dtype=np.uint8)
mean_1 = np.zeros([h, w], dtype=np.float32)
mean_2 = np.zeros([h, w], dtype=np.float32)
std_1 = np.zeros([h, w], dtype=np.float32)
std_2 = np.zeros([h, w], dtype=np.float32)
std_3 = np.zeros([h, w], dtype=np.float32)
std_4 = np.zeros([h, w], dtype=np.float32)
std_5 = np.zeros([h, w], dtype=np.float32)
std_6 = np.zeros([h, w], dtype=np.float32)
B_1_std = np.zeros([h, w], dtype=np.float32)
B_1_mean = np.zeros([h, w], dtype=np.float32)
B_greater = np.zeros([h, w], dtype=np.uint8)
B_2_mean = np.zeros([h, w], dtype=np.float32)
B_less = np.zeros([h, w], dtype=np.uint8)
########## FOR TESTING ##############
img_array = []
C_array = []
# std_data = np.zeros([h,w],dtype=np.float32)
# mean_data = np.zeros([h,w],dtype=np.float32)
save_arr = False
y_data = np.zeros((h, w))
img_temp = np.zeros((3, h, w))
C_temp = np.zeros((3, h, w))
temp_img = np.zeros((h, w))
########## FOR TESTING ##############
last_n_frame = 0
p = c.LEARNING_RATE
time_cumulative = 0
total_time = 0
total_frames = 0
while True:
time.sleep(1E-4)
try:
if event_manager.shutdown.is_set():
proc_complete.set()
return
# get the frames from queue
n_frame, n_frame_2, frame_buf = unprocessed_frames.get_nowait()
# y_data is a numpy array hxw with 8 bit greyscale brightness values
y_data = np.frombuffer(frame_buf, dtype=np.uint8,
count=w * h).reshape(
(h, w)).astype(np.float32)
# every 3 frames update the background
if n_frame_2 > (last_n_frame + c.BACKGROUND_RATE
) and not event_manager.event_change.is_set():
last_n_frame = n_frame_2
# img_mean = (1 - p) * img_mean + p * y_data # calculate mean
np.multiply(1 - p, img_mean, out=mean_1)
np.multiply(p, y_data, out=mean_2)
np.add(mean_1, mean_2, out=img_mean)
# img_std = np.sqrt((1 - p) * (img_std ** 2) + p * ((y_data - img_mean) ** 2)) # calculate std deviation
np.square(img_std, out=std_1)
np.multiply(1 - p, std_1, out=std_2)
np.subtract(y_data, img_mean, out=std_3)
np.square(std_3, out=std_4)
np.multiply(p, std_4, out=std_5)
np.add(std_2, std_5, out=std_6)
np.sqrt(std_6, out=img_std)
if not proc_complete.is_set() and n_frame > -1:
mean_data = img_mean
std_data = img_std
start = time.time_ns()
B_old = np.copy(B)
# B = np.logical_or((y_data > (img_mean + 2*img_std)),
# (y_data < (img_mean - 2*img_std))) # foreground new
np.multiply(img_std, 3, out=B_1_std)
np.add(B_1_std, img_mean, out=B_1_mean)
B_greater = np.greater(y_data, B_1_mean)
np.subtract(img_mean, B_1_std, out=B_2_mean)
B_less = np.less(y_data, B_2_mean)
B = np.logical_or(B_greater, B_less)
A = np.invert(
np.logical_and(B_old, B)
) # difference between prev foreground and new foreground
C = np.logical_and(
A,
B) # different from previous frame and part of new frame
C = 255 * C.astype(np.uint8)
C = cv2.filter2D(C, ddepth=-1, kernel=kernel4)
C[C < c.LPF_THRESH] = 0
C[C >= c.LPF_THRESH] = 255
n_features_cv, labels_cv, stats_cv, centroids_cv = cv2.connectedComponentsWithStats(
C, connectivity=4)
label_mask_cv = np.logical_and(
stats_cv[:, cv2.CC_STAT_AREA] > 2,
stats_cv[:, cv2.CC_STAT_AREA] < 10000)
ball_candidates = np.concatenate(
(stats_cv[label_mask_cv, 2:], centroids_cv[label_mask_cv]),
axis=1)
# sort ball candidates by size and keep the top 100
ball_candidates = ball_candidates[
