Python实现特定场景去除高光算法详解

算法思路

1、求取源图I的平均灰度，并记录rows和cols；

2、按照一定大小，分为N*M个方块，求出每块的平均值，得到子块的亮度矩阵D；

3、用矩阵D的每个元素减去源图的平均灰度，得到子块的亮度差值矩阵E；

4、通过插值算法，将矩阵E差值成与源图一样大小的亮度分布矩阵R；

5、得到矫正后的图像result=I-R；

应用场景

光照不均匀的整体色泽一样的物体，比如工业零件，ocr场景。

代码实现

import cv2
import numpy as np
 
def unevenLightCompensate(gray, blockSize):
    #gray = cv2.cvtColor(img, cv2.COLOR_BGR2GRAY)
    average = np.mean(gray)
    rows_new = int(np.ceil(gray.shape[0] / blockSize))
    cols_new = int(np.ceil(gray.shape[1] / blockSize))
    blockImage = np.zeros((rows_new, cols_new), dtype=np.float32)
    for r in range(rows_new):
        for c in range(cols_new):
            rowmin = r * blockSize
            rowmax = (r + 1) * blockSize
            if (rowmax > gray.shape[0]):
                rowmax = gray.shape[0]
            colmin = c * blockSize
            colmax = (c + 1) * blockSize
            if (colmax > gray.shape[1]):
                colmax = gray.shape[1]
            imageROI = gray[rowmin:rowmax, colmin:colmax]
            temaver = np.mean(imageROI)
 
            blockImage[r, c] = temaver
 
 
    
    blockImage = blockImage - average
    blockImage2 = cv2.resize(blockImage, (gray.shape[1], gray.shape[0]), interpolation=cv2.INTER_CUBIC)
    gray2 = gray.astype(np.float32)
    dst = gray2 - blockImage2
    dst[dst>255]=255
    dst[dst<0]=0
    dst = dst.astype(np.uint8)
    dst = cv2.GaussianBlur(dst, (3, 3), 0)
    #dst = cv2.cvtColor(dst, cv2.COLOR_GRAY2BGR)
    return dst
 
if __name__ == '__main__':
    file = 'www.png'
    blockSize = 8
    img = cv2.imread(file)
    b,g,r = cv2.split(img)
    dstb = unevenLightCompensate(b, blockSize)
    dstg = unevenLightCompensate(g, blockSize)
    dstr = unevenLightCompensate(r, blockSize)
    dst = cv2.merge([dstb, dstg, dstr])
    result = np.concatenate([img, dst], axis=1)
cv2.imwrite('result.jpg', result)

实验效果

补充

OpenCV实现光照去除效果

1.方法一(RGB归一化)

int main(int argc, char *argv[])
{
	//double temp = 255 / log(256);
	//cout << "doubledouble temp ="<< temp<(i, j)[0] = 255 * (float)image.at(i, j)[0] / ((float)image.at(i, j)[0] + (float)image.at(i, j)[2] + (float)image.at(i, j)[1]+0.01);
			src.at(i, j)[1] = 255 * (float)image.at(i, j)[1] / ((float)image.at(i, j)[0] + (float)image.at(i, j)[2] + (float)image.at(i, j)[1]+0.01);
			src.at(i, j)[2] = 255 * (float)image.at(i, j)[2] / ((float)image.at(i, j)[0] + (float)image.at(i, j)[2] + (float)image.at(i, j)[1]+0.01);
		}
	}
	
	normalize(src, src, 0, 255, CV_MINMAX);
      
	convertScaleAbs(src,src);
	imshow("rgb", src);
	imwrite("C://Users//TOPSUN//Desktop//123.jpg", src);
	waitKey(0);
	return 0;
}

实现效果

2.方法二

void unevenLightCompensate(Mat &image, int blockSize)
{
	if (image.channels() == 3) cvtColor(image, image, 7);
	double average = mean(image)[0];
	int rows_new = ceil(double(image.rows) / double(blockSize));
	int cols_new = ceil(double(image.cols) / double(blockSize));
	Mat blockImage;
	blockImage = Mat::zeros(rows_new, cols_new, CV_32FC1);
	for (int i = 0; i < rows_new; i++)
	{
		for (int j = 0; j < cols_new; j++)
		{
			int rowmin = i*blockSize;
			int rowmax = (i + 1)*blockSize;
			if (rowmax > image.rows) rowmax = image.rows;
			int colmin = j*blockSize;
			int colmax = (j + 1)*blockSize;
			if (colmax > image.cols) colmax = image.cols;
			Mat imageROI = image(Range(rowmin, rowmax), Range(colmin, colmax));
			double temaver = mean(imageROI)[0];
			blockImage.at(i, j) = temaver;
		}
	}
	blockImage = blockImage - average;
	Mat blockImage2;
	resize(blockImage, blockImage2, image.size(), (0, 0), (0, 0), INTER_CUBIC);
	Mat image2;
	image.convertTo(image2, CV_32FC1);
	Mat dst = image2 - blockImage2;
	dst.convertTo(image, CV_8UC1);
}
int main(int argc, char *argv[])
{
	//double temp = 255 / log(256);
	//cout << "doubledouble temp ="<< temp<