Simple detection of objects by color
Good day.
In this short post I wanted to show a simple way to find objects by color with OpenCV .
For experiments used the camera Logitech WebCam C270
so, let's begin
')
We connect everything you need
Declare variables
Create trackbar to adjust color
Open the camera
Run the camera loop
In this loop, first convert the frame from RGB to HSV. Then we blur and call the color search function. The inRange () function does a full-blown search, i.e. from what color to which to allocate pixels. Then I process each pixel of the frame, if the pixel is white, fill it with gray. Well, using the rectangle () function, select the object by simply drawing a rectangle around the selected area.
Here is an example of a lemon search:
In this short post I wanted to show a simple way to find objects by color with OpenCV .
For experiments used the camera Logitech WebCam C270
so, let's begin
')
We connect everything you need
#include <opencv2/opencv.hpp> #include <iostream> #include <cstdlib> // exit() using namespace cv; using std::cout; using std::endl; Declare variables
char mainWindow[] = "Main"; char trackbarWindow[] = "Trackbar"; char thresholdWindow[] = "Threshold"; int min = 0, max = 1000; int hmin = 0, smin = 0, vmin = 0, hmax = 255, smax = 255, vmax = 255; Mat frame, HSV, threshold, blurred; VideoCapture capture; Create trackbar to adjust color
createTrackbar("H min:", trackbarWindow, &hmin, hmax); createTrackbar("H max:", trackbarWindow, &hmax, hmax); createTrackbar("S min:", trackbarWindow, &smin, smax); createTrackbar("S max:", trackbarWindow, &smax, smax); createTrackbar("V min:", trackbarWindow, &vmin, vmax); createTrackbar("V max:", trackbarWindow, &vmax, vmax); createTrackbar("Size min:", trackbarWindow, &min, max); createTrackbar("Size max:", trackbarWindow, &max, max); Open the camera
capture.open(1); if(!capture.isOpened()){ cout << " ." << endl; exit(1); } Run the camera loop
for(;;){ capture >> frame; cvtColor(frame, HSV, COLOR_BGR2HSV); medianBlur(HSV, blurred, 21); inRange(blurred, Scalar(hmin, smin, vmin), Scalar(hmax, smax, vmax), threshold); for(int y = 0; y < threshold.rows; y++){ for(int x = 0; x < threshold.cols; x++){ int value = threshold.at<uchar>(y, x); if(value == 255){ Rect rect; int count = floodFill(threshold, Point(x, y), Scalar(200), &rect); if(rect.width >= min && rect.width <= max && rect.height >= min && rect.height <= max){ rectangle(frame, rect, Scalar(255, 0, 255, 4)); } } } } imshow(mainWindow, frame); imshow(thresholdWindow, threshold); if(waitKey(33) == 27) break; } In this loop, first convert the frame from RGB to HSV. Then we blur and call the color search function. The inRange () function does a full-blown search, i.e. from what color to which to allocate pixels. Then I process each pixel of the frame, if the pixel is white, fill it with gray. Well, using the rectangle () function, select the object by simply drawing a rectangle around the selected area.
Here is an example of a lemon search:
Full source code
#include <opencv2/opencv.hpp> #include <iostream> #include <cstdlib> using namespace cv; using std::cout; using std::endl; int main(int argc, char **argv){ char mainWindow[] = "Main"; char trackbarWindow[] = "Trackbar"; char thresholdWindow[] = "Threshold"; int min = 0, max = 1000; int hmin = 0, smin = 0, vmin = 0, hmax = 255, smax = 255, vmax = 255; Mat frame, HSV, threshold, blurred; VideoCapture capture; // namedWindow(mainWindow, 0); namedWindow(trackbarWindow, 0); namedWindow(thresholdWindow, 0); // createTrackbar("H min:", trackbarWindow, &hmin, hmax); createTrackbar("H max:", trackbarWindow, &hmax, hmax); createTrackbar("S min:", trackbarWindow, &smin, smax); createTrackbar("S max:", trackbarWindow, &smax, smax); createTrackbar("V min:", trackbarWindow, &vmin, vmax); createTrackbar("V max:", trackbarWindow, &vmax, vmax); createTrackbar("Size min:", trackbarWindow, &min, max); createTrackbar("Size max:", trackbarWindow, &max, max); // capture.open(1); if(!capture.isOpened()){ cout << " ." << endl; exit(1); } // for(;;){ capture >> frame; cvtColor(frame, HSV, COLOR_BGR2HSV); medianBlur(HSV, blurred, 21); inRange(blurred, Scalar(hmin, smin, vmin), Scalar(hmax, smax, vmax), threshold); for(int y = 0; y < threshold.rows; y++){ for(int x = 0; x < threshold.cols; x++){ int value = threshold.at<uchar>(y, x); if(value == 255){ Rect rect; int count = floodFill(threshold, Point(x, y), Scalar(200), &rect); if(rect.width >= min && rect.width <= max && rect.height >= min && rect.height <= max){ rectangle(frame, rect, Scalar(255, 0, 255, 4)); } } } } imshow(mainWindow, frame); imshow(thresholdWindow, threshold); if(waitKey(33) == 27) break; } return 0; } Source: https://habr.com/ru/post/245139/
All Articles