import cv2 as cv
import numpy as np
import matplotlib.pyplot as plt
img = cv.imread("../SampleImages/pomeranian.png", cv.IMREAD_COLOR)
rows,cols,channels = img.shape
print(rows,cols,channels)
#为图像添加高斯噪声
#使用np.random.normal(loc=0.0, scale=1.0, size=None)
# loc: 高斯分布中心点,分布的均值
# scale: 高斯分布的宽度,分布的标准差
# size:维度。如果给定维度是(m,n,k)则从分布中抽取m*n*k个样本
#参考资料:https://blog.csdn.net/wzy628810/article/details/103807829
# https://blog.csdn.net/sinat_29957455/article/details/123977298
def AddGaussianNoise(image, mean=0, var=0.005):
image = np.array(image/255, dtype=float) #将像素值归一
noise = np.random.normal(mean, var ** 0.5, image.shape) #产生高斯噪声
out = image + noise #直接将归一化的图片与噪声相加
if out.min() < 0:
low_clip = -1.
else:
low_clip = 0.
out = np.clip(out, low_clip, 1.0)
out = np.uint8(out*255)
return out
img_gaussian_noise = img.copy()
gauss_mean = 0
gauss_sigma = 0.003
#增加高斯噪声到图像
img_gaussian_noise = AddGaussianNoise(img_gaussian_noise, gauss_mean, gauss_sigma)
#高斯滤波(高斯模糊)
#cv.GaussianBlur(src, ksize, sigmaX, sigmaY, borderType)
#src: 输入图像
#ksize: kernel大小,高斯卷积和大小。注意卷积核的宽度和高度可以不同,但必须为正数且为奇数,也可以为零。
#sigmaX/Y: X和Y方向上的高斯标准差
#参考资料:https://blog.csdn.net/weixin_52012241/article/details/122284713
img_gaussian_blur_origin = cv.GaussianBlur(img, (3,3), 0)
img_gaussian_blur_noise = cv.GaussianBlur(img_gaussian_noise, (13,13), 0.006)
#显示图像
fig,axes = plt.subplots(nrows=2, ncols=2, figsize=(10,10), dpi=100)
axes[0][0].imshow(img[:,:,::-1])
axes[0][0].set_title("Original")
axes[0][1].imshow(img_gaussian_blur_origin[:,:,::-1])
axes[0][1].set_title("Original Gaussian Blurred")
axes[1][0].imshow(img_gaussian_noise[:,:,::-1])
axes[1][0].set_title("Add Gaussian Noise")
axes[1][1].imshow(img_gaussian_blur_noise[:,:,::-1])
axes[1][1].set_title("Gaussian Noise Blurred")