Target Tracking, Recognition And Coverage-Oriented Localization In Camera Sensor Networks

Intelligent living environment is indispensable when stepping into theinformation age, whose concentration on the comfort and security, such asthe smart entrance guard and visual surveillance, accelerates the research ofcamera sensor networks. Due to the distributed architecture, wireless camerasensor networks can achieve the high-level recognition, including smarthome comfort system and the location-based fast criminal clues search. Forthe former, the smart entrance guard can greatly comfort the daily life. Forthe latter, the inefficiencies of visual monitoring due to fatigue can bealleviated, meeting the needs of large-scale video surveillance systems.Therefore, further research of target tracking, recognition andcoverage-oriented localization in camera sensor networks is of greattheoretical and practical significance in the video surveillance industry ofchina. The main work of the thesis covers the following four aspects.1. The construction of a context-aware smart entrance guard.A centroid based human motion recognition is implemented alongwith the eigenface based face recognition, which are both used todetect an event with the time and place added. Thus, an event drivenapplication is constructed concerning distinctive reactions towards different events detected via MSP430F5438controller andNRF24L01wireless module.2. Complete area coverage in camera sensor networks. In the areacoverage model, point coverage is used and then the area pointcoverage optimization model is derived in the considered scenario.Finally, optimal parameters of the deployed cameras are obtainedthrough the heuristic algorithms.3. Construction of a target tracking evaluation platform.Consisting of an EP2C8Q208C8Altera FPGA controller,2-DOFMG995steering engine and the Logitech web camera, theconstructed platform can achieve the target tracking via the fusion ofthe continuous Additive Increase-Multiple Decrease (AIMD) basedrepaired foreground and the repaired foreground from the Cannyedge extraction of the subtraction image.4. A Simultaneous Localization and Coverage (SLAC) Strategy.Considering the scenario where people move around in the initialcompletely covered area, an optimization model is presented to makea compromise between the overall switch time and remaining areacoverage rate in the selection of the localization camera. Quantitativesimulation data confirm the effectiveness of the SLAC strategy.