Python + OpenCV：枚举矩阵

Question

I am a newbie to Python.

And I want to access Mat (a threshold binary image) element from opencv and create the a histogram according to the x-axis, and so I can do some image segmentation vertically.

What I have done is to preprocess the image and enumerate twice as following.

def crack(src):
    #src is a binary image 

    src = cv2.resize(src, (0,0), fx=6, fy=6)
    thresh = preprocessing(src)

    #create empty histogram
    print(thresh.shape)
    height, width = thresh.shape
    size = height ,255,  3
    hist = np.zeros(size, dtype=np.uint8)

    #enumerate elements
    for y, x in enumerate(thresh):
        val = 0
        for x, pix in enumerate(x):
            val = val+ pix/255
            print y,x, val
        cv2.rectangle(hist, (y, 255-val), (y+val, 255-val+1), (0, 255, 0), 1)

    cv2.imshow("thresh", thresh)
    cv2.imshow("hist", hist)

And if I directly enumerate the threshold Mat like

    for y, x in enumerate(thresh):

I can only enumerate from the outer y axis first then enumerate the x axis. So how can I do this reversely? My aim is to get the image like this:

image1 http://7xsnr6.com2.z0.glb.clouddn.com/QQ%E5%9B%BE%E7%89%8720160509184009.jpg

Image references:

Jeff Yan, A Low-cost Attack on a Microsoft CAPTCHA

Answer 1

If you want the number of black pixels for each x position, following your code this should work:

# Enumerate elements
for x in range(width):
    val = 0
    for y in range(height):
        val += (255-thresh[y, x])/255
        print y, x, val
    cv2.line(thresh, (x, height-val-1), (x, height-1), (0, 255, 0), 1)

So basically you are looping over the x axis and, for each of its position, finding the amount of black pixels over the y axis, which is stored in the variable val . Finally you draw a line that has a height equal to val pixels at the x position you are looping that starts at the bottom of the image.

Hope this helped!