def __findContour(self, filename): #find the contour of images, and save all points in self.vKeyPoints
self.img = highgui.cvLoadImage (filename)
self.grayimg = cv.cvCreateImage(cv.cvSize(self.img.width, self.img.height), 8,1)
self.drawimg = cv.cvCreateImage(cv.cvSize(self.img.width, self.img.height), 8,3)
cv.cvCvtColor (self.img, self.grayimg, cv.CV_BGR2GRAY)
cv.cvSmooth(self.grayimg, self.grayimg, cv.CV_BLUR, 9)
cv.cvSmooth(self.grayimg, self.grayimg, cv.CV_BLUR, 9)
cv.cvSmooth(self.grayimg, self.grayimg, cv.CV_BLUR, 9)
cv.cvThreshold( self.grayimg, self.grayimg, self.threshold, self.threshold +100, cv.CV_THRESH_BINARY )
cv.cvZero(self.drawimg)
storage = cv.cvCreateMemStorage(0)
nb_contours, cont = cv.cvFindContours (self.grayimg,
storage,
cv.sizeof_CvContour,
cv.CV_RETR_LIST,
cv.CV_CHAIN_APPROX_NONE,
cv.cvPoint (0,0))
cv.cvDrawContours (self.drawimg, cont, cv.cvScalar(255,255,255,0), cv.cvScalar(255,255,255,0), 1, 1, cv.CV_AA, cv.cvPoint (0, 0))
self.allcurve = []
idx = 0
for c in cont.hrange():
PointArray = cv.cvCreateMat(1, c.total , cv.CV_32SC2)
PointArray2D32f= cv.cvCreateMat( 1, c.total , cv.CV_32FC2)
cv.cvCvtSeqToArray(c, PointArray, cv.cvSlice(0, cv.CV_WHOLE_SEQ_END_INDEX))
fpoints = []
for i in range(c.total):
kp = myPoint()
kp.x = cv.cvGet2D(PointArray,0, i)[0]
kp.y = cv.cvGet2D(PointArray,0, i)[1]
kp.index = idx
idx += 1
fpoints.append(kp)
self.allcurve.append(fpoints)
self.curvelength = idx
def sync(self):
"""
Synchronizes the Capture image with the Camera image
Arguments:
- self: The main object pointer.
"""
self.__camera.query_image()
if not self.__image:
self.__images_cn = { 1 : cv.cvCreateImage ( self.__camera.imgSize, 8, 1 ),
3 : cv.cvCreateImage ( self.__camera.imgSize, 8, 3 ),
4 : cv.cvCreateImage ( self.__camera.imgSize, 8, 4 ) }
self.__color = "bgr"
self.__image_orig = self.__image = self.__camera.img
if self.__color != self.__color_set:
self.__image = self.color("rgb")
# TODO: Workaround, I've to fix it
if len(self.__camera.img_lkpoints["last"]) > 0:
self.__camera.show_lkpoints()
self.__camera.swap_lkpoints()
self.show_rectangles(self.rectangles())
self.__image = self.resize(200, 160)
return self.async
def evalCurrentImageGradient(self):
key = {}
try:
self._cnt.emit(qt.PYSIGNAL("getImage"), (key, ))
qimage = key['image']
except KeyError:
return
##EVAL IMAGE GRADIENT
x, y = self._cnt.focusPointSelected
rectangleSize = self._cnt.focusRectangleSize
im = qimage.copy(x, y, x + rectangleSize, y + rectangleSize)
srcColorImage = qtTools.getImageOpencvFromQImage(im)
if srcColorImage.nChannels > 1:
srcImage = cv.cvCreateImage(
cv.cvSize(srcColorImage.width, srcColorImage.height),
srcColorImage.depth, 1)
cv.cvCvtColor(srcColorImage, srcImage, cv.CV_RGB2GRAY)
else: # In Fact It's a grey image
srcImage = srcColorImage
destImage = cv.cvCreateImage(
cv.cvSize(srcImage.width, srcImage.height), cv.IPL_DEPTH_16S, 1)
cv.cvSobel(srcImage, destImage, 1, 0, 3)
array = numpy.fromstring(destImage.imageData_get(), dtype=numpy.int16)
focusQuality = array.std()
return focusQuality
def depthmatrix(leftimage, rightimage, precision=4, mask=0):
"""Returns a 3-channel 32bit floating-point distance matrix. Channels 1,2,3 = x,y,z coordinates of that point.
Precision is the number of times to downsample mask. Downsample is the number of loops to
go through with successively smaller match areas. If mask is set, only pixels in the mask are set."""
info = cv.cvGetSize(leftimage)
width = info.width
height = info.height
precision_pixels = (2**precision)
downsampled_size = cv.cvSize(width/precision_pixels, height/precision_pixels)
print "Precision of", downsampled_size.width, downsampled_size.height, "px"
if mask:
downsampled_mask = cv.cvCreateImage(downsampled_size, 8, 1)
cv.cvResize(mask, downsampled_mask)
matx = cv.cvCreateImage(downsampled_size, 8, 1)
maty = cv.cvCreateImage(downsampled_size, 8, 1)
matz = cv.cvCreateImage(downsampled_size, 8, 1)
for i in xrange(width/precision_pixels):
for j in xrange(height/precision_pixels):
if mask:
if (not cv.cvGetReal2D(downsampled_mask, j, i)):
continue
x = i*precision
y = j*precision
depth = depthmatch(x+precision_pixels/2, y+precision_pixels/2, leftimage, rightimage, roi=precision_pixels, buf=precision_pixels*2)
#print i, j
# fill in result matrix if mask wasn't 0 at this point (X,Y,Z)
cv.cvSetReal2D(matx, j, i, int(depth[0][0]))
cv.cvSetReal2D(maty, j, i, int(depth[0][1]))
cv.cvSetReal2D(matz, j, i, int(depth[0][2]))
return matz
def analyzeImage(original): scaleImage = cv.cvCreateImage(cv.cvSize(int(original.width*scale), int(original.height*scale)), 8, 3) cv.cvResize(original, scaleImage) # Create 1-channel image for the egdes edgeImage = cv.cvCreateImage(cv.cvGetSize(scaleImage), 8, 1) # Retrieve edges edgeDetector.findBWEdges(scaleImage, edgeImage, edgeThreshold1, edgeThreshold2) # Get cuts cuts = lib.findGoldenMeans(cv.cvGetSize(scaleImage)) # Run along allComponents = [] for cut in cuts: cutComponents = analyzeCut(scaleImage, edgeImage, cut) allComponents.append(cutComponents) # Get the collected component_dictionaries for dict in allComponents: lib.drawBoundingBoxes(original, dict, scale) # Draw the margins for cut in cuts: lib.drawMargin(original, cut, margin, scale) #include if super margen is need to drawn #lib.drawMargin(original, cut, superMargin, scale) return (original, allComponents)
def findEdges(original, out, threshold1 = 100, threshold2 = None):
"""Return a new edge detected image with a specified threshold"""
warnings.warn("Use findBWEdges instead unless you really need colored edges.", DeprecationWarning)
#Define threshold2
if threshold2 == None:
threshold2 = threshold1 * 3
# Create two pictures with only one channel for a b/w copy
# and one for storring the edges found in the b/w picture
gray = cv.cvCreateImage(cv.cvGetSize(original), 8, 1)
edge = cv.cvCreateImage(cv.cvGetSize(original), 8, 1)
# Create the b/w copy of the original
cv.cvCvtColor(original, gray, cv.CV_BGR2GRAY)
# Blur the b/w copy, but put the result into edge pic
cv.cvSmooth(gray, edge, cv.CV_BLUR, 3, 3, 0)
# Negate the b/w copy of original with newly blurred
# b/w copy. This will make egdes stand out
cv.cvNot(gray, edge)
# Run an edge-finding algorithm called 'Canny'
# It will analyse the first argument and store the
# resulting picture in the second argument
cv.cvCanny(gray, edge, threshold1, threshold2)
# We initialize our out-image to black
cv.cvSetZero(out)
# Finally, we use the found edges, which are b/w, as
# a mask for copying the colored edges from the original
# to the out-image
cv.cvCopy(original, out, edge)
def analyzeCut(scaleImage, edgeImage, cut): """Extract the interesting features respecting the cut""" # Set up constraints constraints = regionSelector.Constraints(cv.cvGetSize(scaleImage), cut, margin, superMargin, 0.002, 0.25) # Create temporary images blurImage = cv.cvCreateImage(cv.cvGetSize(scaleImage), 8, 3) workImage = cv.cvCreateImage(cv.cvGetSize(scaleImage), 8, 3) # Create a blurred copy of the original cv.cvSmooth(scaleImage, blurImage, cv.CV_BLUR, 3, 3, 0) # Superimpose the edges onto the blured image cv.cvNot(edgeImage, edgeImage) cv.cvCopy(blurImage, workImage, edgeImage) # Get the edges back to white cv.cvNot(edgeImage, edgeImage) # We're done with the blurred image now cv.cvReleaseImage(blurImage) # Retrive the regions touching the cut component_dictionary = featureDetector.ribbonFloodFill(scaleImage, edgeImage, workImage, cut, margin, lo, up) # Clean up cv.cvReleaseImage(workImage) # Prune components newComponents = regionSelector.pruneRegions(component_dictionary, constraints) # Return the dictionary of accepted components #transformer.translateBoundingBoxes(newComponents, 1) return newComponents
def detect_lines(self, img_grey, img_orig):
""" Detect lines within the image. To switch between standard and
probabilistic Hough transform, use cv.CV_HOUGH_STANDARD, or
cv.CV_HOUGH_PROBABILISTIC.
