简化从 cv2.findContours 获取坐标

Question

I'm currently working with opencv 3.4 and getting the contours as a list of coordinates from a binary image. My code looks like the following:

_, contours, _ = cv2.findContours(image, cv2.RETR_TREE, cv2.CHAIN_APPROX_NONE
list_of_contours= []
for contour in contours:         
    list_of_contours.append(list(map(tuple, (coord[0] for coord in contour))))

It is working as intended but I'm curious if the loop in list(map(tuple, (c[0] for c in contour))) could be avoided and/or is it even reasonable to do so?

Answer 1

Try this:

cnts = cv2.findContours(gray,cv2.RETR_EXTERNAL,cv2.CHAIN_APPROX_SIMPLE)[-2]
xcnts = np.vstack((x.reshape(-1,2) for x in cnts))
print(xcnts.shape)

It return (698, 2) for this image (from opencv - plot contours in an image ):