OPenCV4-颜色识别（二）三原色识别

OPenCV4-颜色识别（二）三原色识别

这次颜色识别使用C++来实现。颜色识别简单来说就是在 HSV 空间中找到符合颜色的阈值合适区域罢了。可以划分为以下几个步骤：

1、首先获取到图像

2、将获取到的图像颜色空间进行转换，转换为方便识别的 HSV 颜色模型

3、筛选阈值合适区域，作为识别的颜色

4、形态学滤波，对识别的区域进行滤波处理

5、提取被识别区域的轮廓，并标识出来

接下来实现一下吧！

1.带有滑动栏的颜色识别

这个实现主要是用来选择合适的阈值区间的。可以通过调节滑动栏的各个数值，查看颜色识别的效果。

 #include <iostream>
 #include <opencv2/opencv.hpp>
 ​
 using namespace std;
 using namespace cv;
 ​
 /* RED */
 int iLowH = 156;
 int iLowS = 43;
 int iLowV = 46;
 int iHighH = 180;
 int iHighS = 255;
 int iHighV = 255;
 ​
 int main(int argc, char** argv)
 {
     system("color F0");
 ​
     Mat hsvFrame;
     Mat gaussImg;
     Mat kernel;
     Mat REDimg;
 ​
     vector<vector<Point>> REDcontours;
 ​
     /* 1.获取视频源 */
     VideoCapture camera(0);
     if (!camera.isOpened())
     {
         cout << "打开摄像头失败，请确认摄像头是否安装成功" << endl;
         return -1;
     }
     camera.set(CAP_PROP_FRAME_WIDTH, 1280);
     camera.set(CAP_PROP_FRAME_HEIGHT, 720);
 ​
     while (1)
     {
         Mat frame;
 ​
         /* 2.获取视频帧 */
         camera >> frame;
         if (frame.empty())
         {
             cout << "没有获取到图像" << endl;
             break;
         }
 ​
         /* 创建显示窗口和滑动条 */
         namedWindow("Color");
         createTrackbar("LowH", "Color", &iLowH, 179);
         createTrackbar("LowS", "Color", &iLowS, 255);
         createTrackbar("LowV", "Color", &iLowV, 255);
         createTrackbar("HighH", "Color", &iHighH, 179);
         createTrackbar("HighS", "Color", &iHighS, 255);
         createTrackbar("HighV", "Color", &iHighV, 255);
 ​
         /* 3.颜色空间转换 BGR 转 HSV */
         cvtColor(frame, hsvFrame, COLOR_BGR2HSV);
 ​
         /* 4.高斯滤波 */
         GaussianBlur(hsvFrame, gaussImg, Size(5, 5), 0);
 ​
         /* 5.筛选阈值合适区域 */
         inRange(gaussImg, Scalar(iLowH, iLowS, iLowV),
                 Scalar(iHighH, iHighS, iHighV),
                 REDimg);
 ​
         /* 6.形态学滤波 */
         kernel = getStructuringElement(MORPH_RECT, Size(3, 3));
         morphologyEx(REDimg, REDimg, MORPH_CLOSE, kernel);
         morphologyEx(REDimg, REDimg, MORPH_OPEN, kernel);
 ​
         /* 7.提取轮廓 */
         findContours(REDimg, REDcontours, RETR_EXTERNAL, CHAIN_APPROX_NONE);
 ​
         /* 8.绘制轮廓图 */
         drawContours(frame, REDcontours, -1, Scalar(0, 0, 255), 2);
 ​
         imshow("video", frame);
         if (27 == waitKey(1000 / 30))
             break;
     }
 ​
     destroyAllWindows();
     camera.release();
 ​
     return 0;
 }

2.三原色识别

这个实现用于识别R、G、B三原色。

#include <iostream>
#include <opencv2/opencv.hpp>
 ​
 using namespace std;
 using namespace cv;
 ​
 /*
  * 颜色阈值
  */
 int colorThreshold[][6] =
 {
     // {hmin, hmax, smin, smax, vmin, vmax}
     {0,   10,  43,  255, 46,  255},     // RED
     {156, 180, 43,  255, 46,  255},     // RED
     {35,  77,  150, 255, 200, 255},     // GREEN
     {100, 124, 43,  255, 46,  255},     // BULE

