Gamut errors programmaticaly detection

[Valentin Nikin]

Hello!

I want to detect Gamut errors in file. And I try to use OpenCV to do it.
Accordinally EBU R 103 V3 (https://tech.ebu.ch/docs/r/r103.pdf)

Gamut Errors need detection if the RGB components and the corresponding Luminance (Y) signal exceed
the "Preferred Minimum/Maximum" range of digital sample levels

"Preferred Minimum/Maximum" range is [5, 246]

But any black pixels decoded with OpenCV has (0,0,0) values

Please help me. Maybe I'm misinterpreting the standard?

[Valentin Nikin]

Simple code example

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

using namespace cv;
using namespace std;

int main(int, char**) {
    Mat frame;
    VideoCapture cap("./video/Test_zapis_34.mp4");
    
    if (!cap.isOpened()) {
        cerr << "ERROR! Unable to open camera\n";
        return -1;
    }

    int min = 5;
    int max = 246;
    int frameCounter = 0;

    std::cout << "Start grabbing" << std::endl << "Press any key to terminate" << std::endl;
    for (;;) {
        cap.read(frame);
        if (frame.empty()) {
            cerr << "ERROR! blank frame grabbed\n";
            break;
        }

        int countPixelsOutsideGamut = 0;

        for (int i = 0; i < frame.rows; i++) {
            for (int j = 0; j < frame.cols; j++) {
                auto cmp = frame.at<Vec3b>(i, j);

                int r = cmp[0];
                int g = cmp[1];
                int b = cmp[2];

                std::cout << r << " " << g << " " << b << std::endl;

                if (r < min || r > max || g < min || g > max || b < min || b > max) {
                    countPixelsOutsideGamut++;
                }
            }
        }

        double percent = static_cast<double>(countPixelsOutsideGamut) / static_cast<double>(frame.rows * frame.cols) * 100;

        std::cout << frameCounter << " - " << percent << std::endl;

        imshow("Live", frame);

        frameCounter++;

        if (waitKey(5) >= 0) {
            break;
        }
    }
    return 0;
}

[Sharc]

There exist different conversion matrices for converting between full and limited (TV) ranges of various color spaces.
See for example here:
https://software.intel.com/content/w...ml?language=en
https://web.archive.org/web/20120403...onversion.html
https://forum.doom9.org/showthread.p...62#post1936762

The EBU document just defines its "Preferred Range" which includes a typical (or practical) headroom/footroom to allow some overshoot/undershoot of the limited (TV) range. Something like halfway beteen 'limited' and 'full'.

[Valentin Nikin]

Sharc, thank you for reply. I implemented gamut detection algorithm using FFmpeg API and analyzing raw data from AVFrame.
But now I need to check corectness of my algorithm. Are there any free tools (or with trial version) to detection Gamut problems in video file?