Summary

MoveDetect uses the standard image processing library OpenCV. You'll need to link against OpenCV to use MoveDetect.

To detect movement, you instantiate a MoveDetect::Handler object, and then repeatedly call MoveDetect::Handler::detect() when you have frames to be analyzed.

The detect() method will compare the last few frames and return either true or false to indicate whether movement was detected.

Your code would look similar to this:

#include <MoveDetect.hpp>
 
void process_video()
{
    MoveDetect::Handler handler;
    cv::VideoCapture video("movie.m4v");
    while (true)
    {
        cv::Mat frame;
        video >> frame;
        if (frame.emtpy())
        {
            break;
        }
 
        const bool movement_detected = handler.detect(frame);
        if (movement_detected)
        {
            do_something_movement_detected(frame);
        }
        else
        {
            do_something_no_movement(frame);
        }
    }
}

There are many options, including thresholds for the amount of movement required, as well as generating a mask, bounding boxes, and contours.

Several key options to look into:

Configuration Item	Image
MoveDetect::Handler::psnr_threshold
MoveDetect::Handler::key_frame_frequency
MoveDetect::Handler::number_of_control_frames
original video frame
MoveDetect::Handler::mask_enabled MoveDetect::Handler::mask
MoveDetect::Handler::bbox_enabled MoveDetect::Handler::output
MoveDetect::Handler::contours_enabled MoveDetect::Handler::output