 };
 ​
 int main(int argc, char** argv)
 {
     system("color F0");
 ​
     Mat hsvFrame;
     Mat gaussImg;
     Mat kernel;
 ​
     Mat REDimg;
     Mat GREENimg;
     Mat BULEimg;
 ​
     vector<vector<Point>> REDcontours;
     vector<vector<Point>> GREENcontours;
     vector<vector<Point>> BLUEcontours;
     
     /* 1.获取视频源 */
     VideoCapture camera(0);
     if (!camera.isOpened())
     {
         cout << "打开摄像头失败，请确认摄像头是否安装成功" << endl;
         return -1;
     }
     camera.set(CAP_PROP_FRAME_WIDTH, 1280);
     camera.set(CAP_PROP_FRAME_HEIGHT, 720);
 ​
     while (1)
     {
         Mat frame;
 ​
         /* 2.获取视频帧 */
         camera >> frame;
         if (frame.empty())
         {
             cout << "没有获取到图像" << endl;
             break;
         }
 ​
         /* 3.颜色空间转换 BGR 转 HSV */
         cvtColor(frame, hsvFrame, COLOR_BGR2HSV);
 ​
         /* 4.高斯滤波 */
         GaussianBlur(hsvFrame, gaussImg, Size(5, 5), 0);
 ​
         /* 5.关键部分：筛选阈值合适区域 */
         inRange(gaussImg,
             Scalar(colorThreshold[0][0], colorThreshold[0][2], colorThreshold[0][4]),
             Scalar(colorThreshold[0][1], colorThreshold[0][3], colorThreshold[0][5]),
             REDimg);
         inRange(gaussImg,
             Scalar(colorThreshold[2][0], colorThreshold[2][2], colorThreshold[2][4]),
             Scalar(colorThreshold[2][1], colorThreshold[2][3], colorThreshold[2][5]),
             GREENimg);
         inRange(gaussImg,
             Scalar(colorThreshold[3][0], colorThreshold[3][2], colorThreshold[3][4]),
             Scalar(colorThreshold[3][1], colorThreshold[3][3], colorThreshold[3][5]),
             BULEimg);
 ​
         /* 6.形态学滤波 */
         kernel = getStructuringElement(MORPH_RECT, Size(3, 3));
         morphologyEx(REDimg, REDimg, MORPH_CLOSE, kernel);
         morphologyEx(REDimg, REDimg, MORPH_OPEN, kernel);
 ​
         morphologyEx(GREENimg, GREENimg, MORPH_CLOSE, kernel);
         morphologyEx(GREENimg, GREENimg, MORPH_OPEN, kernel);
 ​
         morphologyEx(BULEimg, BULEimg, MORPH_CLOSE, kernel);
         morphologyEx(BULEimg, BULEimg, MORPH_OPEN, kernel);
         
 ​
         /* 7.提取轮廓 */
         findContours(REDimg, REDcontours, RETR_EXTERNAL, CHAIN_APPROX_NONE);
         findContours(GREENimg, GREENcontours, RETR_EXTERNAL, CHAIN_APPROX_NONE);
         findContours(BULEimg, BLUEcontours, RETR_EXTERNAL, CHAIN_APPROX_NONE);
 ​
         /* 8.绘制轮廓图 */
         drawContours(frame, REDcontours, -1, Scalar(0, 0, 255), 2);
         drawContours(frame, GREENcontours, -1, Scalar(0, 255, 0), 2);
         drawContours(frame, BLUEcontours, -1, Scalar(255, 0, 0), 2);
 ​
         imshow("video", frame);
         if (27 == waitKey(1000 / 30))
             break;
     }
 ​
     destroyAllWindows();
     camera.release();
 ​
     return 0;
 }