import cv2
img=cv2.imread('bird.jpg')
cv2.imshow('bird',img)
gray = cv2.cvtColor(img, cv2.COLOR_BGR2GRAY)
# Blur the image using Gaussian Blur
#高斯核服從正態(tài)分布，核數(shù)越大，越模糊
gray_blur = cv2.GaussianBlur(gray, (75, 75), 0)
# Convert the image into pencil sketch
cartoon = cv2.divide(gray, gray_blur, scale=250.0)
cv2.imshow('cartoon.jpg',cartoon)
cv2.waitKey()
cv2.destroyAllWindows()

import cv2
img=cv2.imread('bird.jpg')
cv2.imshow('bird',img)
#convert the image into grayscale image
gray = cv2.cvtColor(img, cv2.COLOR_BGR2GRAY)
#Blur the grayscale image with median blur中值濾波
gray_blur = cv2.medianBlur(gray, 3)
#Apply adaptive thresholding to detect edges檢測(cè)圖像邊緣
edges = cv2.adaptiveThreshold(gray_blur, 255, cv2.ADAPTIVE_THRESH_MEAN_C,cv2.THRESH_BINARY, 9, 9)#自適應(yīng)均值濾波
#Sharpen the image銳化圖像
color = cv2.detailEnhance(img, sigma_s=5, sigma_r=0.5)
cv2.imshow('color',color)
#細(xì)節(jié)增強(qiáng)函數(shù) sigma_s控制著鄰域的大小，該鄰域的大小將被加權(quán)以替換圖像中的像素值。值越高，鄰域越大。這樣可以使圖像更平滑
#：如果要在平滑圖像時(shí)保留邊緣，這很重要。較小的值只會(huì)產(chǎn)生非常相似的顏色進(jìn)行平均（即平滑），而相差很大的顏色將保持不變
#Merge the colors of same images using "edges" as a mask
cartoon = cv2.bitwise_and(color, color, mask=edges)
cv2.imshow('cartoon',cartoon)
cv2.waitKey()
cv2.destroyAllWindows()

import cv2
img=cv2.imread('bird.jpg')

gray = cv2.cvtColor(img, cv2.COLOR_BGR2GRAY)
cv2.imshow('img.jpg', img )

gray = cv2.medianBlur(gray, 3)
edges = cv2.Laplacian(gray, -1, ksize=5)
edges_inv = 255 - edges
dummy, cartoon = cv2.threshold(edges_inv, 0, 255, cv2.THRESH_BINARY)
cv2.imshow('cartoon.jpg', cartoon )
cv2.waitKey()
cv2.destroyAllWindows()