函数陷入无限循环，找不到原因

Question

The code is supposed to take each pixel rgb value and store it in a 2d array (eg pixels[pixelNumber][r, g, b]), where it then gets sent to a user defined method to, in this case, blur it (using Gaussian blur). Width and height are pixel width and height definitions taken from the image, and radius is the user defined radius of which pixels get averaged to be blurred. It gave me a divide by 0 error at

weight_r = pixels[j][0] * weight
weight_g = pixels[j][1] * weight
weight_b = pixels[j][2] * weight

before I changed total_weight from 0 to 1 when defined just below the first for loop, if that helps

def blur(pixels, radius, width, height):
    for i in range(len(pixels)):
        total_weight = 1
        pix_weight = 0
        r_temp = radius
        pixels_2 = copy.deepcopy(pixels)
        x = i % width
        y = (i // width) + 1
        for j in range(len(pixels)):
            x_2 = j % width
            y_2 = (j // width) + 1
            dist = math.sqrt(((x_2 - x) ** 2) + ((y_2 - y) ** 2))
            while(r_temp > 0):
                if((x_2 - radius == x or x_2 + radius == x) and (y_2 - radius == y or y_2 + radius == y) and dist != 0):
                    weight = ((math.e ** (-((dist ** 2) / (2 * (radius ** 2))))) / (2 * math.pi * (radius ** 2)))
                    total_weight = total_weight + weight
                    weight_r = pixels[j][0] * weight
                    weight_g = pixels[j][1] * weight
                    weight_b = pixels[j][2] * weight
                    pix_weight = pix_weight + weight_r + weight_g + weight_b
                r_temp = r_temp - 1
        final_blur = int((pix_weight / total_weight) / 255)
        pixels_2[i][0] = int(pixels[i][0] * final_blur)
        pixels_2[i][1] = int(pixels[i][1] * final_blur)
        pixels_2[i][2] = int(pixels[i][2] * final_blur)
    return pixels_2

Method above, pixels is a 2d array, radius, width, and height are all integers.

Answer 1

I've copied your code exactly, using Python 2.7.11 with some example arguments below, and I cannot reproduce an infinite loop. For which arguments are you seeing infinite loop behavior?

In [13]: p = [[1, 1, 1, 1, 1], 
              [1, 1, 1, 1, 1], 
              [1, 1, 1, 1 ,1], 
              [1, 1, 1, 1, 1], 
              [1, 1, 1, 1, 1]]

In [14]: r = 3

In [15]: w, h = 5, 5

In [16]: blur(p, r, w, h)
Out[16]: 
[[1, 1, 1, 1, 1],
 [1, 1, 1, 1, 1],
 [1, 1, 1, 1, 1],
 [1, 1, 1, 1, 1],
 [0, 0, 0, 1, 1]]

(On a separate note, it looks like your blur kernel is not treating the different image boundaries with the same edge conditions).

Added you mentioned in the comments you are using a 256x256 image -- and also just to check on the format I tried making an example with 256 ** 2 entries (list of all pixels), where each pixel is a 3-element list. I see now this is probably what you were going for.

Those changes are in this example:

In [19]: p = [[1, 1, 1] for _ in range(256 ** 2)]

In [20]: %timeit blur(p, 1, 256, 256)

This indeed is taking a very long time to run, I'll update with time info when it's finished.