Research Of HEVC Optimization Algorithm Based On Region Of Interest

With the continuous improvement of requirements of video analysis,processing and transmission technology,video surveillance system improves from the original analog to the digital,intelligent direction,this results in a sharp increase of surveillance data,making the existing video coding standards cannot meet the video aplication development requirement.High efficiency video coding(H.265/HEVC)provides a number of new technologies,with their own continuous improvement will gradually replace H.264/AVC,and apply in all kinds of video coding business.HEVC makes the coding efficiency reach twice than H.264 by using these new technologies,but these new technologies also increase the computational complexity of coding.In recent years,how to use an effective method to reduce the computational complexity of HEVC has been the most important part of the research work of HEVC.In order to improve the coding efficiency,this thesis combines the intelligent video surveillance technology and HEVC standard.Starting from the key modules of HEVC coding framework,the technical principles of the key coding modules of HEVC standard are studied in depth.After a detialed study of CU partition and entropy coding principle,the improved algorithm of CU partition and entropy coding based on high concern region is proposed.First of all,extracting high concern region from the video sequence,also known as the region of interest that is the moving target in the video surveillance,is served as the preparatory work for the proposed algorithm.The region of interest is obtained by the depth information of Kinect and stereo video and the background information extracted by Gaussian mixture model from the video sequence combine with the depth information definition and background subtraction method;Secondly,the extracted region of interest is applied to the video codec.For the region of interest that has been acquired,an early termination strategy is proposed for the CU partition in HEVC,which reduces the unnecessary rate cost calculation and accelerates the coding efficiency.This method not only ensures the less loss of video coding quality,but also reduces the average coding time by 55.47%,significantly improves the coding efficiency;Finally,taken into account the effect of CU partition on entropy coding,another method of improving original rate-distortion algorithm is proposed based on region of interest.When the video image quality and coding complexity are relatively constant,the bit rate is reduced by 51.6%,which obviously improves the compression performance of original HEVC algorithm.