ball_candidates[:, c.SIZE].argsort()[::-1][:c.N_OBJECTS]]
processed_frames.put((n_frame, ball_candidates))
########## FOR TESTING ##############
C_temp = cv2.cvtColor(C, cv2.COLOR_GRAY2RGB)
img_temp = cv2.cvtColor(y_data, cv2.COLOR_GRAY2RGB)
ball_candidates = ball_candidates.astype(int)
for ball in ball_candidates:
cv2.drawMarker(C_temp, (ball[c.X_COORD], ball[c.Y_COORD]),
(0, 0, 255),
cv2.MARKER_CROSS,
thickness=2,
markerSize=10)
cv2.drawMarker(img_temp,
(ball[c.X_COORD], ball[c.Y_COORD]),
(0, 0, 255),
cv2.MARKER_CROSS,
thickness=2,
markerSize=10)
save_arr = True
img_array.append((n_frame, y_data))
C_array.append(C_temp)
total_time += (time.time_ns() - start)
total_frames += 1
########## FOR TESTING ##############
elif event_manager.record_stream.is_set() and n_frame > -1:
if n_frame % n_calib.value == 0:
processed_frames.put((n_frame, y_data))
except queue.Empty:
if not event_manager.recording.is_set(
) and unprocessed_frames.qsize(
) == 0: # if the recording has finished
proc_complete.set() # set the proc_complete event
######### FOR TESTING ##############
if total_frames > 0:
print((total_time / total_frames) / 1E6)
total_frames = 0
total_time = 0
if img_array is not None and save_arr:
for i, img in enumerate(img_array):
frame, data = img
os.chdir(c.IMG_P)
cv2.imwrite(f"{frame:04d}.png", data)
cv2.imwrite(f"C{frame:04d}.png", C_array[i])
os.chdir(c.DATA_P)
np.save('img_mean', mean_data)
np.save('img_std', std_data)
os.chdir(c.ROOT_P)
img_array = []
C_array = []
save_arr = False
im2, contours, hierarchy = cv2.findContours(maskc, cv2.RETR_TREE,
cv2.CHAIN_APPROX_SIMPLE)
contours = sorted(contours, key=cv2.contourArea, reverse=True)[:1]
if contours:
# compute the center of the contour
M = cv2.moments(contours[0])
if M["m00"] != 0:
cX = int(M["m10"] / M["m00"])
cY = int(M["m01"] / M["m00"])
# draw the contour and center of the shape on the image
# cv2.drawContours(frame, [c], -1, (0, 255, 0), 2)
# cv2.circle(frame, (cX, cY), 7, (0, 255, 0), -1)
# cv2.putText(frame, "center", (cX - 20, cY - 20),
# cv2.FONT_HERSHEY_SIMPLEX, 0.5, (255, 255, 255), 2)
cv2.drawMarker(frame, (cX, cY), (0, 255, 0),
markerSize=200,
thickness=20)
else:
print("Nothing found")
cv2.namedWindow('frame', cv2.WINDOW_NORMAL)
cv2.resizeWindow('frame', 600, 600)
cv2.namedWindow('maskc', cv2.WINDOW_NORMAL)
cv2.resizeWindow('maskc', 600, 600)
cv2.moveWindow('maskc', 600, 0)
cv2.imshow("frame", frame)
cv2.imshow("maskc", maskc)
cv2.waitKey(2)
cv2.destroyAllWindows()
def _visualize_output():
last_frame_index = 0
last_frame_time = time.time()
fps_history = []
all_gaze_histories = []
if args.fullscreen:
cv.namedWindow('vis', cv.WND_PROP_FULLSCREEN)
cv.setWindowProperty('vis', cv.WND_PROP_FULLSCREEN,
cv.WINDOW_FULLSCREEN)
while True:
# If no output to visualize, show unannotated frame
if inferred_stuff_queue.empty():
next_frame_index = last_frame_index + 1
if next_frame_index in data_source._frames:
next_frame = data_source._frames[next_frame_index]
if 'faces' in next_frame and len(
next_frame['faces']) == 0:
if not args.headless:
cv.imshow('vis', next_frame['bgr'])
if args.record_video:
video_out_queue.put_nowait(next_frame_index)
last_frame_index = next_frame_index
