图像阴影调整

import cv2
import numpy as np
def max_filter(image,filter_size):
    empty_image=np.full((image.shape[0]+(filter_size//2)*2,image.shape[1]+(filter_size//2)*2),-1)
    #原始图像填充进矩阵
    empty_image[(filter_size//2):empty_image.shape[0]-(filter_size//2),(filter_size//2):empty_image.shape[1]-(filter_size//2)]=image.copy()
    result = np.full((image.shape[0], image.shape[1]), -1)
    #遍历原图像中的每个像素点，对于点，选取其周围（filter_size*filter_size）个像素中的最大值，作为结果矩阵中的对应位置值
    for h in range(filter_size//2,empty_image.shape[0]-filter_size//2):
        for w in range(filter_size//2, empty_image.shape[1]-filter_size//2):
            filter=empty_image[h-(filter_size//2):h+(filter_size//2)+1,
                     w-(filter_size//2):w+(filter_size//2)+1]
            result[h-filter_size//2,w-filter_size//2]=np.amax(filter)
    return result
def min_filter(image,filter_size):
    empty_image=np.full((image.shape[0]+(filter_size//2)*2,image.shape[1]+(filter_size//2)*2),400)
    #原始图像填充进矩阵
    empty_image[(filter_size//2):empty_image.shape[0]-(filter_size//2),
    (filter_size//2):empty_image.shape[1]-(filter_size//2)]=image.copy()
    result=np.full((image.shape[0],image.shape[1]),400)
    #遍历原图像中的每个像素点，对于点，选取其周围（filter_size*filter_size）个像素中的最小值，作为结果矩阵中的对应位置值
    for h in range(filter_size//2,empty_image.shape[0]-filter_size//2):
        for w in range(filter_size//2, empty_image.shape[1]-filter_size//2):
            filter=empty_image[h-(filter_size//2):h+(filter_size//2)+1,
                     w-(filter_size//2):w+(filter_size//2)+1]
            result[h-filter_size//2, w-filter_size//2]=np.amin(filter)
    return result
def shadow_adjustment(image_path):
    image=cv2.imread(image_path,0)
    max_result=max_filter(image,30)
    min_result=min_filter(max_result,30)
    result=image-min_result
    result=cv2.normalize(result, None,0,255,norm_type=cv2.NORM_MINMAX)
    return result
if __name__=="__main__":
    result=shadow_adjustment('C:/Users/xpp/Desktop/Lena.png')
    cv2.imwrite("C:/Users/xpp/Desktop/result.png",result)