空域增强

from sklearn import preprocessing
import cv2
import matplotlib.pyplot as plt
import numpy as np
from skimage import data, filters, io

# 函数
def im2double(image):
    info = np.iinfo(image.dtype)              # 获取输入图像的数据类型
    return image.astype(np.float) / info.max  # 将所有值除以数据类型中最大可能的值

def rgb2gray(rgb):
    return np.dot(rgb[..., :3], [0.299, 0.587, 0.114])


image = cv2.imread('image01.png')			# 这里是原图的路径名
image = im2double(image)
image = rgb2gray(image)
cv2.imshow("Origin", image)
cv2.waitKey(0)

kernel = np.array([[-1, -1, -1], [-1, 8, -1], [-1, -1, -1]])
laplace = cv2.filter2D(image.astype('float32'), -1,
                       kernel, borderType=cv2.BORDER_CONSTANT)

cv2.imshow('laplacian', laplace)
cv2.waitKey(0)

# 加法操作
add_img = image + laplace
cv2.imshow("add_img", add_img)
cv2.waitKey(0)

# 单方向x求梯度 sobel
sobel_img = filters.sobel(image)
# sobel_img = cv2.Sobel(image, cv2.CV_64F, 0, 1) 出现的结果不一样
cv2.imshow("sobel_img", sobel_img)
cv2.waitKey(0)

# 使用5*5均值滤波器平滑后的Sobel图像
average_img = cv2.blur(sobel_img, (3, 3)) #可以更改核的大小
cv2.imshow("average_img", average_img)
cv2.waitKey(0)

# 乘法运算 sobel平滑 和 拉普拉斯锐化
mask_img = add_img * average_img
cv2.imshow("mask_img", mask_img)
cv2.waitKey(0)

# 加法运算 原图 和 锐化加强mask_img 
sharp_img = image + mask_img
cv2.imshow("sharp_img", sharp_img)
cv2.waitKey(0)

# 进行灰度幂律变换
power_img = sharp_img ** 0.5
cv2.imshow("power_img", power_img)
cv2.waitKey(0)