python - 如何访问numpy中的整个子数组？

Question

我试图将不同的图像结果存储在多维数组中。考虑下面的 m 有 3 个大小为 3x4 的图像，现在我想访问第二个图像，我正在尝试 m[:, :, 1]。但它产生了错误的结果。您能否纠正我如何访问子数组。

m = np.random.random((3,4,3))
array([[[ 0.72474649,  0.88239477,  0.21834498],
        [ 0.3594392 ,  0.06462519,  0.43329582],
        [ 0.83452734,  0.43937307,  0.61267164],
        [ 0.75062416,  0.76516036,  0.35487906]],

       [[ 0.90644054,  0.22091564,  0.94508058],
        [ 0.20201594,  0.60295539,  0.77998197],
        [ 0.10496835,  0.77960017,  0.78249163],
        [ 0.53903028,  0.81245971,  0.38983454]],

       [[ 0.81369716,  0.89243267,  0.49874087],
        [ 0.95248644,  0.0847973 ,  0.59074351],
        [ 0.24507041,  0.03595347,  0.32902477],
        [ 0.48906304,  0.14659161,  0.77392082]]])


m[:, :, 1]
// Actual output
array([[ 0.88239477,  0.06462519,  0.43937307,  0.76516036],
       [ 0.22091564,  0.60295539,  0.77960017,  0.81245971],
       [ 0.89243267,  0.0847973 ,  0.03595347,  0.14659161]])

// Expected  output
[ 0.90644054,  0.22091564,  0.94508058],
[ 0.20201594,  0.60295539,  0.77998197],
[ 0.10496835,  0.77960017,  0.78249163],
[ 0.53903028,  0.81245971,  0.38983454]

Answer 1

您可以访问单个图像及其预期位置，即m[0] 为您提供第一张图像。

证明:

>>> a = numpy.random.random((3, 4, 3))
>>> a
array([[[ 0.2585421 ,  0.6096753 ,  0.70295892],
        [ 0.50408344,  0.37075371,  0.30463057],
        [ 0.76298221,  0.67466292,  0.53305787],
        [ 0.63844013,  0.45100157,  0.1346955 ]],

       [[ 0.54268873,  0.31534909,  0.40414511],
        [ 0.87335605,  0.81278098,  0.12953214],
        [ 0.64353518,  0.22347   ,  0.63712407],
        [ 0.02646421,  0.56478202,  0.57160074]],

       [[ 0.36965073,  0.796066  ,  0.7289024 ],
        [ 0.47232785,  0.43087964,  0.873769  ],
        [ 0.12393581,  0.63266617,  0.0935309 ],
        [ 0.62007608,  0.77474674,  0.28507152]]])
>>> a[0]
array([[ 0.2585421 ,  0.6096753 ,  0.70295892],
       [ 0.50408344,  0.37075371,  0.30463057],
       [ 0.76298221,  0.67466292,  0.53305787],
       [ 0.63844013,  0.45100157,  0.1346955 ]])