如何在列表中的列表的情况下忽略外部列表(不使用 Numpy)

How to ignore external list in case of lists in list(without using Numpy)

我写了一个函数,它的输入是- image_one, image_two, mask)。 该函数通过根据公式-

计算每个新像素来输出新图像
new_image[i][j] = round(image_one[i][j] x mask[i][j] + image_two[i][j] x (1 - mask[i][j]))

写到现在-

def image_plus_mask(image_one, image_two, mask):
    hight = len(image_one)
    width = len(image_one[0])
    new_image = [[0 for _ in range(width)] for _ in range(hight)]

    for i in range(hight):
        for j in range(width):
            new_image[i][j] = round((image_one[i][j] * mask[i][j]) + (image_two[i][j] * (1-mask[i][j])))
    return new_image

函数 dose 适用于双重列表。 例如:

输入-

image_plus_mask([[50, 50, 50]], [[200, 200, 200]], [[0, 0.5, 1]]) 

输出(如预期)-

[[200, 125, 50]]

问题是它不能处理 3 个列表,我需要它同时处理 - 2 和 3 个列表。(2 我的意思是 -[[] ], 3 我的意思是- [[[]]])

例如-

输入-

image_plus_mask([[[1,2,3], [4,5,6]],[[7,8,9],[10,11,12]]],
[[[250,250,250], [0,0,0]],[[250,250,100],[1,11,13]]],[[0,0.5]]*2))

预期输出-

[[[250, 250, 250,], [2,2,3]], [[250, 250, 100], [6, 11, 12]]]

我能做什么?

您的蒙版必须与输入的形状相同。 如果不同,则 image_one[i][j] * mask[i][j] 将不起作用。

我强烈建议您像这样使用 numpy 数组:

image1 = np.array([[50, 50, 50]])
image2 = np.array([[200, 200, 200]])
mask = np.array([[0, 0.5, 1]])

output = np.round(image1 * mask + image2 * (1 - mask))

print(output)

输出:

[[200. 125.  50.]]

如果您输入的形状(图像和蒙版)一致,这将提供您想要的任何尺寸。

如评论所说:

def image_plus_mask(image_one, image_two, mask):
    
    hight = len(image_one)
    width = len(image_one[0])
    
    #case: 3 lists
    if isinstance(image_one[0][0], (list,tuple)):
        depth = len(image_one[0][0])
        new_image = [[[0 for _ in range(depth)] for _ in range(width)] for _ in range(hight)]
        print(new_image)
        for i in range(hight):
            for j in range(width):
                for k in range(depth):
                    new_image[i][j][k] = round((image_one[i][j][k] * mask[i][j]) + (image_two[i][j][k] * (1-mask[i][j])))

    #case: 2 lists                
    else:
        new_image = [[0 for _ in range(width)] for _ in range(hight)]
        for i in range(hight):
            for j in range(width):
                new_image[i][j] = round((image_one[i][j] * mask[i][j]) + (image_two[i][j] * (1-mask[i][j])))
                
    return new_image

您需要调整 new_image 的形状。查看 3 个列表的预期输出并查看函数内部为 new_image 定义的内容。无法工作。

现在输出:

res1 = image_plus_mask([[[1,2,3], [4,5,6]],[[7,8,9],[10,11,12]]],
[[[250,250,250], [0,0,0]],[[250,250,100],[1,11,13]]],[[0,0.5]]*2)

res2 = image_plus_mask([[50, 50, 50]], [[200, 200, 200]], [[0, 0.5, 1]]) 

print(res1)
[[[250, 250, 250], [2, 2, 3]], [[250, 250, 100], [6, 11, 12]]]

print(res2)
[[200, 125, 50]]