"""
# Set transform method ('standard','probabilistic')
transform_method = 'probabilistic'
# Clear out our storage
cv.cvClearMemStorage(self.lines_storage)
sz = cv.cvSize(img_grey.width & -2, img_grey.height & -2)
img_dst_color = cv.cvCreateImage(cv.cvGetSize(img_orig), 8, 3)
tgrey = cv.cvCreateImage(sz, 8, 1)
cv.cvCanny(tgrey, img_grey, 50, 200, 3)
if transform_method == 'standard':
lines = cv.cvHoughLines2(img_grey, self.lines_storage,
cv.CV_HOUGH_STANDARD, 1, cv.CV_PI / 180,
100, 0, 0)
else:
lines = cv.cvHoughLines2(img_grey, self.lines_storage,
cv.CV_HOUGH_PROBABILISTIC, 1,
cv.CV_PI / 180, 50, 50, 10)
return lines
def _cv_to_pygame(self,frame,channel=-1) : # scale the image to size of the window cvt_scale = cv.cvCreateImage(cv.cvSize(self.image_dims[0],self.image_dims[1]),frame.depth,frame.nChannels) #cv.cvResize(frame,cvt_scale,cv.CV_INTER_LINEAR) cv.cvResize(frame,cvt_scale,cv.CV_INTER_NN) # need to convert the colorspace differently depending on where the image came from cvt_color = cv.cvCreateImage(cv.cvSize(cvt_scale.width,cvt_scale.height),cvt_scale.depth,3) if frame.nChannels == 3 : # frame is in BGR format, convert it to RGB so the sky isn't orange cv.cvCvtColor(cvt_scale,cvt_color,cv.CV_BGR2RGB) elif frame.nChannels == 1 : # image has only one channel, iow 1 color if channel == 0 : cv.cvMerge(frame,None,None,None,cvt_color) elif channel == 1 : cv.cvMerge(None,frame,None,None,cvt_color) elif channel == 2 : cv.cvMerge(None,None,frame,None,cvt_color) elif channel == 3 : cv.cvMerge(None,None,None,frame,cvt_color) else : cv.cvCvtColor(cvt_scale,cvt_color,cv.CV_GRAY2RGB) # create a pygame surface frame_surface=pygame.image.frombuffer(cvt_color.imageData,self.image_dims,'RGB') return frame_surface
def detect_lines(self, img_grey, img_orig):
""" Detect lines within the image. To switch between standard and
probabilistic Hough transform, use cv.CV_HOUGH_STANDARD, or
cv.CV_HOUGH_PROBABILISTIC.
"""
# Set transform method ('standard','probabilistic')
transform_method = 'probabilistic'
# Clear out our storage
cv.cvClearMemStorage(self.lines_storage)
sz = cv.cvSize(img_grey.width & -2, img_grey.height & -2)
img_dst_color = cv.cvCreateImage(cv.cvGetSize(img_orig), 8, 3)
tgrey = cv.cvCreateImage(sz, 8, 1)
cv.cvCanny(tgrey, img_grey, 50, 200, 3)
if transform_method == 'standard':
lines = cv.cvHoughLines2(img_grey,
self.lines_storage,
cv.CV_HOUGH_STANDARD,
1,
cv.CV_PI/180,
100,
0,
0)
else:
lines = cv.cvHoughLines2(img_grey,
self.lines_storage,
cv.CV_HOUGH_PROBABILISTIC,
1,
cv.CV_PI/180,
50,
50,
10)
return lines
def pixelInRange(src, rmin, rmax, floor, roof, dst):
if rmax > rmin: # normal case
cvInRangeS(src, rmin, rmax, dst)
else: # considering range as a cycle
dst0 = cvCreateImage(cvSize(frameWidth, frameHeight), IPL_DEPTH_8U, 1)
dst1 = cvCreateImage(cvSize(frameWidth, frameHeight), IPL_DEPTH_8U, 1)
cvInRangeS(src, floor, rmax, dst0)
cvInRangeS(src, rmin, roof, dst1)
cvOr(dst0, dst1, dst)
def getFilter(frameWidht, frameHeight):
cvNamedWindow("Filtred")
cvCreateTrackbar("hmax", "Filtred", getHlsFilter('hmax'), 180, trackBarChangeHmax)
cvCreateTrackbar("hmin", "Filtred", getHlsFilter('hmin'), 180, trackBarChangeHmin)
#cvCreateTrackbar("lmax", "Filtred", hlsFilter['lmax'], 255, trackBarChangeLmax)
#cvCreateTrackbar("lmin", "Filtred", hlsFilter['lmin'], 255, trackBarChangeLmin)
cvCreateTrackbar("smax", "Filtred", getHlsFilter('smax'), 255, trackBarChangeSmax)
cvCreateTrackbar("smin", "Filtred", getHlsFilter('smin'), 255, trackBarChangeSmin)
cvSetMouseCallback("Filtred", mouseClick, None)
frame = cvCreateImage(cvSize(frameWidth, frameHeight), IPL_DEPTH_8U, 3)
hlsFrame = cvCreateImage(cvSize(frameWidth, frameHeight), IPL_DEPTH_8U, 3)
filtredFrame = cvCreateImage(cvSize(frameWidth, frameHeight), IPL_DEPTH_8U, 3)
mask = cvCreateImage(cvSize(frameWidth, frameHeight), IPL_DEPTH_8U, 1)
hFrame = cvCreateImage(cvSize(frameWidth, frameHeight), IPL_DEPTH_8U, 1)
lFrame = cvCreateImage(cvSize(frameWidth, frameHeight), IPL_DEPTH_8U, 1)
sFrame = cvCreateImage(cvSize(frameWidth, frameHeight), IPL_DEPTH_8U, 1)
ThHFrame = cvCreateImage(cvSize(frameWidth, frameHeight), IPL_DEPTH_8U, 1)
ThLFrame = cvCreateImage(cvSize(frameWidth, frameHeight), IPL_DEPTH_8U, 1)
ThSFrame = cvCreateImage(cvSize(frameWidth, frameHeight), IPL_DEPTH_8U, 1)
key = -1
while key == -1:
if not cvGrabFrame(CAM):
print "Could not grab a frame"
exit
frame = cvQueryFrame(CAM)
cvCvtColor(frame, hlsFrame, CV_BGR2HLS)
cvSplit(hlsFrame, hFrame, lFrame, sFrame, None)
pixelInRange(hFrame, getHlsFilter('hmin'), getHlsFilter('hmax'), 0, 180, ThHFrame)
#pixelInRange(lFrame, getHlsFilter('lmin'), getHlsFilter('lmax'), 0, 255, ThLFrame)
pixelInRange(sFrame, getHlsFilter('smin'), getHlsFilter('smax'), 0, 255, ThSFrame)
cvSetZero(mask)
cvAnd(ThHFrame, ThSFrame, mask)
cvSetZero(filtredFrame)
cvCopy(frame, filtredFrame, mask)
cvShowImage("Filtred", filtredFrame)
key = cvWaitKey(10)
if key == 'r':
key = -1
resetHlsFilter()
cvDestroyWindow("Filtred")
def __FindHarris(self, filename): #find the corners of images, and save all corner points in self.vKeyPoints
self.img = highgui.cvLoadImage (filename)
greyimg = cv.cvCreateImage(cv.cvSize(self.img.width, self.img.height), 8,1)
w = cv.cvGetSize(self.img).width
h = cv.cvGetSize(self.img).height
image = cv.cvCreateImage(cv.cvGetSize(self.img), cv.IPL_DEPTH_32F, 1)
cv.cvConvert(image, greyimg)
self.cornerimg = cv.cvCreateImage(cv.cvGetSize(self.img), cv.IPL_DEPTH_32F, 1)
cv.cvCornerHarris(image, self.cornerimg, 11,5,0.1)
def _detect(image):
""" Detects faces on `image`
Parameters:
@image: image file path
Returns:
[((x1, y1), (x2, y2)), ...] List of coordenates for top-left
and bottom-right corner
"""
# the OpenCV API says this function is obsolete, but we can't
# cast the output of cvLoad to a HaarClassifierCascade, so use
# this anyways the size parameter is ignored
capture = cvCreateFileCapture(image)
if not capture:
return []
frame = cvQueryFrame(capture)
if not frame:
return []
img = cvCreateImage(cvSize(frame.width, frame.height),
