A Real Time Surveillance System Using Wired and Wireless Sensor Networks by Multi-algorithmic Approach

摘要

In this paper, real time surveillance system is presented for an efficient identification of a person as an intruder or not in the inhibited location by multi-algorithmic approach. This system is aided with both wired and wireless sensor network. The Wireless Sensor Network detects the presence of person using PIR sensor and identifies the person with RFID and the verification is assured by using the image sensor associated with wired network. The data are gathered at server side and synchronised using time and location information. The image obtained is scanned for human faces and from the detected faces the features are extracted with the help of multiple algorithms (BICA, DCT&FLD, Kalman face) and the results are combined with weighted average based fusion strategy. The extracted features are then compared with the available set of features in database for authentication. The image comparison is done semantically by intra and inter class search in the database using the RFID value. The proposed system identifies a person as an intruder, if there is any mismatch in the sensed information. This system also provides solution to tail-gating problem. The system is implemented in JAVA with the OpenCV functions using JNA.