Pan-Tilt Kameralar İçin Arka Plan Modellemenin Donanımda Gerçekleştirilmesi

Abstract

Background modeling is commonly used in image processing applications. Stationary camera assumption, the basic assumption of background modeling, becomes invalid in some cases such as movement of camera, movement of platform on which the camera is mounted, oscillation of camera due to wind etc. The purpose of this study is to keep providing the background information while the view point of the camera changing. Proposed method models the background with a Gaussian distribution. Previous background model and current frame are registered with the motion information which allows utilizing the previous background model. The purposed method is implemented on real time operating hardware for 1080p@25 video and tested on different image sequences to present the performance.