Hybrid object detection using improved three frame differencing and background subtraction

Abstract

Object Detection and Tracking in video has applied in robotics, video-surveillance; human-computer interaction etc. and different approach of object detection e.g. Background subtraction, frame differencing. Motion based recognition is one of the methods to detect objects in sequence of image. In this method, a video sequence containing a large number of images is used to extract motion information. Two frame differencing is very easy but there is problem of holes. Three frame differencing and background subtraction have solved the problem of holes of two frames till a limit. Background subtraction is used for stable background video but Dynamic Background subtraction is capable to detect object in video with gradual background changes. So there is scope of work such that holes problem should be reduced more and object should be detected better in dynamic changes in background. In this paper, the proposed technique is able to reduce the holes problem in dynamic background updating video.