将彩色图像转换为单行numpy数组（r1，g1，b1； r2，g2，b2；…）

Question

Could anyone show me how to convert a color image into a single-row NumPy array with (r1, g1, b1; r2, g2, b2; ...) ?

So far I've loaded my image into my image with 512x512px size into (512, 512, 3) numpy array. Now I want to convert it into (1 x 786432) array.

Answer 1

ndarray.flatten() should do the trick:

In [15]: image = np.arange(4*4*3).reshape((4, 4, 3))

In [16]: image
Out[16]: 
array([[[ 0,  1,  2],
        [ 3,  4,  5],
        [ 6,  7,  8],
        [ 9, 10, 11]],

       [[12, 13, 14],
        [15, 16, 17],
        [18, 19, 20],
        [21, 22, 23]],

       [[24, 25, 26],
        [27, 28, 29],
        [30, 31, 32],
        [33, 34, 35]],

       [[36, 37, 38],
        [39, 40, 41],
        [42, 43, 44],
        [45, 46, 47]]])

In [17]: image.flatten()
Out[17]: 
array([ 0,  1,  2,  3,  4,  5,  6,  7,  8,  9, 10, 11, 12, 13, 14, 15, 16,
       17, 18, 19, 20, 21, 22, 23, 24, 25, 26, 27, 28, 29, 30, 31, 32, 33,
       34, 35, 36, 37, 38, 39, 40, 41, 42, 43, 44, 45, 46, 47])

Answer 2

To push everything into a single row, you can use reshape .

array.reshape(1, -1)

This will return a 2D view of the array with shape (1, 786432) , a single row. The method views the array as C order by default so the higher dimensions are "flattened" first (which is what you want in this case).

Alternatively, you could use ravel to get a back a flat array:

array.ravel()

This difference is that the returned view of the array is 1D: shape (786432,) .