Research On Video Coding Based On Background Modeling In Surveillance Scene

With the progress of technology and the improvement of people's living stadnards,surveillance cameras used for security can be seen everywhere in the streets and alleys,which also brings about the storage problem of surveillance video.Although the current video coding standards can compress digital signals,the mainstream coding standard only compresses video in the surveillance scene as ordinary video in a general way.In this context,the thesis uses the characteristics of the videos in surveillance scene,combined with the reference structure of video coding,considers that the background in the surveillance video can bring a certain degree of improvement to the performance of video coding.This thesis will use a Hisilicon chip named Hi3559AV100 as the experimental carrier,propose a set of video coding scheme for surveillance scenes on the upper layer of the H.265 standard,and put the scheme into the Hi Silicon chip.Firstly,this dissertation uses the knowledge of background modeling to generate the background frame that needs to be referenced in the encoding,and compares several common low-complexity background modeling methods,then analyzes their advantages and disadvantages,and makes a certain degree of improvement on the basis to obtain a background model of higher reference value.Secondly,in order to carry out video encoding experiments on the Hisilicon chip,this dissertation studies the use of the chip,builds the system required for the experiment,and uses the generated background image in the process of video coding.Finally,this dissertation analyzes the content in the surveillance video and believes that the moving foreground objects are the focus of attention.In order to improve the quality of the regions of interest and reduce the bit rate of other regions,this dissertation discusses the foreground extraction scheme and proposes a foreground extraction based on block-level structured,and combines with a block-level rate control method to further improve the performance of video coding.