Video Survillence using Multifeature Background Subtraction Algorithm: A Self Adaptive Security Mechanism
Neethu Kunjappan1, K.Lakshmanan2
1Neethu Kunjappan, Department of Computer Science and Engineering, Annai Mathammal Sheela Engineering College, Namakkal (Tamil Nadu), India.
2Prof. K Lakshmanan, Department of Computer Science and Engineering, Annai Mathammal Sheela Engineering College, Namakkal (Tamil Nadu), India.
Manuscript received on 15 April 2013 | Revised Manuscript received on 22 April 2013 | Manuscript Published on 30 April 2013 | PP: 75-78 | Volume-2 Issue-5, April 2013 | Retrieval Number: E0668032413/13©BEIESP
Open Access | Editorial and Publishing Policies | Cite | Mendeley | Indexing and Abstracting
© The Authors. Blue Eyes Intelligence Engineering and Sciences Publication (BEIESP). This is an open access article under the CC-BY-NC-ND license (http://creativecommons.org/licenses/by-nc-nd/4.0/)
Abstract: This is a security system based on background subtraction algorithm. Currently existing surveillance systems normally use Closed Circuit TVs. Background modeling and subtraction is a natural technique for object detection in videos captured by a static cameras. The proposed paper uses multi feature background subtraction technique. Here it uses a pixel wise background modeling and subtraction using multiple features. Here generative and discriminative techniques are combined for classification. In this algorithm, gradient, color, and Haar-like features are closely integrated so that they can handle variations in space and time for each and every pixel. A e background model that is pixel wise generative is obtained for each feature by Kernel Density Approximation (KDA). Background subtraction is performed using a Support Vector Machine (SVM). The proposed algorithm is resistant to shadow, illumination changes in light and spatial variations of background. It monitors an already captured environment and if an intruder comes, then it will send message alert to the administrator and it will send current streaming video to the admin system. All these actions are performed so fast that it will be easy to catch the intruder and needs no human interaction which makes the system efficient.
Keywords: Background Subtraction Algorithm, Kernel Density Approximation ,Support Vector Machine, Haar-Like Features.
Scope of the Article: Big Data Security