Background Subtraction Techniques

Abstract

Background subtraction is one of the important and useful step for moving object detection, especially in the domain of video surveillance. There are various methods have been developed over the recent years. This paper gives survey of the recent approaches which concern to statistical background modeling techniques. These background subtraction techniques have benefits and limitations in terms of noise, illumination change. To overcome this problem, this paper provides a review of Background subtraction methods and comparison mainly based on three factors speed, memory requirements and accuracy.
Key Words: Background modeling, Kernel Density Estimation, Mixture of Gaussians, Single Gaussian