1. 程式人生 > 其它 >python opencv 實現Reinhard顏色遷移演算法

python opencv 實現Reinhard顏色遷移演算法

技術標籤:影象特徵

https://www.cnblogs.com/likethanlove/p/6003677.html

Reinhard顏色遷移演算法的過程很簡單,流程如下,細節部分見原文,題目為color transfer between images:

  1. 將參考圖片和目標圖片轉換到LAB空間下
  2. 得到參考圖片和目標圖片的均值和標準差
  3. 對目標圖片的每一個畫素值,減去目標影象均值然後乘上參考圖片和目標圖片標準差的比值,再加上參考影象均值
  4. 將目標圖片轉換到RGB空間

將RGB圖片轉換到LAB空間很重要,因為LAB空間能降低三原色之間的相關性,如果不轉換,結果會有很大的不同

複製程式碼

# -*- coding: utf-8 -*-


import cv2
import numpy as np
image = cv2.imread('des.jpg')
image = cv2.cvtColor(image,cv2.COLOR_BGR2LAB)
original = cv2.imread('src.jpg')
original = cv2.cvtColor(original,cv2.COLOR_BGR2LAB)


def getavgstd(image):    //得到均值和標準差
    avg = []
    std = []
    image_avg_l = np.mean(image[:,:,0])
    image_std_l = np.std(image[:,:,0])
    image_avg_a = np.mean(image[:,:,1])
    image_std_a = np.std(image[:,:,1])
    image_avg_b = np.mean(image[:,:,2])
    image_std_b = np.std(image[:,:,2])
    avg.append(image_avg_l)
    avg.append(image_avg_a)
    avg.append(image_avg_b)
    std.append(image_std_l)
    std.append(image_std_a)
    std.append(image_std_b)
    return (avg,std)

image_avg,image_std = getavgstd(image)
original_avg,original_std = getavgstd(original)

height,width,channel = image.shape
for i in range(0,height):
    for j in range(0,width):
        for k in range(0,channel):
            t = image[i,j,k]
            t = (t-image_avg[k])*(original_std[k]/image_std[k]) + original_avg[k]
            t = 0 if t<0 else t
            t = 255 if t>255 else t
            image[i,j,k] = t
image = cv2.cvtColor(image,cv2.COLOR_LAB2BGR)
cv2.imwrite('out.jpg',image)