OpenCV和Halcon分别實作彩色圖像的白平衡效果

實作白平衡算法中的灰階世界法，能有效改善圖像發紅/發藍/發綠的現象

1、OpenCV

#include <opencv2/opencv.hpp>
using namespace cv;
int main()
{
    Mat g_srcImage,dstImage;
    vector<Mat> g_vChannels;
    g_srcImage = imread("C:/Users/Administrator/Desktop/01.jpg");
    imshow("原圖",g_srcImage);
    //waitKey(0);
    //分離通道
    split(g_srcImage, g_vChannels);
    Mat imageBlueChannel = g_vChannels.at(0);
    Mat imageGreenChannel = g_vChannels.at(1);     
    Mat imageRedChannel = g_vChannels.at(2);
    double imageBlueChannelAvg = 0;
    double imageGreenChannelAvg = 0;
    double imageRedChannelAvg = 0;
    //求各通道的平均值
    imageBlueChannelAvg = mean(imageBlueChannel)[0];
    imageGreenChannelAvg = mean(imageGreenChannel)[0];
    imageRedChannelAvg = mean(imageRedChannel)[0];
    //求出個通道所占增益
    double K = (imageRedChannelAvg+imageGreenChannelAvg+imageRedChannelAvg) / 3;
    double Kb = K / imageBlueChannelAvg;
    double Kg = K / imageGreenChannelAvg;
    double Kr = K / imageRedChannelAvg;
    //更新白平衡後的各通道BGR值
    addWeighted(imageBlueChannel,Kb,0,0,0,imageBlueChannel);
    addWeighted(imageGreenChannel,Kg,0,0,0,imageGreenChannel);
    addWeighted(imageRedChannel,Kr,0,0,0,imageRedChannel);
    merge(g_vChannels,dstImage);//圖像各通道合并
    imshow("白平衡後圖",dstImage);
    waitKey(0);
    return 0;
}      

API詳解：

void cvAddWeighted( const CvArr* src1, double alpha,const CvArr* src2, double beta,double gamma, CvArr* dst );

參數1：src1，圖1

參數2：alpha，圖1數組元素權重

參數3：src2，圖2

參數4：beta，圖2數組元素權重

參數5：gamma，圖1與圖2疊加之後再添加的數值。不要太大，不然圖檔一片白。總和等于255以上就是純白色了。

參數6：dst，輸出圖檔

即：目标圖=src1*alpha+src2*beta+gamma

2、Halcon

dev_close_window ()
read_image (Image, 'D:/hellowprld/2/test777.jpg')
dev_open_window_fit_image (Image, 0, 0, -1, -1, WindowHandle)
dev_display (Image)
*将圖像進行通道分解，分别轉換為三個通道的RGB圖像
decompose3 (Image, Red, Green, Blue)
*實作白平衡算法中的灰階世界法，能有效改善圖像發紅/發藍/發綠的現象
*取RGB各個通道的平均值
get_domain (Red, Domain)
intensity (Domain, Red, MeanRed, DeviationRed)
get_domain (Green, Domain)
intensity (Domain, Green, MeanGreen, DeviationGreen)
get_domain (Blue, Domain)
intensity (Domain, Blue, MeanBlue, DeviationBlue)
*求出個通道所占增益
K := (MeanRed + MeanGreen + MeanBlue) / 3.0
Kr := K / MeanRed
Kg := K / MeanGreen
Kb := K / MeanBlue
*更新白平衡後的各通道值White Balance
scale_image (Red, ImageScaledRed, Kr, 0)
scale_image (Green, ImageScaledGreen, Kg, 0)
scale_image (Blue, ImageScaledBlue, Kb, 0)
compose3(ImageScaledRed, ImageScaledGreen, ImageScaledBlue, Multichannel0)
write_image (Multichannel0, 'jpeg 100', 0, 'D:/opt.jpg')
stop()      

---------------附錄--------------------

兩幅圖像之間處理的算子

1.sub_image(ImageMinuend, ImageSubtrahend : ImageSub : Mult, Add : )

對兩幅圖像做減法   g' := (g1 - g2) * Mult + Add

程式如下：

read_image (Scene00, 'autobahn/scene_00')

read_image (Scene01, 'autobahn/scene_01')

sub_image (Scene00, Scene01, ImageSub1, 1, 0)

dev_display(ImageSub1)

2.abs_image(Image : ImageAbs : : )

計算圖像的絕對值模型

3.crop_part(Image : ImagePart : Row, Column, Width, Height : )

剪切出一個長方形的圖像

4.add_image(Image1, Image2 : ImageResult : Mult, Add : )

兩圖像相疊加 g' := (g1 + g2) * Mult + Add

5.max_image(Image1, Image2 : ImageMax : : )

計算兩幅圖像每個像素點的最大值

6.min_image(Image1, Image2 : ImageMin : : )

計算兩幅圖像每個像素點的最小值

7.div_image(Image1, Image2 : ImageResult : Mult, Add : )

兩幅圖像相除   g' := g1 / g2 * Mult + Add

8.mult_image(Image1, Image2 : ImageResult : Mult, Add : )

兩幅圖像相乘   g' := g1 * g2 * Mult + Add