def get_lines(img): height, width, depth = img.shape scale = SCALE_SIZE / height small_img = np.zeros((height*scale,width*scale,3), np.uint8) cv2.resize(img, small_img) small_img_gray = cv2.cvtColor(small_img,cv2.COLOR_BGR2GRAY) cv2.canny(small_img_gray, small_img_gray, 50, 200, 3); lines = cv2.HoughLines(small_img_gray,OpenCV::CV_HOUGH_PROBABILISTIC, 1, CV_PI / 2, 50, 32, 16) #lines = cv2.HoughLines(small_img_gray,OpenCV::CV_HOUGH_PROBABILISTIC, 1, 1.57, 50, 50, 32 ) return lines
print('Cant load Cascade')
for(x1,y1,x2,y2) in pallet:
#draw rectangle aroud detected pallet
cv2.rectangle(img, (x1,y1), (x1+x2, y1+y2), (0,255,0), 5)
font = cv2.FONT_HERSHEY_SIMPLEX
cv2.putText(img, 'Pallet', (x2+10, y2+10), font, 1, (0,255,0), 3, cv2.LINE_AA
imshow('Obj Det', img)
# pallet is found
if len(pallet) != 0:
#cascade can be release somehow
#perhaps first do canny then get the roi with the next good lines to the detected objects
palletAndTruckROI
# do canny detection to get some clean pics
canny = cv2.canny(blankImage, 100, 200)
cv2.contours
#here we have to get the pixel value of the upper line of the Truck so we can get the length from the ground to the pallet height
getHeightOfPallet()
#getForkToTheRightHeight
driveTheWayToPallet()
#set up the motor for the right time
#release everything not needed
cap.release()
break
# if no pallet is found
else:
#drive a bit forward and try again to recognize it
#octosonarDontCrash
BP.set_motor_power(BP.PORT_B, 20)
def canny(img, low_threshold, high_threshold):
'''canny边缘检测'''
return cv2.canny(low_threshold, high_threshold)
'''
takes inputs from josm, runs edge detection and
outputs ways back to josm
'''
# grab inputs from josm
lat = argv[1]
long = argv[2]
zoom = argv[3]
# grab sat view from provider (likely bing maps)
bing_api_key = ""
# construct request
# https://dev.virtualearth.net/REST/v1/Imagery/Map/Aerial/lat, long/zoom?mapSize=500,500&key=bing_api_key
img = requests.get()
# Reads image from specified path
img = cv2.imread(
file, 0) # 0 arg imports image as gray scale as needed for canny function
# cv2.imshow()
# run canny edge detection algorithm
edges = cv2.canny(img, min_val,
max_val) # image, min, max. aperture, bool edge grad.
# post process edges to remove extras
# filter out areas too small
# filter incomplete ways
# filter shadows
# orthagonalize
# convert to correct format for osm
# send back to josm
import sys
try:
fn = sys.argv[1]
except:
fn = video.presets['chess']
def nothing(*arg):
pass
cv2.namedWindow('edge')
cv2.createTrackbar('thrs1', 'edge', 2000, 5000, nothing)
cv2.createTrackbar('thrs2', 'edge', 4000, 5000, nothing)
cap = video.create_capture(fn)
while True:
flag, img = cap.read()
gray = cv2.cvtColor(img, cv.CV_BGR2GRAY)
thrs1 = cv2.getTrackbarPos('thrs1', 'edge')
thrs2 = cv2.getTrackbarPos('thrs2', 'edge')
edge = cv2.canny(gray, thrs1, thrs2, apertureSize=5)
vis = img.copy()
vis /= 2
vis[edge != 0] = (0, 255, 0)
cv2.imshow('edge', vis)
ch = cv2.waitKey(5)
if ch == 27:
break
def EdgeDetection(img):
width = 400
height = 400
dim = (width, height)
cv2.resize(img, (800, 800))
return cv2.canny(img, 30, 50)
def bgcalc():
im = origin
im = cv2.canny(300, 600, 5, true)
import cv2, cv
import video
import sys
try: fn = sys.argv[1]
except: fn = video.presets['chess']
def nothing(*arg):
pass
cv2.namedWindow('edge')
cv2.createTrackbar('thrs1', 'edge', 2000, 5000, nothing)
cv2.createTrackbar('thrs2', 'edge', 4000, 5000, nothing)
cap = video.create_capture(fn)
while True:
flag, img = cap.read()
gray = cv2.cvtColor(img, cv.CV_BGR2GRAY)
thrs1 = cv2.getTrackbarPos('thrs1', 'edge')
thrs2 = cv2.getTrackbarPos('thrs2', 'edge')
edge = cv2.canny(gray, thrs1, thrs2, apertureSize=5)
vis = img.copy()
vis /= 2
vis[edge != 0] = (0, 255, 0)
cv2.imshow('edge', vis)
ch = cv2.waitKey(5)
if ch == 27:
break
object_detector = ObjectDetector(net, detector, transform, num_classes, args.cuda)
txt = open("F:/数据备份/大四上（电脑）/大创1/task_xuedi（pytorch测试用）/android_result.txt", "w")
image_save = []
'''for image_id in range(len(demo_set)):
image_per_starttime = time.time()
print('##### Start display #####')
# image_file = os.path.join(args.save_folder, demo_set[image_id])
image_file = os.path.join('F:/数据备份/大四上（电脑）/大创1/task_xuedi（pytorch测试用）/test_inter/', demo_set[image_id])
print(image_file)
image = cv2.imdecode(np.fromfile(image_file, dtype=np.uint8), cv2.IMREAD_COLOR)'''
image = cv2.imdecode(np.fromfile('F:/数据备份/大四上（电脑）/大创1/task_xuedi（pytorch测试用）/test_inter/test.jpg', dtype=np.uint8), cv2.IMREAD_COLOR)
detect_bboxes = object_detector.predict(image)
image = cv2.GaussianBlur(image, (3, 3), 0)
canny = cv2.canny(image, 60, 100)
ret, thresh = cv2.threshold(canny, 128, 1, cv2.THRESH_BINARY)
summ = thresh.sum(axis = 1)
def smooth(a,WSZ):
# a:原始数据，NumPy 1-D array containing the data to be smoothed
# 必须是1-D的，如果不是，请使用 np.ravel()或者np.squeeze()转化
# WSZ: smoothing window size needs, which must be odd number,
# as in the original MATLAB implementation
out0 = np.convolve(a,np.ones(WSZ,dtype=int),'valid')/WSZ
r = np.arange(1,WSZ-1,2)
start = np.cumsum(a[:WSZ-1])[::2]/r
stop = (np.cumsum(a[:-WSZ:-1])[::2]/r)[::-1]
return np.concatenate(( start , out0, stop ))
sm = smooth(summ, 19)
peaks, _ = signal.find_peaks(-sm, distance=80)