IPL_DEPTH_8U, frame.nChannels)
cvCopy(frame, img)
# allocate temporary images
gray = cvCreateImage((img.width, img.height),
COPY_DEPTH, COPY_CHANNELS)
width, height = (cvRound(img.width / IMAGE_SCALE),
cvRound(img.height / IMAGE_SCALE))
small_img = cvCreateImage((width, height), COPY_DEPTH, COPY_CHANNELS)
# convert color input image to grayscale
cvCvtColor(img, gray, CV_BGR2GRAY)
# scale input image for faster processing
cvResize(gray, small_img, CV_INTER_LINEAR)
cvEqualizeHist(small_img, small_img)
cvClearMemStorage(STORAGE)
coords = []
for haar_file in CASCADES:
cascade = cvLoadHaarClassifierCascade(haar_file, cvSize(1, 1))
if cascade:
faces = cvHaarDetectObjects(small_img, cascade, STORAGE, HAAR_SCALE,
MIN_NEIGHBORS, HAAR_FLAGS, MIN_SIZE) or []
for face_rect in faces:
# the input to cvHaarDetectObjects was resized, so scale the
# bounding box of each face and convert it to two CvPoints
x, y = face_rect.x, face_rect.y
pt1 = (int(x * IMAGE_SCALE), int(y * IMAGE_SCALE))
pt2 = (int((x + face_rect.width) * IMAGE_SCALE),
int((y + face_rect.height) * IMAGE_SCALE))
coords.append((pt1, pt2))
return coords
def _detect(image):
""" Detects faces on `image`
Parameters:
@image: image file path
Returns:
[((x1, y1), (x2, y2)), ...] List of coordenates for top-left
and bottom-right corner
"""
# the OpenCV API says this function is obsolete, but we can't
# cast the output of cvLoad to a HaarClassifierCascade, so use
# this anyways the size parameter is ignored
capture = cvCreateFileCapture(image)
if not capture:
return []
frame = cvQueryFrame(capture)
if not frame:
return []
img = cvCreateImage(cvSize(frame.width, frame.height), IPL_DEPTH_8U,
frame.nChannels)
cvCopy(frame, img)
# allocate temporary images
gray = cvCreateImage((img.width, img.height), COPY_DEPTH, COPY_CHANNELS)
width, height = (cvRound(img.width / IMAGE_SCALE),
cvRound(img.height / IMAGE_SCALE))
small_img = cvCreateImage((width, height), COPY_DEPTH, COPY_CHANNELS)
# convert color input image to grayscale
cvCvtColor(img, gray, CV_BGR2GRAY)
# scale input image for faster processing
cvResize(gray, small_img, CV_INTER_LINEAR)
cvEqualizeHist(small_img, small_img)
cvClearMemStorage(STORAGE)
coords = []
for haar_file in CASCADES:
cascade = cvLoadHaarClassifierCascade(haar_file, cvSize(1, 1))
if cascade:
faces = cvHaarDetectObjects(small_img, cascade, STORAGE,
HAAR_SCALE, MIN_NEIGHBORS, HAAR_FLAGS,
MIN_SIZE) or []
for face_rect in faces:
# the input to cvHaarDetectObjects was resized, so scale the
# bounding box of each face and convert it to two CvPoints
x, y = face_rect.x, face_rect.y
pt1 = (int(x * IMAGE_SCALE), int(y * IMAGE_SCALE))
pt2 = (int((x + face_rect.width) * IMAGE_SCALE),
int((y + face_rect.height) * IMAGE_SCALE))
coords.append((pt1, pt2))
return coords
def get_nearest_feature( image, this_point, n=2000 ): """ Get the n-nearest features to a specified image coordinate. Features are determined using cvGoodFeaturesToTrack. """ _red = cv.cvScalar (0, 0, 255, 0); _green = cv.cvScalar (0, 255, 0, 0); _blue = cv.cvScalar (255,0,0,0); _white = cv.cvRealScalar (255) _black = cv.cvRealScalar (0) quality = 0.01 min_distance = 4 N_best = n win_size = 11 grey = cv.cvCreateImage (cv.cvGetSize (image), 8, 1) eig = cv.cvCreateImage (cv.cvGetSize (image), 32, 1) temp = cv.cvCreateImage (cv.cvGetSize (image), 32, 1) # create a grey version of the image cv.cvCvtColor ( image, grey, cv.CV_BGR2GRAY) points = cv.cvGoodFeaturesToTrack ( grey, eig, temp, N_best, quality, min_distance, None, 3, 0, 0.04) # refine the corner locations better_points = cv.cvFindCornerSubPix ( grey, points, cv.cvSize (win_size, win_size), cv.cvSize (-1, -1), cv.cvTermCriteria (cv.CV_TERMCRIT_ITER | cv.CV_TERMCRIT_EPS, 20, 0.03)) eigs = [] for i in range(len(points)): eigs.append(cv.cvGetMat(eig)[int(points[i].y)][int(points[i].x)]) mypoints = np.matrix(np.zeros((len(points)*2),dtype=float)).reshape(len(points),2) dists = [] for i,point in enumerate(points): mypoints[i,0]=point.x mypoints[i,1]=point.y dists.append( np.linalg.norm(mypoints[i,:]-this_point) ) dists = np.array(dists) sorteddists = dists.argsort() cv.cvDrawCircle ( image, points[ sorteddists[0] ], 5, _green, 2, 8, 0 ) return better_points[ sorteddists[0] ]
def texture_features(self, block_size=5, filter_size=3):
"""
Calculates the texture features associated with the image.
block_size gives the size of the texture neighborhood to be processed
filter_size gives the size of the Sobel operator used to find gradient information
"""
#block_size = cv.cvSize(block_size, block_size)
#convert to grayscale float
channels = 1
self.gray_image = cv.cvCreateImage(
cv.cvSize(self.im_width, self.im_height),
cv.IPL_DEPTH_8U, #cv.IPL_DEPTH_16U, #cv.IPL_DEPTH_32F,
channels)
#cv.CV_32FC1, #cv.IPL_DEPTH_32F, #cv.IPL_DEPTH_8U, #cv.IPL_DEPTH_16U,
channels = 1
eig_tex = cv.cvCreateImage(
cv.cvSize(self.im_width * 6, self.im_height), cv.IPL_DEPTH_32F,
channels)
cv.cvCvtColor(self.image, self.gray_image, cv.CV_BGR2GRAY)
#cv.cvAdd(const CvArr* src1, const CvArr* src2, CvArr* dst, const CvArr* mask=NULL );
#highgui.cvConvertImage(self.image, self.gray_image)
cv.cvCornerEigenValsAndVecs(
self.gray_image,
eig_tex, #CvArr* eigenvv,
block_size,
filter_size)
eig_tex = ut.cv2np(eig_tex)
eig_tex = np.reshape(eig_tex, [self.im_height, self.im_width, 6])
#print eig_tex.shape ## [480,640,3]
## (l1, l2, x1, y1, x2, y2), where
## l1, l2 - eigenvalues of M; not sorted
## (x1, y1) - eigenvector corresponding to l1
## (x2, y2) - eigenvector corresponding to l2
tex_feat = np.zeros([3, self.im_height * self.im_width],
dtype=np.float32)
tmp = np.reshape(eig_tex, [self.im_height * self.im_width, 6]).T
s = tmp[0] > tmp[1]
tex_feat[1:3, s] = tmp[0, s] * tmp[2:4, s]
tex_feat[0, s] = tmp[1, s]
tex_feat[1:3, -s] = tmp[1, -s] * tmp[4:6, -s]
tex_feat[0, -s] = tmp[0, -s]
self.tex_feat = tex_feat.T
self.tex_image = np.reshape(self.tex_feat,
[self.im_height, self.im_width, 3])
def harrisResponse(image):
"""pyvision/point/DetectorHarris.py
Runs at 10.5 fps...