if cv.waitKey(1) & 0xFF == ord('q'):
return
continue
# Get output from neural network and visualize
output = inferred_stuff_queue.get()
bgr = None
for j in range(batch_size):
frame_index = output['frame_index'][j]
if frame_index not in data_source._frames:
continue
frame = data_source._frames[frame_index]
# Decide which landmarks are usable
heatmaps_amax = np.amax(output['heatmaps'][j, :].reshape(
-1, 18),
axis=0)
can_use_eye = np.all(heatmaps_amax > 0.7)
can_use_eyelid = np.all(heatmaps_amax[0:8] > 0.75)
can_use_iris = np.all(heatmaps_amax[8:16] > 0.8)
start_time = time.time()
eye_index = output['eye_index'][j]
bgr = frame['bgr']
eye = frame['eyes'][eye_index]
eye_image = eye['image']
eye_side = eye['side']
eye_landmarks = output['landmarks'][j, :]
eye_radius = output['radius'][j][0]
if eye_side == 'left':
eye_landmarks[:, 0] = eye_image.shape[
1] - eye_landmarks[:, 0]
eye_image = np.fliplr(eye_image)
# Embed eye image and annotate for picture-in-picture
eye_upscale = 2
eye_image_raw = cv.cvtColor(cv.equalizeHist(eye_image),
cv.COLOR_GRAY2BGR)
eye_image_raw = cv.resize(eye_image_raw, (0, 0),
fx=eye_upscale,
fy=eye_upscale)
eye_image_annotated = np.copy(eye_image_raw)
if can_use_eyelid:
cv.polylines(
eye_image_annotated,
[
np.round(
eye_upscale * eye_landmarks[0:8]).astype(
np.int32).reshape(-1, 1, 2)
],
isClosed=True,
color=(255, 255, 0),
thickness=1,
lineType=cv.LINE_AA,
)
if can_use_iris:
cv.polylines(
eye_image_annotated,
[
np.round(
eye_upscale * eye_landmarks[8:16]).astype(
np.int32).reshape(-1, 1, 2)
],
isClosed=True,
color=(0, 255, 255),
thickness=1,
lineType=cv.LINE_AA,
)
cv.drawMarker(
eye_image_annotated,
tuple(
np.round(eye_upscale *
eye_landmarks[16, :]).astype(
np.int32)),
color=(0, 255, 255),
markerType=cv.MARKER_CROSS,
markerSize=4,
thickness=1,
line_type=cv.LINE_AA,
)
face_index = int(eye_index / 2)
eh, ew, _ = eye_image_raw.shape
v0 = face_index * 2 * eh
v1 = v0 + eh
v2 = v1 + eh
u0 = 0 if eye_side == 'left' else ew
u1 = u0 + ew
bgr[v0:v1, u0:u1] = eye_image_raw
bgr[v1:v2, u0:u1] = eye_image_annotated
# Visualize preprocessing results
frame_landmarks = (frame['smoothed_landmarks']
if 'smoothed_landmarks' in frame else
frame['landmarks'])
for f, face in enumerate(frame['faces']):
try:
for landmark in frame_landmarks[f][:-1]:
cv.drawMarker(
bgr,
tuple(np.round(landmark).astype(np.int32)),
color=(0, 0, 255),
markerType=cv.MARKER_STAR,
markerSize=2,
thickness=1,
line_type=cv.LINE_AA)
except IndexError:
print(
f"Catch index error on {frame['frame_index']}")
pass
cv.rectangle(
bgr,
tuple(np.round(face[:2]).astype(np.int32)),
tuple(
np.round(np.add(face[:2],
face[2:])).astype(np.int32)),
color=(0, 255, 255),
thickness=1,
lineType=cv.LINE_AA,
)
# Transform predictions
eye_landmarks = np.concatenate([
eye_landmarks,
[[
eye_landmarks[-1, 0] + eye_radius,
eye_landmarks[-1, 1]
]]
])
eye_landmarks = np.asmatrix(
np.pad(eye_landmarks, ((0, 0), (0, 1)),
'constant',
constant_values=1.0))
eye_landmarks = (
eye_landmarks *
eye['inv_landmarks_transform_mat'].T)[:, :2]
eye_landmarks = np.asarray(eye_landmarks)
eyelid_landmarks = eye_landmarks[0:8, :]
iris_landmarks = eye_landmarks[8:16, :]
iris_centre = eye_landmarks[16, :]
