Research On The Optimization Problem Of Video Surveillance System

Video surveillance system has the functions of prevention,supervision and evidence collection,which plays a key role in ensuring social security and maintaining social stability.In recent years,with the in-depth development of projects such as "Safe City" and "Bright Project",cities across the country have increased their investment in monitoring equipment.With the increase of the number of monitoring equipment,in the space construction layout of the monitoring system,the staff generally choose the installation location according to the work experience or simple estimation,and there are problems such as unreasonable distribution,repeated construction,low utilization rate,which will affect the construction and development of the monitoring system.Therefore,it is necessary to study the location selection and regional coverage of monitoring equipment in order to improve the quality of monitoring coverage and the efficiency of resource use.To this end,this paper studies the monitoring point planning and scene coverage optimization in the monitoring system,and the research content is as follows:Firstly,for the monitoring point layout of urban road networks,this paper uses vector data to express urban traffic roads,uses the network topology to abstract roads into a graph,takes the road intersections as the vertex of the graph,and the roads between the intersections as the edges of the graph;Then,based on the road network condition,traffic condition,population number and key units and other influencing factors,the perception probability model is established,and the quantitative index reflecting the advantages and disadvantages of the perception point layout is constructed.Finally,a permutation iteration algorithm based on the contribution of intersection points is designed to solve the model,so as to maximize the probability of the travel object being perceived.The algorithm is applied to a certain urban road network,which can effectively avoid low efficiency or resource waste caused by unreasonable point arrangement.Secondly,an improved artificial bee colony optimization algorithm based on area coverage is proposed for the two-dimensional camera perception model.In this algorithm,the global optimal nectar source is introduced in the population renewal stage,and the leading bees and the following bees search nearby,and the reverse learning idea is used to generate a new nectar source to replace the worst nectar source after the iteration.Experiments show that the algorithm can effectively improve the coverage of monitoring area and its performance is better than the traditional method.Thirdly,a hybrid algorithm based on particle swarm optimization and gradient algorithm is proposed for the 3D camera perception model.In this algorithm,a new gradient iteration formula is constructed,and a partial increment calculation method is given to reduce the time complexity.The improved gradient algorithm is embedded into the iteration of particle swarm optimization algorithm,and the global search of particle swarm optimization algorithm and the local search of gradient algorithm are used to accelerate the convergence of the algorithm.Experiments show that the algorithm is superior to other algorithms in improving the coverage rate of the monitoring network and the convergence of the algorithm.