"""
gray = cv.cvCreateImage( cv.cvGetSize(image), 8, 1 )
corners = cv.cvCreateImage( cv.cvGetSize(image), 32, 1 )
cv.cvCvtColor( image, gray, cv.CV_BGR2GRAY )
cv.cvCornerHarris(gray,corners,3)
image = filter_and_render_cv(image,corners)
#IPShellEmbed()()
return image
def read(self) :
frame=self.input.read()
if self.debug :
raw_frame = cv.cvCreateImage(cv.cvSize(frame.width,frame.height),frame.depth,frame.nChannels)
cv.cvCopy(frame,raw_frame,None)
self.raw_frame_surface=pygame.image.frombuffer(frame.imageData,(frame.width,frame.height),'RGB')
if self.enabled :
cv_rs = cv.cvCreateImage(cv.cvSize(frame.width,frame.height),frame.depth,1)
# convert color
cv.cvCvtColor(frame,cv_rs,cv.CV_BGR2GRAY)
# invert the image
cv.cvSubRS(cv_rs, 255, cv_rs, None);
# threshold the image
frame = cv.cvCreateImage(cv.cvSize(frame.width,frame.height),frame.depth,1)
cv.cvThreshold(cv_rs, frame, self.threshold, 255, cv.CV_THRESH_BINARY)
if self.debug :
thresh_frame = cv.cvCreateImage(cv.cvSize(frame.width,frame.height),frame.depth,3)
cv.cvCvtColor(frame,thresh_frame,cv.CV_GRAY2RGB)
self.thresh_frame_surface=pygame.image.frombuffer(thresh_frame.imageData,(frame.width,frame.height),'RGB')
# I think these functions are too specialized for transforms
cv.cvSmooth(frame,frame,cv.CV_GAUSSIAN,3, 0, 0, 0 )
cv.cvErode(frame, frame, None, 1)
cv.cvDilate(frame, frame, None, 1)
num_contours,contours=cv.cvFindContours(frame,self.storage,cv.sizeof_CvContour,cv.CV_RETR_LIST,cv.CV_CHAIN_APPROX_NONE,cv.cvPoint(0,0))
if contours is None :
return []
else :
contours = cv.cvApproxPoly( contours, cv.sizeof_CvContour, self.storage, cv.CV_POLY_APPROX_DP, 3, 1 );
if contours is None :
return []
else :
final_contours = []
for c in contours.hrange() :
area = abs(cv.cvContourArea(c))
#self.debug_print('Polygon Area: %f'%area)
if area >= self.min_area :
lst = []
for pt in c :
lst.append((pt.x,pt.y))
final_contours.append(lst)
contours = contours.h_next
return final_contours
return []
def getBackground(frameWidht, frameHeight):
cvNamedWindow("Background")
text = cvCreateImage(cvSize(frameWidth, frameHeight), IPL_DEPTH_8U, 3)
frame = cvCreateImage(cvSize(frameWidth, frameHeight), IPL_DEPTH_8U, 3)
background = cvCreateImage(cvSize(frameWidth, frameHeight), IPL_DEPTH_8U, 3)
font = cvInitFont(CV_FONT_HERSHEY_COMPLEX, 1.0, 1.0, 0.0, 2)
pt1 = cvPoint(50, 100)
pt2 = cvPoint(50, 150)
center = cvPoint(frameWidth/2, frameHeight/2)
cvPutText(text, "Press enter, run away and wait", pt1, font, CV_RGB(150, 100, 150))
cvPutText(text, str(delayS) + " seconds to capture background", pt2, font, CV_RGB(150, 100, 150))
cvShowImage("Background", text)
key = -1
while key == -1:
key = cvWaitKey(10)
like = False
while not like:
for i in range(delayS):
cvZero(text)
cvPutText(text, str(delayS-i), center, font, CV_RGB(150, 100, 150))
cvShowImage("Background", text)
cvWaitKey(1000)
csut = camStartUpTime
while (csut): # Stats capturing frames in order to give time to the cam to auto-adjust colors
if not cvGrabFrame(CAM):
print "Could not grab a frame"
exit
cvWaitKey(10)
csut -= 1
frame = cvQueryFrame(CAM)
cvCopy(frame, background)
cvCopy(frame, text)
cvPutText(text, "Is correct? [y/n]", center, font, CV_RGB(150, 100, 150))
cvShowImage("Background", text)
key = -1
while key != 'n' and key != 'y':
key = cvWaitKey(10)
if key == 'y':
like = True
return background
cvDestroyWindow("Background")
def main(): # ctrl+c to end
global h,s,v,h2,v2,s2,d,e
highgui.cvNamedWindow("Camera 1", 1)
highgui.cvNamedWindow("Orig", 1)
highgui.cvCreateTrackbar("H", "Camera 1", h, 256, tb_h)
highgui.cvCreateTrackbar("S", "Camera 1", s, 256, tb_s)
highgui.cvCreateTrackbar("V", "Camera 1", v, 256, tb_v)
highgui.cvCreateTrackbar("H2", "Camera 1", h2, 256, tb_h2)
highgui.cvCreateTrackbar("S2", "Camera 1", s2, 256, tb_s2)
highgui.cvCreateTrackbar("V2", "Camera 1", v2, 256, tb_v2)
highgui.cvCreateTrackbar("Dilate", "Camera 1", d, 30, tb_d)
highgui.cvCreateTrackbar("Erode", "Camera 1", e, 30, tb_e)
cap = highgui.cvCreateCameraCapture(1)
highgui.cvSetCaptureProperty(cap, highgui.CV_CAP_PROP_FRAME_WIDTH, IMGW)
highgui.cvSetCaptureProperty(cap, highgui.CV_CAP_PROP_FRAME_HEIGHT, IMGH)
c = 0
t1 = tdraw = time.clock()
t = 1
font = cv.cvInitFont(cv.CV_FONT_HERSHEY_PLAIN, 1, 1)
while c != 0x27:
image = highgui.cvQueryFrame(cap)
if not image:
print "capture failed"
break
thresh = cv.cvCreateImage(cv.cvSize(IMGW,IMGH),8,1)
cv.cvSetZero(thresh)
cv.cvCvtColor(image,image,cv.CV_RGB2HSV)
cv.cvInRangeS(image, (h,s,v,0), (h2,s2,v2,0), thresh)
result = cv.cvCreateImage(cv.cvSize(IMGW,IMGH),8,3)
cv.cvSetZero(result)
cv.cvOr(image,image,result,thresh)
for i in range(1,e):
cv.cvErode(result,result)
for i in range(1,d):
cv.cvDilate(result,result)
# floodfill objects back in, allowing threshold differences outwards
t2 = time.clock()
if t2 > tdraw+0.3:
t = t2-t1
tdraw=t2
cv.cvPutText(result, "FPS: " + str(1 / (t)), (0,25), font, (255,255,255))
t1 = t2
highgui.cvShowImage("Orig", image)
highgui.cvShowImage("Camera 1", result)
c = highgui.cvWaitKey(10)
def get_nearest_feature(image, this_point, n=2000):
"""
Get the n-nearest features to a specified image coordinate.
Features are determined using cvGoodFeaturesToTrack.
"""
_red = cv.cvScalar(0, 0, 255, 0)
_green = cv.cvScalar(0, 255, 0, 0)
_blue = cv.cvScalar(255, 0, 0, 0)
_white = cv.cvRealScalar(255)
_black = cv.cvRealScalar(0)
quality = 0.01
min_distance = 4
N_best = n
win_size = 11
grey = cv.cvCreateImage(cv.cvGetSize(image), 8, 1)
eig = cv.cvCreateImage(cv.cvGetSize(image), 32, 1)
temp = cv.cvCreateImage(cv.cvGetSize(image), 32, 1)
# create a grey version of the image
cv.cvCvtColor(image, grey, cv.CV_BGR2GRAY)
points = cv.cvGoodFeaturesToTrack(grey, eig, temp, N_best, quality,
min_distance, None, 3, 0, 0.04)
# refine the corner locations
better_points = cv.cvFindCornerSubPix(
grey, points, cv.cvSize(win_size, win_size), cv.cvSize(-1, -1),
cv.cvTermCriteria(cv.CV_TERMCRIT_ITER | cv.CV_TERMCRIT_EPS, 20, 0.03))
eigs = []
for i in range(len(points)):
eigs.append(cv.cvGetMat(eig)[int(points[i].y)][int(points[i].x)])
mypoints = np.matrix(np.zeros((len(points) * 2),
dtype=float)).reshape(len(points), 2)
dists = []
for i, point in enumerate(points):
mypoints[i, 0] = point.x
mypoints[i, 1] = point.y
dists.append(np.linalg.norm(mypoints[i, :] - this_point))
dists = np.array(dists)
sorteddists = dists.argsort()
cv.cvDrawCircle(image, points[sorteddists[0]], 5, _green, 2, 8, 0)
return better_points[sorteddists[0]]
def __FindCorner(self, filename): #find the corners of images, and save all corner points in self.vKeyPoints
self.img = highgui.cvLoadImage (filename)
greyimg = cv.cvCreateImage(cv.cvSize(self.img.width, self.img.height), 8,1)
hsvimg = cv.cvCreateImage(cv.cvGetSize(self.img), 8, 3)
cv.cvCvtColor(self.img, hsvimg, cv.CV_RGB2HSV)
cv.cvCvtColor (hsvimg, greyimg, cv.CV_BGR2GRAY)
eigImage = cv.cvCreateImage(cv.cvGetSize(greyimg), cv.IPL_DEPTH_32F, 1)
tempImage = cv.cvCreateImage(cv.cvGetSize(greyimg), cv.IPL_DEPTH_32F, 1)
self.points = cv.cvGoodFeaturesToTrack(greyimg, eigImage,tempImage, 2000, 0.01, 5, None, 3,0,0.01 )
self.points2 = cv.cvFindCornerSubPix(greyimg, self.points,cv.cvSize(20, 20),
cv.cvSize(-1, -1), cv.cvTermCriteria(cv.CV_TERMCRIT_ITER |cv.CV_TERMCRIT_EPS, 20, 0.03))
cv.cvReleaseImage(eigImage)
cv.cvReleaseImage(tempImage)
def analyzeCut(original, edgeImage, cut, settings, showBlobs=False): """Extract the interesting features in the vicinity of a given cut""" # Get all data from the settings lo = settings.lo up = settings.up # Set up the margin with respect to the cut margin = marginCalculator.getPixels(original, cut, settings.marginPercentage) superMargin = 0 # ^^ We don't use superMargin # Set up constraints constraints = regionSelector.Constraints(cv.cvGetSize(original), cut, margin, superMargin, 0.002, 0.25) # Create temporary images blurImage = cv.cvCreateImage(cv.cvGetSize(original), 8, 3) workImage = cv.cvCreateImage(cv.cvGetSize(original), 8, 3) # Create a blurred copy of the original cv.cvSmooth(original, blurImage, cv.CV_BLUR, 3, 3, 0) # Superimpose the edges onto the blured image cv.cvNot(edgeImage, edgeImage) cv.cvCopy(blurImage, workImage, edgeImage) # We're done with the blurred image now cv.cvReleaseImage(blurImage) # Get the edges back to white cv.cvNot(edgeImage, edgeImage) # Retrive the regions touching the cut component_dictionary = featureDetector.ribbonFloodFill(original, edgeImage, workImage, cut, margin, lo, up) #start expanded # Prune components BEFORE we delete the workImage tmpnewComponents = regionSelector.pruneExpandedRegions(component_dictionary, constraints) newComponents = regionSelector.pruneExpandedRagionsto(tmpnewComponents, constraints, cut, workImage) # Clean up only if we do not return the image if not showBlobs: cv.cvReleaseImage(workImage) # Return the dictionary of accepted components or both if not showBlobs: return newComponents else: return (workImage, newComponents)
def texture_features(self, block_size=5, filter_size=3):
"""
Calculates the texture features associated with the image.
