Abstract: The moving object detection deals with the detection of moving object in terms of moving person, vehicle,
animal etc. It identify and detects the moving object and avoids the background challenges such as motion of the
background, illumination variation etc. Such kind of work can be applicable in many useful for real-time computer
vision applications like indoor-outdoor visual surveillance security systems, robotics, driver assistance system, traffic
analysis, vehicle counting, navy, defence, army, target based object identification and surveillance of restricted zone. In
this research work, the proposed method detects the moving person and handles the problem of background motion and
illumination variation. It shows how a system can be developed by means of controlling the variance and standard
deviation based threshold value. The proposed automatic threshold is used to trade-off with the difference between
background and foreground pixel. It generates effective and appropriate threshold value for each pixel classification. It
also classifies the moving pixels more accurately and improves the detection quality. The proposed work shows a
significant improvement in the results of all considered video sequences and also compared with peer method i.e.
DECOLOR. It presents the strength by eliminating various environmental and illumination effects.
Keywords: Background Subtraction Algorithm, Real Time, MATLAB
| DOI: 10.17148/IJARCCE.2018.75556