Testing the Efficiency of Background Subtraction Techniques

Abstract

In many video surveillance applications, for instance, detection of abandoned/stolen items or parked vehicles, the detection of desk-sure foreground items is an important task. In this work, we recommend the implementation of a green item detection algorithm that may be employed in actual time embedded devices because of its fast processing. The morphological approach is used for similarly processing to remove noise and to maintain the form of transferring object. First video is converted into streams and the applied filter which remove high frequency noise components to obtain smoothened subtraction algorithm with an adaptive threshold that gives detected object is then implemented to a convolution clear out to eliminate the distorted pixels which enhance the quality of an image.