Moving Detection And Localization In Video Surveillance Systems

Abstract

We propose complex algorithm for moving objects detection and localization which represents the consecutive analysis of changes in image contours and optic flow field configuration within observed area. The use of contour analogue of initial image permits to improve quality of motion detection and decrease computational cost. Optic flow calculation enables more accurate and steady moving objects localization. Preliminary experiments show that the developed approach may be applied to real-time video surveillance systems and to other applications that require preparatory localization of moving objects.