python中matlab函数图像处理,python – 如何有效地处理像Matlab的blkproc(blockproc)函数的块中的numpy数组...

以下是使用块的另一种(循环自由)方式的一些示例：

import numpy as np

from numpy.lib.stride_tricks import as_strided as ast

A= np.arange(36).reshape(6, 6)

print A

#[[ 0 1 2 3 4 5]

# [ 6 7 8 9 10 11]

# ...

# [30 31 32 33 34 35]]

# 2x2 block view

B= ast(A, shape= (3, 3, 2, 2), strides= (48, 8, 24, 4))

print B[1, 1]

#[[14 15]

# [20 21]]

# for preserving original shape

B[:, :]= np.dot(B[:, :], np.array([[0, 1], [1, 0]]))

print A

#[[ 1 0 3 2 5 4]

# [ 7 6 9 8 11 10]

# ...

# [31 30 33 32 35 34]]

print B[1, 1]

#[[15 14]

# [21 20]]

# for reducing shape, processing in 3D is enough

C= B.reshape(3, 3, -1)

print C.sum(-1)

#[[ 14 22 30]

# [ 62 70 78]

# [110 118 126]]

所以只是试图简单地将matlab功能复制到numpy并不是所有的方法进行最好的方法。有时需要“脱帽”思维。

警告：

一般来说，基于步幅技巧的实现可能(但不是必须)遭受一些性能惩罚。所以做好准备，以各种方式衡量你的表现。无论如何，首先检查所需的功能(或类似的功能，以便轻松适应)已经准备好以numpy或scipy的方式实现了。

更新：

请注意，这里并没有任何真正的魔力，因此我将提供一个简单的函数来获取任何适合的2D numpy数组的block_view。所以这里我们去：

from numpy.lib.stride_tricks import as_strided as ast

def block_view(A, block= (3, 3)):

"""Provide a 2D block view to 2D array. No error checking made.

Therefore meaningful (as implemented) only for blocks strictly

compatible with the shape of A."""

# simple shape and strides computations may seem at first strange

# unless one is able to recognize the 'tuple additions' involved ;-)

shape= (A.shape[0]/ block[0], A.shape[1]/ block[1])+ block

strides= (block[0]* A.strides[0], block[1]* A.strides[1])+ A.strides

return ast(A, shape= shape, strides= strides)

if __name__ == '__main__':

from numpy import arange

A= arange(144).reshape(12, 12)

print block_view(A)[0, 0]

#[[ 0 1 2]

# [12 13 14]

# [24 25 26]]

print block_view(A, (2, 6))[0, 0]

#[[ 0 1 2 3 4 5]

# [12 13 14 15 16 17]]

print block_view(A, (3, 12))[0, 0]

#[[ 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]]