eyeball_centre = eye_landmarks[17, :]
eyeball_radius = np.linalg.norm(eye_landmarks[18, :] -
eye_landmarks[17, :])
# Smooth and visualize gaze direction
num_total_eyes_in_frame = len(frame['eyes'])
if len(all_gaze_histories) != num_total_eyes_in_frame:
all_gaze_histories = [
list() for _ in range(num_total_eyes_in_frame)
]
gaze_history = all_gaze_histories[eye_index]
if can_use_eye:
# Visualize landmarks
cv.drawMarker( # Eyeball centre
bgr,
tuple(np.round(eyeball_centre).astype(np.int32)),
color=(0, 255, 0),
markerType=cv.MARKER_CROSS,
markerSize=4,
thickness=1,
line_type=cv.LINE_AA,
)
# cv.circle( # Eyeball outline
# bgr, tuple(np.round(eyeball_centre).astype(np.int32)),
# int(np.round(eyeball_radius)), color=(0, 255, 0),
# thickness=1, lineType=cv.LINE_AA,
# )
# Draw "gaze"
# from models.elg import estimate_gaze_from_landmarks
# current_gaze = estimate_gaze_from_landmarks(
# iris_landmarks, iris_centre, eyeball_centre, eyeball_radius)
i_x0, i_y0 = iris_centre
e_x0, e_y0 = eyeball_centre
theta = -np.arcsin(
np.clip((i_y0 - e_y0) / eyeball_radius, -1.0, 1.0))
phi = np.arcsin(
np.clip((i_x0 - e_x0) /
(eyeball_radius * -np.cos(theta)), -1.0,
1.0))
current_gaze = np.array([theta, phi])
gaze_history.append(current_gaze)
gaze_history_max_len = 10
if len(gaze_history) > gaze_history_max_len:
gaze_history = gaze_history[-gaze_history_max_len:]
util.gaze.draw_gaze(bgr,
iris_centre,
np.mean(gaze_history, axis=0),
length=120.0,
thickness=1)
else:
gaze_history.clear()
if can_use_eyelid:
cv.polylines(
bgr,
[
np.round(eyelid_landmarks).astype(
np.int32).reshape(-1, 1, 2)
],
isClosed=True,
color=(255, 255, 0),
thickness=1,
lineType=cv.LINE_AA,
)
if can_use_iris:
cv.polylines(
bgr,
[
np.round(iris_landmarks).astype(
np.int32).reshape(-1, 1, 2)
],
isClosed=True,
color=(0, 255, 255),
thickness=1,
lineType=cv.LINE_AA,
)
cv.drawMarker(
bgr,
tuple(np.round(iris_centre).astype(np.int32)),
color=(0, 255, 255),
markerType=cv.MARKER_CROSS,
markerSize=4,
thickness=1,
line_type=cv.LINE_AA,
)
dtime = 1e3 * (time.time() - start_time)
if 'visualization' not in frame['time']:
frame['time']['visualization'] = dtime
else:
frame['time']['visualization'] += dtime
def _dtime(before_id, after_id):
return int(1e3 * (frame['time'][after_id] -
frame['time'][before_id]))
def _dstr(title, before_id, after_id):
return '%s: %dms' % (title, _dtime(
before_id, after_id))
if eye_index == len(frame['eyes']) - 1:
# Calculate timings
frame['time']['after_visualization'] = time.time()
fps = int(
np.round(1.0 / (time.time() - last_frame_time)))
fps_history.append(fps)
if len(fps_history) > 60:
fps_history = fps_history[-60:]
fps_str = '%d FPS' % np.mean(fps_history)
last_frame_time = time.time()
fh, fw, _ = bgr.shape
cv.putText(bgr,
fps_str,
org=(fw - 110, fh - 20),
fontFace=cv.FONT_HERSHEY_DUPLEX,
fontScale=0.8,
color=(0, 0, 0),
thickness=1,
lineType=cv.LINE_AA)
cv.putText(bgr,
fps_str,
org=(fw - 111, fh - 21),
fontFace=cv.FONT_HERSHEY_DUPLEX,
fontScale=0.79,
color=(255, 255, 255),
thickness=1,
lineType=cv.LINE_AA)
if not args.headless:
cv.imshow('vis', bgr)
last_frame_index = frame_index
# Record frame?
if args.record_video:
video_out_queue.put_nowait(frame_index)
