import cv2 as cv
import numpy as np


#均值滤波
def meansBlur(src, ksize):
 '''
 :param src: input image
 :param ksize:kernel size
 :return dst: output image
 '''
 dst = np.copy(src) #创建输出图像
 kernel = np.ones((ksize, ksize)) # 卷积核
 padding_num = int((ksize - 1) / 2) #需要补0
 dst = np.pad(dst, (padding_num, padding_num), mode="constant", constant_values=0)
 w, h = dst.shape
 dst = np.copy(dst)
 for i in range(padding_num, w - padding_num):
  for j in range(padding_num, h - padding_num):
   dst[i, j] = np.sum(kernel * dst[i - padding_num:i + padding_num + 1, j - padding_num:j + padding_num + 1])       // (ksize ** 2)
 dst = dst[padding_num:w - padding_num, padding_num:h - padding_num] #把操作完多余的0去除，保证尺寸一样大
 return dst


img_path = r"F:\03.png"
img = cv.imread(img_path,0)
dst = meansBlur(img,5)
cv.imshow('src',img)
cv.imshow('dst',dst)
print(dst)
cv.waitKey(0)