Study Of Efficient Video Decoding And Display System Based On Attention Guidance

For video communications,its final recipient is a human observer.Many of the traditional video codecs did not take this into account.This thesis focuses on this issue,taking the relevant characteristics of the human visual system into account.This thesis designed an efficient video decode and display system based on attention guidance,the main work is shown as follows:(1)Firstly,the characteristics of the human visual system are studied systematically.The function of the human visual system is studied from the physiological structure of the human eye,and then the characteristics of the human visual system in sight and field of view,intensity and contrast are studied.Based on the above basic research,a structure of efficient video decode and display system based on attention guidance is designed.(2)In the realization of the design of efficient video decode and display system,two key technical points to is focused on the study,the first one is frame fusion of a number of data stream.By binarization,corrosion,Gaussian blur and then combined with a weighted formula to get the fusion image,can avoid the edge effect which is left when try simple splicing to multi-stream image.The second one is high-resolution reconstruction of the background area based on the redundancy of video sequence in the time dimension.The background of the current frame is reconstructed using some of the information in the previous frame,thereby improving the clarity of the area.Usually the method only considers the similarity between the background region and the high resolution region,and the reconstruction effect is poor.In this thesis,the objective function of genetic algorithm is optimized by using the penalty function method in the optimization theory,and the connection between the two high resolution regions is combined,which makes the result of reconstruction have a better matching effect with the surrounding image.(3)In order to quantify the difference of video quality between this thesis' method with the traditional method under the same bit rate,the peak SNR(PSNR)and the structural similarity(SSIM)of the video objective quality evaluation algorithm are studied,especially the widely used SSIM algorithm.In the process of research,SSIM algorithm is found to be poorly performed under fuzzy distortion,and the evaluation value is contrary to human eye view.In this thesis,an improved SSIM algorithm is proposed to combine the edge information of the image and the region of interest.The experimental results show that after nonlinear fitting,the relative error of the estimation of subjective evaluation is reduced by about 50% compared with the traditional SSIM algorithm.This indicates that the improved SSIM algorithm is closer to the subjective perception of the human eye.(4)Using the C++ programming language and ffmpeg open source library,this thesis completed an efficient video decode and display program to realize the video decode and display system.The experiment shows that the system achieves the design goal.Using the improved SSIM algorithm,the comparison between the proposed method and the traditional H.264 method is carried out.The results show that the objective quality of the video is about 15% higher than that of H.264,and the subjective quality is about 30% when the bit rate is limited.