Cropping an triangle from captured frame - OpenCV and C++

Question

I have a video file from which I'm capturing a frames. I want to crop a triangle from captured frame and display it, but my program shows just a source frame.

Here is my code:

cv::Mat Detector::cropRegionOfInterest(cv::Mat& frame)
{
    cv::Point corners[1][3];
    corners[0][0] = cv::Point(0, frameHeight);
    corners[0][1] = cv::Point(frameWidth, frameHeight);
    corners[0][2] = cv::Point(frameWidth / 2, frameHeight / 2);

    const cv::Point* cornerList[1] = { corners[0] };

    int numPoints = 3;
    int numPolygons = 1;

    cv::Mat mask(frame.size(), CV_8UC1, cv::Scalar(0, 0, 0));
    cv::fillPoly(mask, cornerList, &numPoints, numPolygons, cv::Scalar(255, 255, 255), 8);

    cv::Mat result(frame.size(), CV_8UC3);
    cv::bitwise_and(frame, mask, result);

    return result;
}

Instead of displaying source frame I want it to display cropped triangle.

Answer 1

Since you're using CV_8UC3 as the type of result, I'm assuming (see the Edit at the end of the answer if that's not the case) that the input image frame also has 3 channels. In that case, I'm a bit surprised that you can even see the non-cropped image, as running your code simply throws an exception on my machine at the call to bitwise_and:

OpenCV(3.4.1) Error: Sizes of input arguments do not match

From the documentation, it seems to me that you can't mix different input and mask types. A quick and dirty solution is to split the input image into a vector of three channels, call bitwise_and for each of them, and then merge them back. The code below works for me:

#include <stdio.h>
#include <opencv2/opencv.hpp>

using namespace cv;


cv::Mat cropRegionOfInterest(cv::Mat& frame)
{

    const int frameWidth=frame.cols-1;
    const int frameHeight=frame.rows-1;
    cv::Point corners[1][3];
    corners[0][0] = cv::Point(0, frameHeight);
    corners[0][1] = cv::Point(frameWidth, frameHeight);
    corners[0][2] = cv::Point(frameWidth / 2, frameHeight / 2);

    const cv::Point* cornerList[1] = { corners[0] }; 

    int numPoints = 3; 
    int numPolygons = 1; 

    cv::Mat mask(frame.rows,frame.cols, CV_8UC1, cv::Scalar(0, 0, 0));
    cv::fillPoly(mask, cornerList, &numPoints, numPolygons, cv::Scalar(255, 255, 255), 8);

    std::vector<cv::Mat> src_channels;
    std::vector<cv::Mat> result_channels;
    cv::split(frame,src_channels);

    for(int idx=0;idx<3;++idx)
    {  
      result_channels.emplace_back(frame.rows,frame.cols,CV_8UC1);
      cv::bitwise_and(src_channels[idx], mask,result_channels[idx]);

    }  
    cv::Mat result;
    cv::merge(result_channels,result);
    return result;
}


int main(int argc, char** argv )
{
    if ( argc != 2 )
    {  
        printf("usage: DisplayImage.out <Image_Path>\n");
        return -1;
    }  

    Mat image;
    image = imread( argv[1], 1 ); 

    if ( !image.data )
    {  
        printf("No image data \n");
        return -1;
    }  

    cv::Mat cropped=cropRegionOfInterest(image);

    namedWindow("cropped Image", WINDOW_AUTOSIZE ); 
    imshow("cropped Image", cropped);

    waitKey(0);

    return 0; 
}

---

Edit: From your comments it seems that frame is actually grayscale. In that case, nevermind all the code above, and just change cv::Mat result(frame.size(), CV_8UC3); to

cv::Mat result(frame.rows,frame.cols,CV_8UC1);

in your original code.