주근깨 얼굴 이미지 수정 - 파이썬 코드

파이썬 프로그램을 실행할 수만 있다면 누구나 쉽게 주근깨 얼굴의 이미지도 깔끔하게 

 

 

# Enter your code here

import cv2

import matplotlib.pyplot as plt

import numpy as np

import math

import os

ImagePath =None

winname = "BlemishRemover"

def EdgesMatricExtracter (Mat):

Edgesmat = np.zeros_like(Mat)

Edgesmat[:,0] = Mat[:,0]

Edgesmat[0,:] = Mat[0,:]

Edgesmat[:,-1] = Mat[:,-1]

Edgesmat[-1,:] = Mat[-1,:]

return Edgesmat

def Extractneighbour(Source,x,y,r):

SrC_hieght,Src_width=Source.shape[:2]

X_Start = max(x-r,0)

Y_Start = max(y-r,0)

X_End = min(x+r+1,SrC_hieght)

Y_End = min(y+r+1,Src_width)

Neighbourhood = Source[X_Start:X_End,Y_Start:Y_End]

return Neighbourhood

def BlemishRemover(action, x, y, flags, userdata):

global source

if action == cv2.EVENT_LBUTTONDOWN:

radii = 30

while(True):

BlemishROI = Extractneighbour(source, y, x, radii) # Ensure (row, col) order

Boundary_BlemishROI = EdgesMatricExtracter(BlemishROI)

Boundary_BlemishROI = Boundary_BlemishROI.astype(np.float32) / 255.0

candidate_neighbors = []

ROI_Mask = 255 * np.ones(BlemishROI.shape, BlemishROI.dtype)

 

for i in range(-5, 6):

for j in range(-5, 6):

if i == 0 or j == 0: continue

X_inc = radii * (i / abs(i)) + i

Y_inc = radii * (j / abs(j)) + j

NewCenter = (y + X_inc, x + Y_inc) # Maintain correct order

New_Nei = Extractneighbour(source, int(NewCenter[0]), int(NewCenter[1]), radii)

 

if New_Nei.shape == BlemishROI.shape:

candidate_neighbors.append(New_Nei)

 

selected_neighbor = None

Xgradient_BlemishROI = cv2.Sobel(BlemishROI, cv2.CV_32F, 1, 0)

Ygradient_BlemishROI = cv2.Sobel(BlemishROI, cv2.CV_32F, 0, 1)

MinX_range = abs(np.max(Xgradient_BlemishROI) - np.min(Xgradient_BlemishROI))

MinY_range = abs(np.max(Ygradient_BlemishROI) - np.min(Ygradient_BlemishROI))

for candidate in candidate_neighbors:

Boundary_candidate = EdgesMatricExtracter(candidate)

Boundary_candidate = Boundary_candidate.astype(np.float32) / 255.0

Dist = np.linalg.norm(Boundary_candidate - Boundary_BlemishROI)

 

if Dist < 1.5:

Xgradient_candidate = cv2.Sobel(candidate, cv2.CV_32F, 1, 0)

Ygradient_candidate = cv2.Sobel(candidate, cv2.CV_32F, 0, 1)

MinX_range_Candidate = abs(np.max(Xgradient_candidate) - np.min(Xgradient_candidate))

MinY_range_Candidate = abs(np.max(Ygradient_candidate) - np.min(Ygradient_candidate))

 

if MinX_range_Candidate <= MinX_range and MinY_range_Candidate <= MinY_range:

MinX_range = MinX_range_Candidate

MinY_range = MinY_range_Candidate

selected_neighbor = candidate

 

if selected_neighbor is not None:

source = cv2.seamlessClone(selected_neighbor, source, ROI_Mask, (x, y), cv2.NORMAL_CLONE)

break

radii = math.ceil(radii/2)

if(radii<4): break

 

cv2.imshow(winname, source)

def RunGUI_BlmeshRemover_RunGUI(path):

global source

source = cv2.imread(path,1)

dummy = source.copy()

cv2.namedWindow(winname)

cv2.setMouseCallback(winname, BlemishRemover)

k=0

while k!=27 :

cv2.imshow(winname,source)

k = cv2.waitKey(20) & 0xFF

if k==101 or k==69:

source= dummy.copy()

cv2.imshow(winname, source)

cv2.destroyAllWindows()

if __name__ == "__main__":

ImagePath = "face-002.jpg"

RunGUI_BlmeshRemover_RunGUI(ImagePath)

skin-color-005.py

0.00MB