| With the rapid development of computer science and technology and more attention people pay to public security,network video monitoring system has become an important part of city security,so that large-scale video fusion and monitoring system based on distributed network architecture have gradually become a hot research topic in the industry.In this paper,a distributed network computing node suitable for Panoramic video processing center is designed to solve the problem of large scale multi camera panoramic video surveillance.As the core distributed processing unit of the system,the nodes will complete the distributed calibration and registration of the network camera,video stream processing and computer vision algorithm processing.In response to the above requirements,we have carried out the relevant research and development work.The specific research results are as follows:We realized the distributed network computing nodes remote calibration and calibration method of camera,so that we can get spatial position and geometric transformation relation of the network camera in a panoramic image,in the meantime we solve the image mismatch problem caused by inconformity between registration of the image zooming scale and resolution.The algorithm is composed of two steps: rough calibration and registration,fine calibration and registration.Combining with the two calibration parameters with the registration process,the distributed network computing nodes will calculate the coordinate corresponding relation from panorama network cameras to reproduction and upload data remote to the database management system.We designed the network video segmentation node in panoramic video computing successfully.According to functional requirement of the node parameters of remote configuration,video segmentation,video data compression encoding and streaming video release,we designed a network video segmentation node operation framework,and discuss the technical details and operation mechanism related to each module.Firstly,the video data format provided by the network camera is analyzed,and the parameters of the video frames are specified,so that a network video segmentation algorithm is analyzed.Secondly,based on FFmpeg open source framework,this paper implements the H.264 video coding module which is suitable for distributed network computing nodes.Finally,based on the general and mature streaming media transmission protocol,a released video module based on real-time messaging protocol was implemented.The network video fusion display node in panoramic video computing is designed and implemented.According to the functional requirements of remote configuration,remote video streaming,compressed video data analysis,panoramic video fusion and display,we studied the design scheme and technical details of each module.On this basis,this paper proposes a node event synchronization mechanism based on NTP network time protocol.Aiming at the problems in the execution efficiency of node computer vision algorithm,we put forward a prioritization scheme video stitching and the fusion algorithm based on OpenCL so as to complete part of the complex calculation using the graphics processor in a distributed network computing node replaced CPU and improve the computing performance of the node.In this paper,a set of distributed network computing node verification system was implemented based on the existing algorithms and software platform.This system includes video processing nodes supporting supplying base services,transit server storing and dispatching various data,calibration and registration software of panoramic video computing,database management software and workstation topology calculation supplied by other group members.The system verifies the functional modules of distributed network computing nodes effectively,and finally displays the results of each node in the panoramic monitoring screen array.At the end of this paper,we analyzed the shortcomings of the current work,and prospected the future research directions. |
PDF Full Text Request