block_size gives the size of the texture neighborhood to be processed
filter_size gives the size of the Sobel operator used to find gradient information
"""
#block_size = cv.cvSize(block_size, block_size)
#convert to grayscale float
channels = 1
self.gray_image = cv.cvCreateImage(cv.cvSize(self.im_width, self.im_height),
cv.IPL_DEPTH_8U, #cv.IPL_DEPTH_16U, #cv.IPL_DEPTH_32F,
channels)
#cv.CV_32FC1, #cv.IPL_DEPTH_32F, #cv.IPL_DEPTH_8U, #cv.IPL_DEPTH_16U,
channels = 1
eig_tex = cv.cvCreateImage(cv.cvSize(self.im_width*6, self.im_height),
cv.IPL_DEPTH_32F,
channels)
cv.cvCvtColor(self.image, self.gray_image, cv.CV_BGR2GRAY);
#cv.cvAdd(const CvArr* src1, const CvArr* src2, CvArr* dst, const CvArr* mask=NULL );
#highgui.cvConvertImage(self.image, self.gray_image)
cv.cvCornerEigenValsAndVecs(self.gray_image, eig_tex,#CvArr* eigenvv,
block_size, filter_size)
eig_tex = ut.cv2np(eig_tex)
eig_tex = np.reshape(eig_tex, [self.im_height, self.im_width, 6])
#print eig_tex.shape ## [480,640,3]
## (l1, l2, x1, y1, x2, y2), where
## l1, l2 - eigenvalues of M; not sorted
## (x1, y1) - eigenvector corresponding to l1
## (x2, y2) - eigenvector corresponding to l2
tex_feat = np.zeros([3, self.im_height * self.im_width], dtype=np.float32)
tmp = np.reshape(eig_tex, [self.im_height * self.im_width, 6]).T
s = tmp[0] > tmp[1]
tex_feat[1:3, s] = tmp[0, s] * tmp[2:4, s]
tex_feat[0, s] = tmp[1, s]
tex_feat[1:3, -s] = tmp[1, -s] * tmp[4:6, -s]
tex_feat[0, -s] = tmp[0, -s]
self.tex_feat = tex_feat.T
self.tex_image = np.reshape(self.tex_feat, [self.im_height, self.im_width, 3])
def _get_cv_frame(self): frame = CameraInputProvider.get_frame(self) dst = cv.cvCreateImage(cv.cvSize(self.capture_dims[0],self.capture_dims[1]),frame.depth,frame.nChannels) cv.cvWarpPerspective( frame, dst, self.matrix) return dst
def on_trackbar (position):
# create the image for putting in it the founded contours
contours_image = cv.cvCreateImage (cv.cvSize (_SIZE, _SIZE), 8, 3)
# compute the real level of display, given the current position
levels = position - 3
# initialisation
_contours = contours
if levels <= 0:
# zero or negative value
# => get to the nearest face to make it look more funny
_contours = contours.h_next.h_next.h_next
# first, clear the image where we will draw contours
cv.cvSetZero (contours_image)
# draw contours in red and green
cv.cvDrawContours (contours_image, _contours,
_red, _green,
levels, 3, cv.CV_AA,
cv.cvPoint (0, 0))
# finally, show the image
highgui.cvShowImage ("contours", contours_image)
def __normImage(self, img, length):
#print "Generating norm image..."
width = length
height = length
gray = cv.cvCreateImage(cv.cvSize(img.width,img.height), 8, 1);
small_img = cv.cvCreateImage(cv.cvSize(cv.cvRound(width),
cv.cvRound(height)), 8, 1 );
# convert color input image to grayscale
cv.cvCvtColor(img, gray, cv.CV_BGR2GRAY);
# scale input image for faster processing
cv.cvResize(gray, small_img, cv.CV_INTER_LINEAR);
cv.cvEqualizeHist(small_img, small_img);
#cvClearMemStorage(self.storage);
norm_image = small_img # save the 'normalized image'
return norm_image
def detectObject(image):
grayscale = cv.cvCreateImage(size, 8, 1)
cv.cvFlip(image, None, 1)
cv.cvCvtColor(image, grayscale, cv.CV_BGR2GRAY)
storage = cv.cvCreateMemStorage(0)
cv.cvClearMemStorage(storage)
cv.cvEqualizeHist(grayscale, grayscale)
cascade = cv.cvLoadHaarClassifierCascade(haar_file, cv.cvSize(1,1))
objects = cv.cvHaarDetectObjects(grayscale, cascade, storage, 1.2, 2,
cv.CV_HAAR_DO_CANNY_PRUNING,
cv.cvSize(100,100))
# Draw dots where hands are
if objects:
for i in objects:
#cv.cvRectangle(image, cv.cvPoint( int(i.x), int(i.y)),
# cv.cvPoint(int(i.x+i.width), int(i.y+i.height)),
# cv.CV_RGB(0,255,0), 3, 8, 0)
center = cv.cvPoint(int(i.x+i.width/2), int(i.y+i.height/2))
cv.cvCircle(image, center, 10, cv.CV_RGB(0,0,0), 5,8, 0)
# Left side check
if center.x > box_forward_left[0].x and center.x < box_backwards_left[1].x and center.y > box_forward_left[0].y and center.y < box_backwards_left[1].y:
set_speed('left', center)
# Right side check
if center.x > box_forward_right[0].x and center.x < box_backwards_right[1].x and center.y > box_forward_right[0].y and center.y < box_backwards_right[1].y:
set_speed('right', center)
def on_trackbar(position):
# create the image for putting in it the founded contours
contours_image = cv.cvCreateImage(cv.cvSize(_SIZE, _SIZE), 8, 3)
# compute the real level of display, given the current position
levels = position - 3
# initialisation
_contours = contours
if levels <= 0:
# zero or negative value
# => get to the nearest face to make it look more funny
_contours = contours.h_next.h_next.h_next
# first, clear the image where we will draw contours
cv.cvSetZero(contours_image)
# draw contours in red and green
cv.cvDrawContours(contours_image, _contours, _red, _green, levels, 3,
cv.CV_AA, cv.cvPoint(0, 0))
# finally, show the image
highgui.cvShowImage("contours", contours_image)
def PIL2Ipl(input):
"""Converts a PIL image to the OpenCV/IPL CvMat data format.
Supported input image formats are:
RGB
L
F
"""
if not (isinstance(input, PIL.Image.Image) or isinstance(input, Image.Image)):
raise TypeError, 'Must be called with PIL.Image.Image or Image.Image!'
# mode dictionary:
# (pil_mode : (ipl_depth, ipl_channels)
mode_list = {
"RGB" : (cv.IPL_DEPTH_8U, 3),
"L" : (cv.IPL_DEPTH_8U, 1),
"F" : (cv.IPL_DEPTH_32F, 1)
}
if not mode_list.has_key(input.mode):
raise ValueError, 'unknown or unsupported input mode'
result = cv.cvCreateImage(
cv.cvSize(input.size[0], input.size[1]), # size
mode_list[input.mode][0], # depth
mode_list[input.mode][1] # channels
)
# set imageData
result.imageData = input.tostring()
return result
def __init__(self,
name,
size=2,
draw_center=True,
draw_grid=True,
meters_radius=4.0):
"""
name = name of window
meter_radus = 4.0
size = multiple of 400x200 to use for screen
meter_radius = how many per metrer
"""
self.draw_center = draw_center
self.draw_grid = draw_grid
self.w = (int)(round(size * 400.0))
self.h = (int)(round(size * 200.0))
self.meters_disp = 4.0 #Range in meters of area around robot to display
self.laser_win = name
self.buffer = cv.cvCreateImage(cv.cvSize(self.w, 2 * self.h),
cv.IPL_DEPTH_8U, 3)
#print "RobotDisp: window width", self.buffer.width
#print "RobotDisp: window height", self.buffer.height
self.pixels_per_meter = self.h / self.meters_disp
hg.cvNamedWindow(name, hg.CV_WINDOW_AUTOSIZE)
hg.cvMoveWindow(name, 0, 50)
self.font = cv.cvInitFont(cv.CV_FONT_HERSHEY_PLAIN, as_int(1),
as_int(1), 0, 1, cv.CV_AA)
def detectObject(image):
grayscale = cv.cvCreateImage(size, 8, 1)
cv.cvFlip(image, None, 1)
cv.cvCvtColor(image, grayscale, cv.CV_BGR2GRAY)
storage = cv.cvCreateMemStorage(0)
cv.cvClearMemStorage(storage)
cv.cvEqualizeHist(grayscale, grayscale)
cascade = cv.cvLoadHaarClassifierCascade(haar_file, cv.cvSize(1, 1))
objects = cv.cvHaarDetectObjects(grayscale, cascade, storage, 1.2, 2,
cv.CV_HAAR_DO_CANNY_PRUNING,
cv.cvSize(100, 100))
# Draw dots where hands are
if objects:
for i in objects:
#cv.cvRectangle(image, cv.cvPoint( int(i.x), int(i.y)),
# cv.cvPoint(int(i.x+i.width), int(i.y+i.height)),
# cv.CV_RGB(0,255,0), 3, 8, 0)
center = cv.cvPoint(int(i.x + i.width / 2),
int(i.y + i.height / 2))
cv.cvCircle(image, center, 10, cv.CV_RGB(0, 0, 0), 5, 8, 0)
# Left side check
if center.x > box_forward_left[
0].x and center.x < box_backwards_left[
1].x and center.y > box_forward_left[
0].y and center.y < box_backwards_left[1].y:
set_speed('left', center)
# Right side check
if center.x > box_forward_right[
0].x and center.x < box_backwards_right[
1].x and center.y > box_forward_right[
0].y and center.y < box_backwards_right[1].y:
set_speed('right', center)
def read(self): frame=self.input.read() if self.enabled: cv_rs = [None]*4 cv_thresh = [0]*4 cv_max = [255]*4 for i in self.channels : cv_rs[i] = cv.cvCreateImage(cv.cvSize(frame.width,frame.height),frame.depth,1) cv_thresh[i] = self.thresholds[i] cv_max[i] = self.max_thresholds[i] # extract the color channel cv.cvSplit(frame,cv_rs[0],cv_rs[1],cv_rs[2],cv_rs[3]) #self.debug_print(cv_rs) for i in self.channels : cv.cvThreshold(cv_rs[i],cv_rs[i],cv_thresh[i],cv_max[i],self.type) #cv_thresh = cv.cvCreateImage(cv.cvSize(frame.width,frame.height),frame.depth,3) cv.cvZero(frame) cv.cvMerge(cv_rs[0],cv_rs[1],cv_rs[2],cv_rs[3],frame) #frame = cv_thresh return frame
def detect_faces(self, img_grey):
""" Detect faces within an image, then draw around them.
