| With the development of digital video technology,the needs of HD video and UHD video increase rapidly in recent years.Particularly,the use of intelligent mobile devices has become common,while video applications have been expanded more and more.In spite of the rapid development of the network and storage technology,the growth of the video data is more striking.Consequently,high efficiency video coding(H.265/HEVC)standard,which could save up to nearly 50% bit-stream at the same mean opinion score(MOS)compared to the previous standard H.264/AVC,would definitely become the future mainstream standard of the next ten years.Evaluating the video quality efficiently has been an important link in measuring quality of digital video system and improving user experience.The methods for video quality assessment could be classified into three categories: full reference method(FR),reduced method(RR),no reference method(NR).Particularly,NR method could realize automatic quality assessment without using original video for reference,which makes it a research hotspot gradually.For NR methods,there are plenty of studies for H.264/AVC,but relatively few studies that took a hard look at H.265/HEVC.In this paper,we propose an NR video quality assessment model based on H.265/HEVC bit-stream.The main contents of this paper are as follows: firstly,a software tool is designed for subjective video quality assessment and a subjective video database is established to store subjective quality of video sequences based on H.265/HEVC Video codec at different quantization parameters;secondly,the specific parameters which can reflect the degradation of video quality are extracted after analyzing the compressed bit-stream of H.265/HEVC;finally,a low complexity valuation model is built by using the extracted parameters,which is trained and tested in this paper.The experiment results show that the scores obtained by the proposed model have high correlation with subjective scores. |
PDF Full Text Request