Optical Flow Motion Detection on Raspberry Pi

摘要

This paper presents the implementation of Optical Flow Motion Detection algorithm on Raspberry Pi. The Lucas-Kanade method was chosen for the implementation. The algorithm works by comparing two successive image frames. To find out a displaced object, the algorithm tries to guess the direction of displaced object rather than scanning the second image for the matching pixel. This can be done by solving for the optical flow vector by assuming that the vector will be similar to a small neighbourhood surrounding the pixel. The algorithm was simulated using Python OpenCV. The implementation of Lucas-Kanade algorithm was successfully done on Raspberry Pi.