Research Of Surveillance Video Background-model Technology For Transmission Bandwidth Saving

In recent years,with the rapid development of economy and the expansion of city scale,the social security situation has become more and more complicated.In order to solve the problem of urban security,more than 300 cities in China are carrying out the construction of smart cities.As the basic supporting part of smart city,video monitoring market has been rapidly expanded in recent years,and a large number of video monitoring data enter the network to bring huge transmission bandwidth pressure.The existing method of alleviating bandwidth pressure is mainly the video compression coding technology represented by MPEGX/ H.26 X,but the existing compression coding technology still has a large number of secondary information which occupies a large amount of bandwidth.In order to meet the increasing demand of video monitoring data transmission,the paper [11] proposed a bandwidth-saving video transmission architecture.Under the condition of transmitting the same quality video,bandwidth-saving video transmission architecture saves about 50% of the bandwidth efficiency compared with the h.264 transmission mechanism.Based on this framework,this paper studies the background model technology that include the background extraction,storage and transmission.Firstly,the existing common background modeling methods are studied and simulated.Based on the problems of “ghosting” phenomenon and low update efficiency in common algorithms,an adaptive background extraction and update algorithm is proposed.The algorithm is based on the method of frame difference and segmentation,and it adds new image state value parameters,then uses a transitional background update strategy to eliminate the ghost phenomenon.At the same time,it recognizes the scene based on the similarity detection results of the newly generated background and historical background.Finally,the update rate threshold is adaptively adjusted according to the scene category.Experimental results show that the proposed algorithm can effectively eliminate "ghosting" phenomenon,the background extraction rate is faster,and has a good performance in real-time processing.Secondly,there is less research on the background model management field.Combined with the idea of the design of bandwidth-saving video transmission architecture and the requirements of the remote video surveillance system,an adaptive background-model library management mechanism based on the value scale(A-BMVS)is proposed.It mainly includes four modules: add,delete,check,and change.The mechanism introduces new value parameter and values scale concepts based on actual background changes.Then,the background classification and storage based on the value scale are carried out.Finally,the background of the storage rules is used to manage and transfer information.The experimental results show that the background synchronization of this mechanism is highly accurate,and the synchronization delay meets the requirement of bandwidth saving video surveillance,and this mechanism can reduce the transmission flow rate by more than 70% compared with H.264.