from itertools import chain
import cv2
import os
import numpy as np
import datetime
class BrightnessBalance:
def __init__(self):
pass
def arrayToHist(self,gray):
'''
计算灰度直方图，并归一化
:param gray_path:
:return:
'''
w,h = gray.shape
hist = list(chain.from_iterable(cv2.calcHist([gray],[0],None,[256],[0,256])))
hist = np.array(hist)/(w*h)
return hist
def histMatch(self, gray1, gray2):
'''
gray2向gray1校准
:param image_path1:
:param image_path2:
:return:
'''
if not os.path.exists(image1_path) or not os.path.exists(image2_path):
print('image1_path or image2_path is not exist!')
return
gray2Array = np.array(gray2)
hist1 = self.arrayToHist(gray1)
hist2 = self.arrayToHist(gray2Array)
tmp1 = 0.0
tmp2 = 0.0
h_acc1 = hist1.copy()
h_acc2 = hist2.copy()
for i in range(256):
tmp1 += hist1[i]
tmp2 += hist2[i]
h_acc1[i] = tmp1
h_acc2[i] = tmp2
M = np.zeros(256)
for i in range(256):
idx = 0
minv = 1
for j in range(0,len(h_acc1)):
if (np.fabs(h_acc1[j] - h_acc2[i]) < minv):
minv = np.fabs(h_acc1[j] - h_acc2[i])
idx = int(j)
M[i] = idx
result = M[gray2Array]
return result
def colors_histMatch(self,image_path1,image_path2):
if not os.path.exists(image1_path) or not os.path.exists(image2_path):
print('image1_path or image2_path is not exist!')
return
img1 = cv2.imread(image1_path)
img2 = cv2.imread(image2_path)
result1_chaneel_b = self.histMatch(img1[:, :, 0], img2[:, :, 0])
result1_chaneel_g = self.histMatch(img1[:, :, 1], img2[:, :, 1])
result1_chaneel_r = self.histMatch(img1[:, :, 2], img2[:, :, 2])
result = cv2.merge([result1_chaneel_b,result1_chaneel_g,result1_chaneel_r])
return result
if __name__ == '__main__':
image1_path = r"D:\CV_Data\n\current.jpg"
image2_path = r"D:\CV_Data\n\history.jpg"
starttime = datetime.datetime.now()
BB = BrightnessBalance()
img1 = cv2.imread(image1_path,0)
img2 = cv2.imread(image2_path,0)
histMatch = BB.histMatch(img1,img2)
endtime = datetime.datetime.now()
print(endtime - starttime)
save_path = image2_path.replace('history.jpg', 'result1.jpg')
cv2.imwrite(save_path, histMatch)