The default parameters (scale_factor=1.1, min_neighbors=3, flags=0) are tuned
for accurate yet slow object detection. For a faster operation on real video
images the settings are:
scale_factor=1.2, min_neighbors=2, flags=CV_HAAR_DO_CANNY_PRUNING,
min_size=<minimum possible face size
"""
min_size = cv.cvSize(20, 20)
self.image_scale = 1.3
haar_scale = 1.2
min_neighbors = 2
haar_flags = 0
# Create a small image for better performance
small_size = cv.cvSize(cv.cvRound(img_grey.width / self.image_scale),
cv.cvRound(img_grey.height / self.image_scale))
small_img = cv.cvCreateImage(small_size, 8, 1)
cv.cvResize(img_grey, small_img, cv.CV_INTER_LINEAR)
cv.cvEqualizeHist(small_img, small_img)
cv.cvClearMemStorage(self.faces_storage)
if (self.cascade):
t = cv.cvGetTickCount()
faces = cv.cvHaarDetectObjects(small_img, self.cascade,
self.faces_storage, haar_scale,
min_neighbors, haar_flags, min_size)
t = cv.cvGetTickCount() - t
cv.cvReleaseImage(small_img)
#print "detection time = %gms" % (t/(cvGetTickFrequency()*1000.));
return faces
def read(self) :
frame=self.input.read()
cv_rs = cv.cvCreateImage(cv.cvSize(frame.width,frame.height),frame.depth,1)
cv.cvCvtColor(frame,cv_rs,cv.CV_RGB2GRAY)
frame = cv_rs
if self.enabled :
# I think these functions are too specialized for transforms
cv.cvSmooth(frame,frame,cv.CV_GAUSSIAN,3, 0, 0, 0 )
cv.cvErode(frame, frame, None, 1)
cv.cvDilate(frame, frame, None, 1)
num_contours,contours=cv.cvFindContours(frame,self.storage,cv.sizeof_CvContour,cv.CV_RETR_LIST,cv.CV_CHAIN_APPROX_NONE,cv.cvPoint(0,0))
if contours is None :
return []
else :
contours = cv.cvApproxPoly( contours, cv.sizeof_CvContour, self.storage, cv.CV_POLY_APPROX_DP, 3, 1 );
if contours is None :
return []
else :
final_contours = []
for c in contours.hrange() :
area = abs(cv.cvContourArea(c))
#self.debug_print('Polygon Area: %f'%area)
if area >= self.min_area :
lst = []
for pt in c :
lst.append((pt.x,pt.y))
final_contours.append(lst)
contours = contours.h_next
return final_contours
return []
def detect_faces(self, img_grey): """ Detect faces within an image, then draw around them. The default parameters (scale_factor=1.1, min_neighbors=3, flags=0) are tuned for accurate yet slow object detection. For a faster operation on real video images the settings are: scale_factor=1.2, min_neighbors=2, flags=CV_HAAR_DO_CANNY_PRUNING, min_size=<minimum possible face size """ min_size = cv.cvSize(20,20) self.image_scale = 1.3 haar_scale = 1.2 min_neighbors = 2 haar_flags = 0 # Create a small image for better performance small_size = cv.cvSize(cv.cvRound(img_grey.width/self.image_scale),cv.cvRound(img_grey.height/self.image_scale)) small_img = cv.cvCreateImage(small_size, 8, 1) cv.cvResize(img_grey, small_img, cv.CV_INTER_LINEAR) cv.cvEqualizeHist(small_img, small_img) cv.cvClearMemStorage(self.faces_storage) if(self.cascade): t = cv.cvGetTickCount(); faces = cv.cvHaarDetectObjects(small_img, self.cascade, self.faces_storage, haar_scale, min_neighbors, haar_flags, min_size) t = cv.cvGetTickCount() - t cv.cvReleaseImage(small_img) #print "detection time = %gms" % (t/(cvGetTickFrequency()*1000.)); return faces
def read(self):
frame = self.input.read()
# which channels to combine
cv_rs = [None]*4
#self.debug_print('channels:%s'%self.channels)
# if frame only has one channel, just return it
if frame.nChannels == 1 :
for i in self.channels :
cv_rs[i] = frame
else :
for i in self.channels :
cv_rs[i] = cv.cvCreateImage(cv.cvSize(frame.width,frame.height),frame.depth,1)
#self.debug_print(cv_rs)
# extract the color channel
#print 'frame.nChannels',frame.nChannels
cv.cvSplit(frame,cv_rs[0],cv_rs[1],cv_rs[2],cv_rs[3])
#cvt_im = cv.cvCreateImage(cv.cvSize(frame.width,frame.height),frame.depth,3)
cv.cvMerge(cv_rs[0],cv_rs[1],cv_rs[2],cv_rs[3],frame)
return frame
def process(self, videofile, progress):
progress(0, _("Extracting histogram"))
video = hg.cvCreateFileCapture(str(videofile).encode(sys.getfilesystemencoding()))
if not video:
raise Exception("Could not open video file")
histo = cv.cvCreateHist([256],cv.CV_HIST_ARRAY,[[0,256]], 1)
frame = hg.cvQueryFrame(video)
frame_gray = cv.cvCreateImage(cv.cvGetSize(frame), frame.depth, 1);
hists = []
nbframes = 0
fps = hg.cvGetCaptureProperty(video, hg.CV_CAP_PROP_FPS)
while frame :
if not progress(hg.cvGetCaptureProperty(video, hg.CV_CAP_PROP_POS_AVI_RATIO)):
break
hg.cvConvertImage(frame,frame_gray)
cv.cvCalcHist(frame_gray,histo,0,None)
h = [cv.cvGetReal1D(histo.bins,i) for i in range(255) ]
h = numpy.array(h,dtype='int32')
hists.append(h)
frame = hg.cvQueryFrame(video)
nbframes += 1
hists = numpy.array(hists)
return hists.reshape(nbframes, -1), fps
def show(fr,width,height,name):
image = cv.cvCreateImage(cv.cvSize (width, height),8,1)
l = 0
for j in range(0,image.width):
for i in range(0,image.height):
cv.cvSet2D(image,i,j,int(fr[l][0]));
l=l+1
highgui.cvShowImage(name,image)
highgui.cvWaitKey(1000/29)
def HarrisPoints(self, imgfile):
self.points = []
self.drawimg = highgui.cvLoadImage(imgfile)
c = 1
try:
gray = cv.cvCreateImage(cv.cvGetSize(self.drawimg), 8, 1)
cv.cvCvtColor(self.drawimg, gray, cv.CV_BGR2GRAY)
eig = cv.cvCreateImage(cv.cvGetSize(self.drawimg), 32, 1)
tmpimg = cv.cvCreateImage(cv.cvGetSize(self.drawimg), 32, 1)
p = cv.cvGoodFeaturesToTrack(gray, eig, tmpimg, 100, 0.1, 20, None,
7, 1, 0.04)
for x in p:
cv.cvCircle(self.drawimg, x, 3, cv.CV_RGB(0, 255, 0), 8, 0)
self.points.append(x)
except Exception, e:
print e
print 'ERROR: problem handling ' + imgfile
def putoriginal(fname, img):
ori_img = highgui.cvLoadImage(fname)
ori_img_thumb = cv.cvCreateImage(
cv.cvSize(ori_img.width / 4, ori_img.height / 4), 8, 3)
cv.cvResize(ori_img, ori_img_thumb)
for x in range(ori_img_thumb.height):
for y in range(ori_img_thumb.width):
cv.cvSet2D(img, x, y, cv.cvGet2D(ori_img_thumb, x, y))
return
def __findedge(self, filename):
tmpimg = highgui.cvLoadImage(filename)
self.img = cv.cvCreateImage(
cv.cvSize(int(tmpimg.width * self.enlarge),
int(tmpimg.height * self.enlarge)), 8, 3)
cv.cvResize(tmpimg, self.img, cv.CV_INTER_LINEAR)
if (self.drawimage):
self.drawimg = cv.cvCloneImage(self.img)
else:
self.drawimg = cv.cvCreateImage(cv.cvGetSize(self.img), 8, 3)
greyimg = cv.cvCreateImage(cv.cvSize(self.img.width, self.img.height),
8, 1)
cv.cvCvtColor(self.img, greyimg, cv.CV_BGR2GRAY)
self.allcurve = []
for i in range(80, 200, 20):
bimg = cv.cvCloneImage(greyimg)
cv.cvSmooth(bimg, bimg, cv.CV_MEDIAN, 9)
# cv.cvSmooth(bimg, bimg, cv.CV_BILATERAL, 9)
# cv.cvSmooth(bimg, bimg, cv.CV_BLUR, 9)
# cv.cvSmooth(bimg, bimg, cv.CV_BLUR, 9)
cv.cvThreshold(greyimg, bimg, i, 255, cv.CV_THRESH_BINARY)
self.__findcurve(bimg)
def mask_image(im, mask):
if mask.depth == 8:
bim = cv.cvCreateImage(cv.cvSize(mask.width, mask.height),
cv.IPL_DEPTH_32F, mask.nChannels)
cv.cvConvertScale(mask, bim, 1.0 / 255.0)
if im.depth == 8:
newim = cv.cvCreateImage(cv.cvSize(im.width, im.height),
cv.IPL_DEPTH_32F, im.nChannels)
cv.cvConvertScale(im, newim, 1.0 / 255.0)
print 'newim.depth = ', newim.depth
print 'newim.nChannels = ', newim.nChannels
print 'bim.depth = ', bim.depth
print 'bim.nChannels = ', bim.nChannels
if newim.nChannels == 3 and newim.depth == 32 and bim.nChannels == 3 and bim.depth == 32:
outputIm = cv.cvCloneImage(bim)
cv.cvMul(bim, newim, outputIm, 1)
return outputIm
else:
print 'oops problem with formats'
return mask
def detect_face(self, img):
""" Detect faces within an image, then draw around them.
