python - 如何根据模式从 2D NumPy 数组中提取 2D NumPy 子数组？

Question

我有一个二维 NumPy 数组，如下所示:

Array=
[
[0,0,0,0,0,0,0,2,2,2],
[0,0,0,0,0,0,0,2,2,2].
[0,0,1,1,1,0,0,2,2,2],
[0,0,1,1,1,0,0,2,2,2],
[0,0,1,1,1,0,0,1,1,1],
[0,0,0,0,0,0,0,1,1,1]
]

我需要将非零元素数组显示为:

Array1:
[
[1,1,1],
[1,1,1],
[1,1,1]
]

Array2:
[
[2,2,2],
[2,2,2],
[2,2,2],
[2,2,2]
]

Array3:
[
[1,1,1],
[1,1,1]
]

有人可以帮我看看我可以使用什么逻辑来实现以下输出吗？我不能使用固定索引(如 array[a:b, c:d])，因为我创建的逻辑应该能够用于具有类似模式的任何 NumPy 数组。

Answer 1

这使用 scipy.ndimage.label 递归地识别断开连接的子数组。

import numpy as np
from scipy.ndimage import label

array = np.array(
    [[0,0,0,0,0,0,0,2,2,2,3,3,3],
     [0,0,0,0,0,0,0,2,2,2,0,0,1],
     [0,0,1,1,1,0,0,2,2,2,0,2,1],
     [0,0,1,1,1,0,0,2,2,2,0,2,0],
     [0,0,1,1,1,0,0,1,1,1,0,0,0],
     [0,0,0,0,0,0,0,1,1,1,0,0,0]])

# initialize list to collect sub-arrays
arr_list = []

def append_subarrays(arr, val, val_0):
    '''
    arr : 2D array
    val : the value used for filtering
    val_0 : the original value, which we want to preserve
    '''

    # remove everything that's not the current val
    arr[arr != val] = 0
    if 0 in arr:  # <-- not a single rectangle yet
        # get relevant indices as well as their minima and maxima
        x_ind, y_ind = np.where(arr != 0)
        min_x, max_x, min_y, max_y = min(x_ind), max(x_ind) + 1, min(y_ind), max(y_ind) + 1
        # cut subarray (everything corresponding to val)
        arr = arr[min_x:max_x, min_y:max_y]
        # use the label function to assign different values to disconnected regions
        labeled_arr = label(arr)[0]
        # recursively apply append_subarrays to each disconnected region 
        for sub_val in np.unique(labeled_arr[labeled_arr != 0]):
            append_subarrays(labeled_arr.copy(), sub_val, val_0)

    else:  # <-- we only have a single rectangle left ==> append
        arr_list.append(arr * val_0)

for i in np.unique(array[array > 0]):
    append_subarrays(array.copy(), i, i)

for arr in arr_list:
    print(arr, end='\n'*2)

输出(注意:修改后的示例数组):

[[1]
 [1]]

[[1 1 1]
 [1 1 1]
 [1 1 1]]

[[1 1 1]
 [1 1 1]]

[[2 2 2]
 [2 2 2]
 [2 2 2]
 [2 2 2]]

[[2]
 [2]]

[[3 3 3]]