OpenCV学习之图像加噪与滤波的实现详解

# @Time : 2022/10/10 16:29
# @Author : 是Dream呀
# @File : 图像加噪与滤波.py
import cv2 as cv
import numpy as np

img = cv.imread('photo.jpg')
cv.imshow('test',img)
cv.waitKey(0)
cv.destroyAllWindows()

#  定义sp_Noise（）来作为椒盐噪声
def sp_Noise(image, per):
    img = image.copy()
    threshold = per/2
    w,h = img.shape[0], img.shape[1]
    rand = np.random.random([w,h])
    #  标注噪声区域的二值图像，背景为黑色，噪声区域为白色
    noise = np.zeros(img.shape)
    noise[rand<per] = 1
    img[rand<threshold] = 0
    img[(rand<per) & (rand>=threshold)] = 255
    return noise*255,img

#  定义white_Noise（）来作为白噪声
def white_Noise(image, per):
    img = image.copy()
    w,h = img.shape[0], img.shape[1]
    rand = np.random.random([w,h])
    #  标注噪声区域的二值图像，背景为黑色，噪声区域为白色
    noise = np.zeros(img.shape)
    noise[rand < per] = 1
    img[rand < per] = 255
    return noise*255, img

white_noise, white_img = white_Noise(img, 0.05)
sp_noise, sp_img = sp_Noise(img, 0.02)

#  使用图像减法与二值化处理获取噪声点
sp_noise1 = np.float32(sp_img) - np.float32(img)
sp_noise1 = np.where(sp_noise1 == 0, 0, 255)
sp_noise1 = np.uint8(sp_noise1)

white_noise1 = np.float32(white_img) - np.float32(img)
white_noise1 = np.where(white_noise1 == 0, 0, 255)
white_noise1 = np.uint8(white_noise1)
cv.imshow('white_noise1',white_noise1)
cv.imshow('sp_noise1',sp_noise1)
cv.waitKey(0)
cv.destroyAllWindows()

cv.imshow('img',img)
cv.imshow('white_img',white_img)
cv.imshow('sp_img',sp_img)
#  cv.imshow('white_noise',white_noise)
#  cv.imshow('sp_noise',sp_noise)
cv.waitKey(0)
cv.destroyAllWindows()

img_sp_blur = cv.blur(sp_img,(3,3))
img_white_blur = cv.blur(white_img,(3,3))
cv.imshow('img',img)
cv.imshow('white_img',white_img)
cv.imshow('sp_img',sp_img)
cv.imshow('img_sp_blur',img_sp_blur)
cv.imshow('img_white_blur',img_white_blur)
cv.waitKey(0)
cv.destroyAllWindows()


img_sp_med = cv.medianBlur(sp_img, 3)
img_white_med = cv.medianBlur(white_img, 3)
cv.imshow('img',img)
cv.imshow('white_img',white_img)
cv.imshow('sp_img',sp_img)
cv.imshow('img_sp_med',img_sp_med)
cv.imshow('img_white_med',img_white_med)
cv.waitKey(0)
cv.destroyAllWindows()

img_sp_gs = cv.GaussianBlur(sp_img,(5,5),1)
img_white_gs = cv.GaussianBlur(white_img,(5,5),1)
cv.imshow('img',img)
cv.imshow('white_img',white_img)
cv.imshow('sp_img',sp_img)
cv.imshow('img_sp_gs',img_sp_gs)
cv.imshow('img_white_gs',img_white_gs)
cv.waitKey(0)
cv.destroyAllWindows()

cv.imwrite('white_noise1.jpg',white_noise1)
cv.imwrite('sp_noise1.jpg',sp_noise1)
cv.imwrite('white_img.jpg',white_img)
cv.imwrite('sp_img.jpg',sp_img)

cv.imwrite('img_sp_blur.jpg',img_sp_blur)
cv.imwrite('img_white_blur.jpg',img_white_blur)
cv.imwrite('img_sp_med.jpg',img_sp_med)
cv.imwrite('img_white_med.jpg',img_white_med)
cv.imwrite('img_sp_gs.jpg',img_sp_gs)
cv.imwrite('img_white_gs.jpg',img_white_gs)

dx = cv.Sobel(sp_img,-1,1,0,3)
dy = cv.Sobel(sp_img,-1,0,1,3)
d = cv.addWeighted(dx,0.5,dy,0.5,0)
d = cv.convertScaleAbs(d)
cv.imshow('img_gs_med',d)
cv.waitKey(0)
cv.destroyAllWindows()