Optimization Of Resource Allocation And Object Handover Algorithms For Smart Visual Surveillance Networks

With the rapid development of the Internet of Things and the in-creased demand of social security needs,the Smart Visual Surveillance Networks(SVSN)based on Camera Sensor Networks(CSN)has drawnmuch attention.Not only is SVSN able to sense visual information andsound informantion,but also it owns ability of information processing,inormation transmission wirelessly.It is because SVSN has many advantages,research aimed at SVSN owns vital theoretical significance and practical value.This dissertation mainly focuses on the resource allocation under limited resource and object handover of SVSN.The main contribution of this dissertation is summarized below.1)The crucial area coverage of Smart Visual Surveillance Networks with limited resource.Full coverage of an area is hardly achieved due to the vast resource expenditure.Covering curcial part of an area is a more feasible way to realize.A weighting function is proposed to describe the target area.We utilize Partical Swarm Optimization to adjust locations and orientations of sensors which enhance the coverage effect.Furthermor,we apply the Simulatied Annealling algorithm to help our approach to the global optimization.Simulation results show that our algorithms can clearly guide snesors cover crucial part of the arget area.2)The method of object handover in Smart Visual Surveillance Networks with non-overlapping coverage area.In realtiy,dut to the limited resource,sensors hardly form overlapping coverage area.In this case,this thesis proposed a methed to accomplish object handover with the approach to merge multi-feature of object.Color features and ORB feature are extracted and similarity of different features are calcutated separately.We take advantage of D-S evidence reasoning to merge every similarity and utilize the outcome to accomplish object handover.Experimental results show that our method is effective and reliable.3)The installation of Smart Visual Surveillance Networks and the implementation o f the Indoor Surveillance System.We have applied Raspberry Pi to construct the SVSN platform.With this platform,we implement an Indoor Surveillance System.This system can detect object in real-time and warn client with e-mail.Client can view the surveillance video on their broswers because of the Web server runned on the Raspberry Pi.The web page is designed in responsive layout,so that different broswers can bring same user experience to client.