Algorithm for tracking the movements of a person in the video stream from surveillance cameras

Abstract

The task of tracking the movement of objects in the video stream (tracing) is one of the most difficult tasks of modern video surveillance and video analytics systems. The tracing task is understood as the task of automated recognition and further tracking of objects throughout the sequence of frames of the video stream. The purpose of this article is not only a brief analysis of some of the existing methods and algorithms necessary for visualizing the movement of objects, but also the development of a new algorithm for tracing human movements in a video stream with the ability to display motion trajectories. The proposed approach makes it possible to eliminate some of the shortcomings of currently existing methods (presence of special equipment, the need for preliminary training, overlapping of the object with other objects, going beyond the camera’s field of view). The paper describes the main advantages of this algorithm, its most important functions and capabilities. The main stages of the algorithm operation include: selecting a person in the video stream, implementing the digital processing procedure, which is based on dividing the human body into its constituent parts and obtaining a color histogram of these parts, predicting the location and recognizing the person under study on all subsequent frames of the video stream, where the RGB pixel color analysis is performed using the method of comparing the obtained data with the color histogram of the constituent parts of the main object. The output data of the proposed algorithm is used in the procedure of forming and displaying a general picture of the movement of a particular object within the entire video stream. This article may be of interest to specialists and experts who use computer vision methods in their work to automatically obtain the necessary data when analyzing video fragments.