A Single Tier Homogeneous Visual Sensor Network for Real Time face Recognition

• Published in: 2017 IEEE International Conference on Computational Intelligence and Computing Research (ICCIC) pp. 1 - 6
• Main Authors: Sriharsha, K V, Alphonse, P.J.A
• Format: Conference Proceeding
• Language: English
• Published: IEEE 01.12.2017
• Subjects: Cameras; Face; Face detection; Face recognition; Feature extraction; image resizing; on board processing; similarity prediction; visual sensor node; Visualization; ZigBee
• ISSN: 2473-943X
• DOI: 10.1109/ICCIC.2017.8524274

In centralized based IP surveillance architectures, the information pertaining to the event captured in each camera is sent to the master data base that is located in central control server. Later for crime or theft detection in surveillance region, the recorded data is accessed and analyzed. As on today, with ongoing acts of vandalism, it is highly essential for surveillance to be reactive rather than proactive. The objective is fulfilled by incorporating on board processing features such as face detection and recognition with in the sensor node(s).Soon person enters the vicinity of camera, features are captured, processed and then recognized with in the node.In order to demonstrate this on board processing capability, we came up with a single Tier homogeneous sensor network design featuring the real time face recognition without using database(s) and also optimal transmission in the network using image dimensionality reduction technique. In order to assess the similarity in face recognition statistical measures like mean square error and correlation coefficient are taken for consideration and the system is tested and verified with a confidence level of 98 % in a resource constrained environment.