UDC 004.932
AUTOMATIC EXTRACTION AND DETECTION OF MOVING OBJECTS USING VIDEOCAMERA WITH VARIABLE FIELD OF VIEW
P. V. Babayan, Ph.D. (Tech.), associated professor, Department of automation and information technologies in control, Head of the Department, RSREU, Ryazan, Russia;
orcid.org/0000-0003-4041-3101, e-mail: This email address is being protected from spambots. You need JavaScript enabled to view it.
E. S. Kozhina, Master's student, Department of automation and information technologies in control, RSREU, Ryazan, Russia;
orcid.org/0000-0003-4498-2694, e-mail: This email address is being protected from spambots. You need JavaScript enabled to view it.
The paper considers the problems of extraction and moving objects detection in a sequence of images. To solve the described tasks, a method based on background subtraction is used. The aim of the work is to develop an algorithm that is oriented to work in the presence of scale geometric transformations on an image. To evaluate scale geometric transformations, a multi-template matching approach is used. In the process of developing object detection and extraction algorithm an updated mathematical model of an image has been used. This model takes into account geometric transformations caused by changes in the focal length of a camera or changes in the distance to a scene due to spatial movement of image sensor. The results of experimental studies with various real video sequences are presented. During the research of the algorithm, the influence of scale distortion estimation standard deviation on the quality of object localization has been considered. Based on the results of the experiments, a conclusion about relatively high efficiency of the proposed algorithm has been made. The algorithm developed can be used in computer vision systems of aerial vehicles, mobile robots, in video surveillance systems.
Key words: object extraction, object detection, background subtraction, scale estimation, evaluation of geometric transformations, multi-template matching, vision system