Video Analytics for Abandoned Object Detection and its evaluation on Atom and ARM processor

The goal of proposed method is to enhance the safety and security by identifying the abandoned objects in the environment under consideration. This paper mainly exploits some of the properties of image processing and embedded system to implement the Video Analytics based robust and simple Security System for the surveillance of environment for twenty four hours a day. Proposed method supports a human operator by automatically detecting abandoned objects and drawing operator's attention to such events. This method is based on the various methods of the image processing and pattern recognition such as Gaussian Mixture model, Absolute background subtraction, image segmentation, connected component analysis and Histogram of oriented gradient. The algorithm evaluation is done on the hardware platform viz. Friendly Arm and INTEL'S IVI board. Here we find results which explain whether the accuracy and reliability can be achieved at the cost of processing power. This is done to reduce the cost of system. Abandoned object can be decided by time basis. Proposed algorithm fits for the PETS2006 dataset and works real time. We have used OpenCV as software platform.