Implementation Of AVS2 Scene Coding Foreground Background Separation Transmission And Background Modeling Interpolation Simplification Algorithm

As a Chinese video coding standard,AVS2 has made great progress compared with the previous generation video coding standard,especially in the field of scene coding,which makes its coding efficiency at the international leading level.At the same time,for the huge surveillance video market demand,the current video coding efficiency still needs to be further improved to achieve the purpose of reducing storage consumption,so the research of surveillance video coding is still of great significance.This paper mainly studies the coding technology for surveillance video in the AVS2 standard,including a simplified algorithm for foreground background separation and its codec implementation and background modeling.The specific work and results of this paper are as follows:1.Introduced the basic video coding framework and related technologies.The basic ideas and coding methods are common to all major video coding frameworks.The technical characteristics of the AVS2 video coding framework are introduced in detail.The scheme of the scene coding technology and the surveillance video coding technology used by AVS2 are described in detail,including the background model establishment process and the scene coding algorithm.The front background separation transmission algorithm based on AVS2 framework is implemented.(1)The foreground background separation algorithm is implemented.The detailed process of background model coding and decoding is analyzed based on RD14.0 reference code.A feasible pre-background separation transmission solution is given.Under this scheme,a large number of experiments on typical video are passed.The granularity of the foreground background separation,the selection of the control frame,and the algorithm partitioning parameters are established.Finally,the pre-background separation algorithm is obtained and the rationality of the algorithm is verified by the result analysis.(2)The background block coding algorithm is implemented.The algorithm sets the PSkipDirect mode as the background macroblock flag for the background macroblocks divided by the previous algorithm,and uses the non-PSkipDirect mode for encoding the 8*8size foreground blocks.For other foreground macroblocks.Use the original coding method.The coding efficiency is improved by 35% to 50% when the number of coded bits per frame is increased by about 1% to 10%.(3)The pre-background synthesis algorithm is implemented.By determining whether the current block is an 8*8 size background block for the decoded macroblock type,and thendirectly assigning the decoded background image to the background block,the image is successfully decoded and the video precision loss is not obvious in a typical scene.(4)The video accuracy loss of the overall algorithm in this particular scenario is analyzed.By dividing the foreground background of the macroblocks under all sizes and then performing the encoding and decoding process,the image quality of the original video is compared and the loss of the video image generated by the algorithm directly using the background macroblock to replace the image is analyzed..Aiming at the background model establishment process in AVS2,an interpolation-based simplification algorithm is proposed.By reducing the frequency of the moving average in AVS2,the interpolation method is used to complement the pixel value of this process in the low-frequency moving average process.The background image quality established by the simplified algorithm and the original algorithm under the same frame number is compared and analyzed.Under the condition that the background image quality is almost unchanged,the background modeling time is reduced by about 0.3 seconds to 2 seconds.