操作二值图像中不同连通区域

在深度学习模型预测过程中，对生成的mask(0,1二值图)有多个连通区域的情况，有时我们只需要对其中某一个连通区域操作，思路就是对不同连通区域赋给不同的值。

from skimage.measure import label as label_function
import numpy as np
copy_label_brain = copy.deepcopy(label_brain)
z,x,y = np.shape(label_brain)
temporal_r = np.zeros((z,x,y), np.uint8)
temporal_l = np.zeros((z,x,y), np.uint8)
for i in range(z):
    copy_label_brain[i,:,:] = label_function(copy_label_brain[i,:,:])
    if np.max(copy_label_brain[i])==3:
        #保留2,3区域
        temporal_r[i,:,:] = np.where(copy_label_brain[i,:,:]>1,1,0)
        temporal_r[i,:,:] = label_brain[i,:,:] - temporal_r[i,:,:]
        #保留3区域
        temporal_l[i,:,:] = np.where(copy_label_brain[i,:,:]>2,1,0)
        temporal_l[i,:,:] = label_brain[i,:,:] - temporal_r[i,:,:] - temporal_l[i,:,:]

三个连通区域赋不同值.png