# Quit?
if cv.waitKey(1) & 0xFF == ord('q'):
return
# Print timings
if frame_index % 60 == 0:
latency = _dtime('before_frame_read',
'after_visualization')
processing = _dtime('after_frame_read',
'after_visualization')
timing_string = ', '.join([
_dstr('read', 'before_frame_read',
'after_frame_read'),
_dstr('preproc', 'after_frame_read',
'after_preprocessing'),
'infer: %dms' %
int(frame['time']['inference']),
'vis: %dms' %
int(frame['time']['visualization']),
'proc: %dms' % processing,
'latency: %dms' % latency,
])
print('%08d [%s] %s' %
(frame_index, fps_str, timing_string))
myMatcher = FeatureMatcher(features1,cornerIndexOrig[0:10],features2,cornerIndexRotated[0:10])
matchingPoints=myMatcher.getMatchingPoints()
print "matching Points " , type(matchingPoints),len(matchingPoints)
print "matching Points " , matchingPoints[0],len(matchingPoints[0])
print "matching Points " , matchingPoints[1],len(matchingPoints[1])
m1 = matchingPoints[0]
m2 = matchingPoints[1]
cornerTest=m1.tolist()
cornerTest2=m2
for index in cornerTest:
x = index[1]
y = index[0]
cv2.drawMarker( imageColor, (x,y), (255,0,0), markerSize=30 )
cv2.namedWindow('image',cv2.WINDOW_NORMAL)
cv2.imshow('image', imageColor)
for index in cornerTest2:
x = index[1]
y = index[0]
cv2.drawMarker( imageRotatedColor, (x,y), (0,255,0), markerSize=30 )
cv2.namedWindow('image Rotated',cv2.WINDOW_NORMAL)
cv2.imshow('image Rotated', imageRotatedColor)
# test matching
# matchedFeatures = myMatcher.getMatchingFeatures()
"position": [30, 300],
"initialSize": 100,
"initialThickness": 5,
"brigntness": 250,
"velocity": 4
}
shortestLenhth2border = min(videoParam["size"][0] - marker["position"][0],
marker["position"][0],
videoParam["size"][1] - marker["position"][1],
marker["position"][1])
for t in range(0, videoParam["time"] * videoParam["fps"]):
img = np.zeros(videoParam["size"], dtype=np.uint8)
if t == 0:
size = marker["initialSize"]
thic = marker["initialThickness"]
marker["position"][0] += marker["velocity"]
if marker["position"][0] >= videoParam["size"][0] - marker["initialSize"]:
print("Warning: Graph is out of border. Interrupted.")
break
cv2.drawMarker(img, tuple(marker["position"]), marker["brigntness"],
marker["type"], int(size), int(thic))
outImg = img[verticalCropPoint[0]:verticalCropPoint[1],
horizontalCropPoint[0]:horizontalCropPoint[1]]
out.write(cv2.cvtColor(outImg, cv2.COLOR_GRAY2BGR))
cv2.imshow("test", outImg)
cv2.waitKey(10)
out.release()
cv2.destroyAllWindows()
imageColor = cv2.imread(imageFileName)
imageGray = cv2.imread(imageFileName, 0)
# get current time before harris
start = cv2.getTickCount()
harrisCorners = HarrisCorner( 1, imageGray )
harrisCorners.findCorners( 100000)
cornerIndex = harrisCorners.localMaxima( 25)
# get current time after harris
end = cv2.getTickCount()
time = (end - start)/ cv2.getTickFrequency()
print "Time taken by harris" , time , "seconds"
print imageGray.shape, cornerIndex.shape
# print cornerIndex
for index in cornerIndex:
x = index[1]
y = index[0]
cv2.drawMarker( imageColor, (x,y), (0,0,0), markerSize=3 )
cv2.namedWindow('image',cv2.WINDOW_NORMAL)
cv2.imshow('image', imageColor)
cv2.waitKey(0)
cv2.destroyAllWindows()
def ColorDetector(Color):
thresholdValueGreen = 50
thresholdValueRed = 50
thresholdValueYellow = 50
thresholdValueBlue = 65
# Read image
frame = cv2.imread('/home/ros/catkin_repo/ros/src/grp6_proj/nodes/top1')
#resize frame
#frame = cv2.resize(frame, (640,480))
#convert til HSV colorspace
hsv = cv2.cvtColor(frame,cv2.COLOR_BGR2HSV)
