The Research And Implementation On Target Location In Video Sensor Network

In recent years, the demand for fine-grained and precise environment surveillance increases, which leads to the advent of the research on video sensor network. As a part of this research, study on target location technology in video sensor network is prerequisite.This thesis firstly studies the principles of computer vision based target location technology and analyzes the key points of the technology; then according to the characteristic of video sensor network and experimental scene, this thesis proposes several methods based on previous work. These methods estimate the feature points to reduce sensor node computing load and network data quantity, aggregate multi-node data to improve the accuracy of locating and utilize geometry feature to simplify the camera parameter calibration and node management. In the light of these methods, a prototype system of C-Location is designed and implemented. Based on this system, several experiments are performed. The experimental data shows the result of target location and the effectiveness of applying computer vision based target locating into video sensor network.