Research And Implementation Of Intelligent Video Surveillance System Based On GPU

Intelligent video surveillance technology has been widely used in all aspects of securityarea and social life, by mapping relation is established between the image and the descriptionof the event, the intelligent video surveillance technology can solve all kinds of shortcomingsand defects of the traditional video surveillance technology. With the continuous developmentof computer technology and hardware and the sheer amount of computing data, thecurrent reliance on intelligent video surveillance system based on the CPU processingcapacity has been unable to meet the application requirements of complicated algorithmand advanced features. Therefore, we propose a new intelligent video surveillancesystem based on the graphics processor unit (GPU).Aiming at the intelligent video surveillance system needs to consume a large amountof CPU resources in the video decoding module, YUV to RGB format module, imagebackground extraction module and the operation of the whole process of the algorithm,which makes the system stability and accuracy are not up to the requirements ofcustomers. We have improved the system resource consumption in the stream data process ofintelligent video analysis processing. Make full use of the parallel processing capabilities ofGPU. A heterogeneous computing lifting mode of CPU+GPU makes the whole system effectgreatly. The GPU work in the video stream decoding, YUV to RGB converse andthe optimization algorithms can guarantee the high processing ability of CPU, which can havemore resources available to the whole IVS system, so as to ensure the stability, accuracyand controllability of IVS system.Based on the analysis of intelligent video surveillance system,we use an embeddedproduct with GPU, transplanted the video stream decoding and format conversionmodule and algorithm of the CPU processing to the GPU processing in intelligent videosurveillance system, redesign and transplantation of the new IVS system, by comparingthe system performance and resource usage, which demonstrate that the new intelligentvideo surveillance system based on GPU has obvious advantages and very good application,which also provides a new direction for the intelligent video surveillance system research anddevelopment.