Research On Video Surveillance Technology Based On Multi-view Synthesis Algorithm

With the development of science and technology,video communication has become one of the main ways of data communication.Nowadays,the society has put forward higher and higher requirements on the function and quality of videos.It is desirable to replace the single view images with full view images.However,due to the limitations of the scene,cameras can not be placed at some views.In order to get images for these views,it is necessary to synthesize the adjacent two view video images.Based on this requirement,multi-view video coding(MVC)technology is proposed.This technology laid a good theoretical foundation for the development of modern television,such as 3DTV,FTV.In the future,this technology will also be widely used in many fields.Multi-view synthesis technology is the most important part of MVC,but there are still many problems to be solved: the accuracy of the parallax graph is not high;the initial synthesis graph has a lot of holes;the choice of empty-trap-algorithm greatly affect the quality of the composite graph;the algorithm takes too long and is not suitable for real-time systems.Therefore,there are two key points in the research of MVC: the choice of the disparity estimation algorithm;the choice of the algorithm to fill the hole efficiently and accurately according to the existing image information.In this paper,two modes of view synthesis--parallel camera array and ring camera array--are studied respectively.For the parallel camera array,an adaptive disparity estimation mechanism is proposed,and an improved view synthesis algorithm is also proposed.Only one parallax estimation process is needed to obtain the synthetic graph,which reduces the time-consumption of the algorithm.For the ring camera array,one set of view synthesis system is proposed.The 8-point algorithm is used to correct the polar line,and the depth information is used for 3D mapping,through the background image to finish hole filling.Finally,the two-mode view synthesis algorithm is implemented in the software platform,and one sub-model indoor video monitoring software of virtual view synthesis is completed.The simulation results show that,the algorithm proposed in this paper is time-efficient and effective.