The default parameters (scale_factor=1.1, min_neighbors=3, flags=0) are tuned
for accurate yet slow object detection. For a faster operation on real video
images the settings are:
scale_factor=1.2, min_neighbors=2, flags=CV_HAAR_DO_CANNY_PRUNING,
min_size=<minimum possible face size
"""
min_size = cv.cvSize(20, 20)
image_scale = 1.3
haar_scale = 1.2
min_neighbors = 2
haar_flags = 0
gray = cv.cvCreateImage(cv.cvSize(img.width, img.height), 8, 1)
small_img = cv.cvCreateImage(
cv.cvSize(cv.cvRound(img.width / image_scale),
cv.cvRound(img.height / image_scale)), 8, 1)
cv.cvCvtColor(img, gray, cv.CV_BGR2GRAY)
cv.cvResize(gray, small_img, cv.CV_INTER_LINEAR)
cv.cvEqualizeHist(small_img, small_img)
cv.cvClearMemStorage(self.storage)
if (self.cascade):
t = cv.cvGetTickCount()
faces = cv.cvHaarDetectObjects(small_img, self.cascade,
self.storage, haar_scale,
min_neighbors, haar_flags, min_size)
t = cv.cvGetTickCount() - t
#print "detection time = %gms" % (t/(cvGetTickFrequency()*1000.));
if faces:
for r in faces:
pt1 = cv.cvPoint(int(r.x * image_scale),
int(r.y * image_scale))
pt2 = cv.cvPoint(int((r.x + r.width) * image_scale),
int((r.y + r.height) * image_scale))
cv.cvRectangle(img, pt1, pt2, cv.CV_RGB(255, 0, 0), 3, 8,
0)
return img
def run(self):
if self.capture:
webcam_frame = highgui.cvQueryFrame(self.capture)
else:
print "Capture failed!"
return
if self.inverted_video.get_active():
highgui.cvConvertImage(webcam_frame, webcam_frame,
highgui.CV_CVTIMG_FLIP)
highgui.cvConvertImage(webcam_frame, self.display_frame,
highgui.CV_CVTIMG_SWAP_RB)
if False:
# PROCESS WEBCAM FRAME HERE...
inputImage = cv.cvCreateImage(cv.cvGetSize(webcam_frame),
cv.IPL_DEPTH_8U, 1)
cv.cvCvtColor(webcam_frame, inputImage, cv.CV_RGB2GRAY)
cv.cvThreshold(inputImage, inputImage, 128, 255,
cv.CV_THRESH_BINARY)
mysize = cv.cvGetSize(webcam_frame)
height = mysize.height
width = mysize.width
# Find horizontal first-moment:
if False:
mysum = 0
for i in range(height):
mysum += sum(inputImage[i, :])
print "Sum:", mysum
cv.cvMerge(inputImage, inputImage, inputImage, None,
self.display_frame)
incoming_pixbuf = gtk.gdk.pixbuf_new_from_data(
self.display_frame.imageData, gtk.gdk.COLORSPACE_RGB, False, 8,
self.display_frame.width, self.display_frame.height,
self.display_frame.widthStep)
incoming_pixbuf.copy_area(0, 0, self.display_frame.width,
self.display_frame.height,
self.webcam_pixbuf, 0, 0)
self.video_image.queue_draw()
return self.video_enabled_button.get_active()
def detect_faces_on(path):
faces = []
image = cvLoadImage(path)
# convert to grayscale for faster results
grayscale = cvCreateImage(cvSize(image.width, image.height), 8, 1)
cvCvtColor(image, grayscale, CV_BGR2GRAY)
# smooth picture for better results
cvSmooth(grayscale, grayscale, CV_GAUSSIAN, 3, 3)
storage = cvCreateMemStorage(0)
cvClearMemStorage(storage)
cvEqualizeHist(grayscale, grayscale)
cascade_files = [
# ('/usr/share/opencv/haarcascades/haarcascade_eye_tree_eyeglasses.xml', (50, 50)),
# ('/usr/share/opencv/haarcascades/haarcascade_frontalface_alt.xml', (50, 50)),
# ('/usr/share/opencv/haarcascades/haarcascade_lowerbody.xml', (50, 50)),
# ('/usr/share/opencv/haarcascades/haarcascade_mcs_mouth.xml', (50, 50)),
# ('/usr/share/opencv/haarcascades/haarcascade_profileface.xml', (50, 50)),
# ('/usr/share/opencv/haarcascades/haarcascade_eye.xml', (50, 50)),
# ('/usr/share/opencv/haarcascades/haarcascade_frontalface_default.xml', (50, 50)),
# ('/usr/share/opencv/haarcascades/haarcascade_mcs_eyepair_big.xml', (50, 50)),
# ('/usr/share/opencv/haarcascades/haarcascade_mcs_nose.xml', (50, 50)),
# ('/usr/share/opencv/haarcascades/haarcascade_righteye_2splits.xml', (50, 50)),
# ('/usr/share/opencv/haarcascades/haarcascade_frontalface_alt2.xml', (50, 50)),
# ('/usr/share/opencv/haarcascades/haarcascade_fullbody.xml', (50, 50)),
# ('/usr/share/opencv/haarcascades/haarcascade_mcs_eyepair_small.xml', (50, 50)),
# ('/usr/share/opencv/haarcascades/haarcascade_mcs_righteye.xml', (50, 50)),
# ('/usr/share/opencv/haarcascades/haarcascade_upperbody.xml', (50, 50)),
('/usr/share/opencv/haarcascades/haarcascade_frontalface_alt_tree.xml',
(50, 50)),
# ('/usr/share/opencv/haarcascades/haarcascade_lefteye_2splits.xml', (50, 50)),
# ('/usr/share/opencv/haarcascades/haarcascade_mcs_lefteye.xml', (50, 50)),
# ('/usr/share/opencv/haarcascades/haarcascade_mcs_upperbody.xml', (50, 50)),
# ('parojos_22_5.1.xml', (22, 5)),
# ('Mouth.xml', (22, 15)),
]
for cascade_file, cascade_sizes in cascade_files:
cascade = cvLoadHaarClassifierCascade(os.path.join(cascade_file),
cvSize(1, 1))
faces += cvHaarDetectObjects(grayscale, cascade, storage, HAAR_SCALE,
HAAR_NEIGHBORS, CV_HAAR_DO_CANNY_PRUNING,
cvSize(*cascade_sizes))
return [{'x': f.x, 'y': f.y, 'w': f.width, 'h': f.height} for f in faces]
def detect_faces(image):
"""Converts an image to grayscale and prints the locations of any
faces found"""
grayscale = cvCreateImage(cvSize(image.width, image.height), 8, 1)
cvCvtColor(image, grayscale, CV_BGR2GRAY)
storage = cvCreateMemStorage(0)
cvClearMemStorage(storage)
cvEqualizeHist(grayscale, grayscale)
