ALGORITHM FOR CALCULATING THE POSITION AND SPEED OF OBJECTS ON A PLANE USING DATA FROM A MONOCULAR CAMERA

This article presents an algorithm for the automatic calculation of a homography matrix, intended for estimating the speed of forklifts in warehouse conditions. The proposed algorithm uses video streams from surveillance cameras to transform object coordinates from a perspective image into a metric space, enabling the determination of forklift positions in real measurement units. The algorithm implements automatic calibration based on the recognition of pallets with known dimensions and the use of camera characteristics. An automatic parameter selection algorithm is proposed for the monocular system in cases where these parameters are not known in advance. An experiment using real video recordings demonstrates the applicability of the proposed algorithm for monitoring logistics processes and improving the efficiency of warehouse transport management. The experiment showed that when calculating average forklift speed, 98% of the obtained values fall within an error margin of ±1 km/h.