Motion Detection Using Adaptive Temporal Averaging Method

Abstract

Motion detection methods are widely integrated in modern intelligent video surveillance systems. Many of these methods use background subtraction techniques to separate the foreground objects from the background. Temporal averaging is one of the most commonly used and simple method for background subtraction. In this paper we propose new version of the original Temporal averaging algorithm. The speed of updating the background model has been modified to be adaptive and determined by pixel difference. Another approach with simultaneously adaptive threshold and background update speed is also proposed. Our goal is increasing the F-measure of the method by making the algorithm more versatile for different scene scenarios. Experimental results are shown and analyzed. The quality parameters of the original method and the proposed method are compared.