Research Of Assessment Method On No-reference Network Video Quality Based Channel Impairment

With the rapid development of network and communication technologies,video has received increasingly attention due to its advantages of being i ntuitive,vivid,accurate,efficient,and rich in information.The volume of data for video services sent over wired and wireless networks has grown exponentially.While the network is under heavy load,it is necessary to ensure the quality of network vide o services so that the ultimate receiver of the video can maintain the perceived quality of the video within an acceptable range.However,in the process of encoding and transmission,the video suffers from encoding distortion caused by quantization and tr ansmission distortion caused by network impairments such as packet loss,delay and jitter,respectively,which affects the user experience.How to evaluate video quality accurately and effectively has become an urgent problem in the field of video compression,processing,and transmission.A method of no-reference network video quality assessment is proposed by this study,aiming to improve the real-time and effectiveness of objective assessment of video quality.Firstly,from the perspective of network Quality of Service(Qo S),a no-reference video quality assessment method suitable for network video is proposed,which considers the characteristics of video content.This evaluation method not only considers the distortion caused by the quantization process during encoding to the original video,but also fully considers the impact of video content and network channel transmission conditions on the final perceived quality of the video.Then,from the perspective of video Quality of Experience(Qo E),a no-reference video quality assessment method considering frame freeze events is proposed.This method fully considers the impact of single frame freeze events and multiple frame freeze events on the subjective perception experience of the video.And the two method s have low computational complexity and can make self-adaptive evaluation of videos with different contents.The research content of this study mainly includes the following aspects:(1)Study the distortion of the video when it is compressed by the H.264 encoding standard.Since the H.264 coding standard is lossy compression,for the H.264 coding standard,this study researches its quantization,motion estimation,intra-frame prediction and inter-frame prediction and other encode methods,as well as the influence of video parameters such as frame rate and coding bit rate on the video quality,which provides the basis for the subsequent objective representation of the video content characteristics.(2)Study video transmission quality distortion caused by th e network channel when the video passes through the network channel.Since the network for transmitting video data uses the best-effort UDP(User Datagram Protocol)transmission protocol,so the quality of the transmitted video cannot be guaranteed.In this study,the frame loss rate is used to characterize the condition of the channel,the effect of different frame loss rate on the video quality is also studied,and carefully analyzes the impact of different video types on video quality at the same frame l oss rate.Besides,under the same frame loss rate with difference in frame loss concentration,this study also makes a detailed analysis of the impact of the final video experience.(3)The objective representation of the characteristics of video content a nd its effect on the subjective experience of video are studied.Due to the selective focus of the human eye on the target object,the characteristics of the video area that the human eye focuses on are deeply studied,and objectively describe the content characteristics of the video with the as low as possible complexity.We find that the motion vector in the video P frame and the number of intra-coded macroblocks in the intra-frame prediction method adopted by P frame are the most representative of the co ntent characteristics of the video.Moreover,the influence of different video content under the same encoding bit rate on the video experience is also studied.At the same time,the influence of different video content on the subjective perception of the video under the same channel transmission conditions is also carefully analyzed.(4)The relationship between the frame freeze that occurs during video playback and the subjective perception experience of the final video is studied.Because the fluency of video is the main factor affecting the subjective experience of video,the frame freeze event is exactly the most important reason that affects video fluency.Therefore,the relationship between the single frame freeze event and the subjective experience of the video is deeply studied,and the connection between the multiple frame freeze events and the video experience during video playback is carefully analyzed.At the same time,the relationship between the content complexity of the video frame and the frame freeze duration and the subjective perceived quality of the video is analyzed in detail.In the experiment,the performance of the video evaluation model constructed in this study is compared with the international standard G.1070 model.The results sh ow that the evaluation model proposed in this study has better evaluation performance.