The paper describes a new approach to background/foreground estimation in images from a video stream. The proposed method is based on the analysis of small gradients in images. The video stream is recorded with the camera mounted in the measuring station above the road and directed along the road. Input images are converted into binary form and divided into square blocks. For each block the sum of binary values is calculated. Blocks are assigned to the background or to the foreground of an input image. Background/foreground assignment depends on sums of binary values within individual blocks. Vehicle detection is carried out through the analysis of blocks assigned to the foreground. The proposed method of background/foreground estimation is intended for systems of traffic measurements and surveillance.
Video-Based Background/Foreground Estimation for Vehicle Detection
Lect. Notes in Networks, Syst.
Scientific And Technical Conference Transport Systems Theory And Practice ; 2021 ; Katowice, Poland September 21, 2021 - September 23, 2021
2021-11-03
10 pages
Article/Chapter (Book)
Electronic Resource
English
Smooth Foreground-Background Segmentation for Video Processing
British Library Conference Proceedings | 2006
|Two video analysis applications using foreground/background segmentation
British Library Conference Proceedings | 2003
|Video Repairing: Inference of Foreground and Background under Severe Occlusion
British Library Conference Proceedings | 2004
|