# Show image on screen
cv2.imshow('frame', frame)
#wait untill keypress
cv2.waitKey()
#find color green
if Color == 'green':
# Range for green
#Hue has value from 0-180 (value from Photoshop/2)
#Saturation 255 is full color, 0 is white (photoshop values is in %)
#Value/brightness interval from 0-255. 255 is full color, 0 is black (photoshop values is in %)
green_lower = np.array([40,65,100])
green_upper = np.array([75,255,255])
mask_green = cv2.inRange(hsv, green_lower, green_upper)
exr = cv2.bitwise_and(frame, frame, mask=mask_green)
cv2.imshow("Frame", exr)
#HSVtoGRAY(exr)
exr = cv2.cvtColor(exr,cv2.COLOR_HSV2BGR)
exr = cv2.cvtColor(exr, cv2.COLOR_BGR2GRAY)
#Threshold for hvornaar farven bliver talt med
thresholded = ((exr>thresholdValueGreen)*255).astype('uint8')
elif Color == 'red':
# Range for lower red
red_lower = np.array([0,150,100])
red_upper = np.array([10,255,255])
mask_red1 = cv2.inRange(hsv, red_lower, red_upper)
# Range for upper range
red_lower = np.array([170,150,50])
red_upper = np.array([180,255,255])
mask_red2 = cv2.inRange(hsv, red_lower, red_upper)
mask_red = mask_red1 + mask_red2
exr = cv2.bitwise_and(frame, frame, mask=mask_red)
cv2.imshow("Frame", exr)
exr = cv2.cvtColor(exr,cv2.COLOR_HSV2BGR)
exr = cv2.cvtColor(exr, cv2.COLOR_BGR2GRAY)
thresholded = ((exr>thresholdValueRed)*255).astype('uint8')
elif Color == 'blue':
# Range for blue
blue_lower = np.array([105,50,50])
blue_upper = np.array([125,255,255])
mask_blue = cv2.inRange(hsv, blue_lower, blue_upper)
exr = cv2.bitwise_and(frame, frame, mask=mask_blue)
cv2.imshow("Frame", exr)
exr = cv2.cvtColor(exr,cv2.COLOR_HSV2BGR)
exr = cv2.cvtColor(exr, cv2.COLOR_BGR2GRAY)
thresholded = ((exr>thresholdValueBlue)*255).astype('uint8')
elif Color == 'yellow':
# Range for yellow
yellow_lower = np.array([25,80,20])
yellow_upper = np.array([35,255,255])
mask_yellow = cv2.inRange(hsv, yellow_lower, yellow_upper)
exr = cv2.bitwise_and(frame, frame, mask=mask_yellow)
cv2.imshow("Frame", exr)
exr = cv2.cvtColor(exr,cv2.COLOR_HSV2BGR)
exr = cv2.cvtColor(exr, cv2.COLOR_BGR2GRAY)
thresholded = ((exr>thresholdValueYellow)*255).astype('uint8')
else:
print('***Please choose a valid color***')
# show those areas
#cv2.imshow('exrWithThreshold', thresholded)
cv2.waitKey()
#dette kode er taget fra undervisningen - Skrevet af: Mads Dyrmann and Henrik Skov Midtiby
# Find connected components
_, contours, _ = cv2.findContours(thresholded, cv2.RETR_EXTERNAL, cv2.CHAIN_APPROX_SIMPLE)
# loop through all components, one at the time
for cnt in contours:
# Whats the area of the component?
areal = cv2.contourArea(cnt)
# Gr noget hvis arealet er strre end 1.
if(areal > 1):
# get the center of mass
M = cv2.moments(cnt)
cx = int(M['m10']/M['m00'])
cy = int(M['m01']/M['m00'])
#print(cx, cy)
#coordinates = "x: %s, y: %d" % (cx,cy)
#alternativ
# cx=sum(cnt[:,0][:,0])/len(cnt[:,0][:,0])
# cy=sum(cnt[:,0][:,1])/len(cnt[:,0][:,1])
#Print the coordinates that are within our limits/table
A1 = (cy>34 and cy<310) and (cx>1 and cx<237)
A2 = (cy>34 and cy<310) and (cx>381 and cx<600)
A3 = (cy>34 and cy<233) and (cx>237 and cx<381)
if (A1 or A2 or A3):
print (cx)
print (cy)
#print ('x:', cx,'y:', cy)
# Tegn et sigtekorn p billedet, der markerer elementet.
# <alter these lines>
frame = cv2.drawMarker(frame, (cx, cy), (255,0,255),markerType=cv2.MARKER_CROSS, thickness=2)
# </alter these lines>
#viser billede med kryds p
cv2.imshow('frame annotated', frame)
cv2.waitKey(0)
return