# The default parameters (scale_factor=1.1, min_neighbors=3,
# flags=0) are tuned for accurate yet slow face detection. For
# faster face detection on real video images the better settings are
# (scale_factor=1.2, min_neighbors=2, flags=CV_HAAR_DO_CANNY_PRUNING).
# --- http://www710.univ-lyon1.fr/~bouakaz/OpenCV-0.9.5/docs/ref/OpenCVRef_Experimental.htm#decl_cvHaarDetectObjects
# The size box is of the *minimum* detectable object size. Smaller box = more processing time. - http://cell.fixstars.com/opencv/index.php/Facedetect
minsize = (int(MINFACEWIDTH_PERCENT * image.width + 0.5),
int(MINFACEHEIGHT_PERCENT * image.height))
print >> sys.stderr, "Min size of face: %s" % ` minsize `
faces = []
for cascadefile in [
'/usr/share/opencv/haarcascades/haarcascade_frontalface_default.xml'
]:
# for cascadefile in ['/usr/share/opencv/haarcascades/haarcascade_frontalface_default.xml', '/usr/share/opencv/haarcascades/haarcascade_profileface.xml']:
cascade = cvLoadHaarClassifierCascade(cascadefile, cvSize(1, 1))
# faces += cvHaarDetectObjects(grayscale, cascade, storage, 1.2, 2, CV_HAAR_DO_CANNY_PRUNING, cvSize(50,50))
# faces += cvHaarDetectObjects(grayscale, cascade, storage, 1.1, 3, 0, cvSize(MINFACEWIDTH,MINFACEHEIGHT))
# faces += cvHaarDetectObjects(grayscale, cascade, storage, 1.1, 3, 0, cvSize(MINFACEWIDTH,MINFACEHEIGHT))
# faces += cvHaarDetectObjects(grayscale, cascade, storage, 1.1, 3, CV_HAAR_DO_CANNY_PRUNING, cvSize(*minsize))
faces += cvHaarDetectObjects(grayscale, cascade, storage, 1.1,
4, CV_HAAR_DO_CANNY_PRUNING,
cvSize(*minsize))
# faces += cvHaarDetectObjects(grayscale, cascade, storage, scale_factor=1.1, min_neighbors=3, flags=0, cvSize(50,50))
# print dir(faces)
bboxes = []
if faces:
for f in faces:
print >> sys.stderr, "\tFace at [(%d,%d) -> (%d,%d)]" % (
f.x, f.y, f.x + f.width, f.y + f.height)
bboxes = [Face(f.x, f.y, f.x + f.width, f.y + f.height) for f in faces]
return bboxes
def detect(image):
image_size = opencv.cvGetSize(image)
# create grayscale version
grayscale = opencv.cvCreateImage(image_size, 8, 1)
opencv.cvCvtColor(image, grayscale, opencv.CV_BGR2GRAY)
# create storage
storage = opencv.cvCreateMemStorage(0)
opencv.cvClearMemStorage(storage)
# equalize histogram
opencv.cvEqualizeHist(grayscale, grayscale)
# detect objects
faces = opencv.cvHaarDetectObjects(grayscale, cascade, storage, 1.2, 2, opencv.CV_HAAR_DO_CANNY_PRUNING, opencv.cvSize(100, 100))
# eyes = opencv.cvHaarDetectObjects(grayscale, eye_cascade, storage, 1.2, 2, opencv.CV_HAAR_DO_CANNY_PRUNING, opencv.cvSize(60,60))
draw_bounding_boxes(faces, image, 127,255,0, 3)
def initialize_video(self):
webcam_frame = highgui.cvQueryFrame(self.capture)
if not webcam_frame:
print "Frame acquisition failed."
return False
self.webcam_pixbuf = gtk.gdk.pixbuf_new_from_data(
webcam_frame.imageData, gtk.gdk.COLORSPACE_RGB, False, 8,
webcam_frame.width, webcam_frame.height, webcam_frame.widthStep)
self.video_image.set_from_pixbuf(self.webcam_pixbuf)
self.display_frame = cv.cvCreateImage(
cv.cvSize(webcam_frame.width, webcam_frame.height),
cv.IPL_DEPTH_8U, 3)
return True
def np2cv(im):
print 'WARNING: np2cv is not reliable or well tested (it is a bit flakey...)'
#raise AssertionError('np2cv does not work :-(')
if len(im.shape) == 3:
shp = im.shape
channels = shp[2]
height = shp[0]
width = shp[1]
#height, width, channels = im.shape
elif len(im.shape) == 2:
height, width = im.shape
channels = 1
else:
raise AssertionError(
"unrecognized shape for the input image. should be 3 or 2, but was %d."
% len(im.shape))
key = str(im.dtype)
cv_type = np2cv_type_dict[key]
print 'attempt to create opencv image with (key, width, height, channels) =', (
key, width, height, channels)
cv_im = cv.cvCreateImage(cv.cvSize(width, height), cv_type, channels)
#cv_im.imageData = im.tostring()
if True:
if len(im.shape) == 3:
for y in xrange(height):
for x in xrange(width):
pix = [float(v) for v in im[y, x]]
scalar = cv.cvScalar(*pix)
#print scalar
cv_im[y, x] = scalar
else:
for y in xrange(height):
for x in xrange(width):
pix = float(im[y, x])
cv_im[y, x] = cv.cvScalar(pix, pix, pix)
#print 'im[y,x], cv_im[y,x] =', im[y,x], cv_im[y,x]
print 'resulted in an image openCV image with the following properties:'
numpy_type, nchannels = cv2np_type_dict[cv.cvGetElemType(cv_im)]
print '(numpy_type, nchannels, cvmat.width, cvmat.height) =', (
numpy_type, nchannels, cv_im.width, cv_im.height)
return cv_im
def threshold_image(image, n=[]):
"""Record the first 5 images to get a background, then diff current frame with the last saved frame.
"""
if len(n) < 5:
# n[4] will be our background
# First capture a few images
n.append(cv.cvCloneMat(image))
if len(n) == 5:
# last time here
# could do averaging here.
pass
return image
original = n[4]
differenceImage = cv.cvCloneMat(image)
cv.cvAbsDiff(image, original, differenceImage)
"""The threshold value determines the amount of "Change" required
before something will show up"""
thresholdValue = 50 # 32
cv.cvThreshold(differenceImage, differenceImage, thresholdValue, 255,
cv.CV_THRESH_BINARY)
# Convert to one channel
gray = cv.cvCreateImage(cv.cvGetSize(differenceImage), 8, 1)
cv.cvCvtColor(differenceImage, gray, cv.CV_BGR2GRAY)
# Use median filter to remove salt and pepper noise.
cv.cvSmooth(gray, gray, cv.CV_MEDIAN, 15)
# Dilate and the threshold image
# It adds a border to the object.
#cv.cvDilate(gray,gray, None, 9)
# Add a bit of Blur to the threshold mask
cv.cvSmooth(gray, gray, cv.CV_GAUSSIAN, 5)
result = cv.cvCloneMat(image)
cv.cvSetZero(result)
cv.cvAnd(image, image, result, gray)
return result
def timerEvent(self, ev):
# Fetch a frame from the video camera
frame = highgui.cvQueryFrame(self.cap)
img_orig = cv.cvCreateImage(cv.cvSize(frame.width, frame.height),
cv.IPL_DEPTH_8U, frame.nChannels)
if (frame.origin == cv.IPL_ORIGIN_TL):
cv.cvCopy(frame, img_orig)
else:
cv.cvFlip(frame, img_orig, 0)
# Create a grey frame to clarify data
#img = self.detect_face(frame_copy)
img = self.detect_squares(frame_copy)
img_pil = adaptors.Ipl2PIL(img)
s = StringIO()
img_pil.save(s, "PNG")
s.seek(0)
q_img = QImage()
q_img.loadFromData(s.read())
bitBlt(self, 0, 0, q_img)
def display_images(image_list, max_x=1200, max_y=1000, save_images=False):
"""
Display a list of OpenCV images tiled across the screen
with maximum width of max_x and maximum height of max_y
save_images - will save the images(with timestamp)
"""
curtime = time.localtime()
date_name = time.strftime('%Y_%m_%d_%I%M%S', curtime)
loc_x, loc_y = 0, 0
wins = []
for i, im in enumerate(image_list):
if save_images:
if im.nChannels == 1 and im.depth == cv.IPL_DEPTH_32F:
clr = cv.cvCreateImage(cv.cvSize(im.width, im.height),
cv.IPL_DEPTH_8U, 1)
cv.cvConvertScale(im, clr, 255.0)
im = clr
highgui.cvSaveImage('image%d_' % i + date_name + '.png', im)
window_name = 'image %d' % i
wins.append((window_name, im))
highgui.cvNamedWindow(window_name, highgui.CV_WINDOW_AUTOSIZE)
highgui.cvMoveWindow(window_name, loc_x, loc_y)
loc_x = loc_x + im.width
if loc_x > max_x:
loc_x = 0
loc_y = loc_y + im.height
if loc_y > max_y:
loc_y = 0
while True:
for name, im in wins:
highgui.cvShowImage(name, im)
keypress = highgui.cvWaitKey(10)
if keypress == '\x1b':
break
def detect(self, pil_image, cascade_name, recogn_w=50, recogn_h=50):
# Get cascade:
cascade = self.get_cascade(cascade_name)
image = opencv.PIL2Ipl(pil_image)
image_size = opencv.cvGetSize(image)
grayscale = image
if pil_image.mode == "RGB":
# create grayscale version
grayscale = opencv.cvCreateImage(image_size, 8, 1)
# Change to RGB2Gray - I dont think itll affect the conversion
opencv.cvCvtColor(image, grayscale, opencv.CV_BGR2GRAY)
# create storage
storage = opencv.cvCreateMemStorage(0)
opencv.cvClearMemStorage(storage)
# equalize histogram
opencv.cvEqualizeHist(grayscale, grayscale)
# detect objects
return opencv.cvHaarDetectObjects(grayscale, cascade, storage, 1.2, 2,
opencv.CV_HAAR_DO_CANNY_PRUNING,
opencv.cvSize(recogn_w, recogn_h))
