A Mixture of Distributions Background Model for Traffic Video Surveillance

Abstract

Video surveillance gathers more and more ground in the area of traffic measurement. Such computer vision applications use a static camera and some kinds of a detection algorithm for identifying moving objects. Background estimation plays an important part in these applications playing the part of generating the ideal background of the video stream, enabling the application to easily subtract the recent frame, and separate the foreground pixels from the background. The paper takes a short survey on the recently applied background estimation models, and introduces a distribution mixture model.