Based On The Subjective Quality Of Low Bit Rate Video Coding Research

We has been paying more attention to video communications since the network wasborn. If the video doesn’t be compressed before communication, video communicationwill occupy a very large bandwidth, which will be a serious impediment to theapplication of video of communication. It’s an inevitable way to compress the massivevideo redundancy. ISO and ITU launched a series of standards through a lot of work likeseries of JPEG, the MPEG series and H.26X series which played a significant role to thepopularity of the video image.Existing video coding techniques (such as H.264/AVC) can not only achieve highercompression ratio but also get relatively good compression efficiency. But when it usedin application scenes with low bit rate such as mobile video communication, real-timedesktop video, video quality always descend because of the low requirements of bit rate,especially the face areas will appear blurs or cubes where the human eyes are moresensitive to.First, this paper introduces the requirements of the low bit rate video codingtechnology and the research of enhancing the subjective video quality methods in detail.The image enhancement method is the mainly method to enhance the subjective videoquality for now. the image enhancement method can improve image quality, rich amountof information through clear image clarity or to emphasize certain characteristics ofattention and inhibition of the characteristics of non-concern. Image enhancementtechnology allows images to improve the subjective quality to a large extent, but itsdrawback is that when the edge enhanced the noise will increased at the same time.What’s more, when you try to filter noise, the edge will be blurred to a certain extent. Themethod can hardly enhance the video communication. Moreover, it will worsen thequality of the video subjective in some cases.Second,based on H.264, this paper presents a region of interest-based adaptiveimage enhancement method to solve the above problems. To start with, we will dividevideo image into two regions: interest and not interest. Then we access to the video frame image of the interest region. Next, we use ASAD algorithm to obtain the gray-scale andget mask map of the region on the basic of gray-scale. We get adaptive enhancementfactor according to the mask, meanwhile, we will use Gaussian high-pass filtering to theregion of interest. As the region of not interest, we will use a Gaussian low-pass filtering.Finally, boundary drop processing will be taken on the entire frame image. The proposedmethod can process video adaptively with high target in low bit rate case so as toachieve the purpose of enhancing subjective video quality.