4K, ultra high-definition (UHD), and higher resolution video contents have become increasingly popular recently. The largely increased data rate casts great challenges to video compression and communication technologies. Emerging video coding methods are claimed to achieve superior performance for high-resolution video content, but thorough and independent validations are lacking. In this study, we carry out an independent and so far the most comprehensive subjective testing and performance evaluation on videos of diverse resolutions, bit rates and content variations, and compressed by popular and emerging video coding methods including H.264/AVC, H.265/HEVC, VP9, AVS2 and AV1. Our statistical analysis derived from a total of more than 36,000 raw subjective ratings on 1,200 test videos suggests that significant improvement in terms of rate-quality performance against the AVC encoder has been achieved by state-of-the-art encoders, and such improvement is increasingly manifest with the increase of resolution. Furthermore, we evaluate state-of-the-art objective video quality assessment models, and our results show that the SSIMplus measure performs the best in predicting 4K subjective video quality. The database will be made available online to the public to facilitate future video encoding and video quality research.
CITATION STYLE
Li, Z., Duanmu, Z., Liu, W., & Wang, Z. (2019). AVC, HEVC, VP9, AVS2 OR AV1? — A comparative study of state-of-the-art video encoders on 4K videos. In Lecture Notes in Computer Science (including subseries Lecture Notes in Artificial Intelligence and Lecture Notes in Bioinformatics) (Vol. 11662 LNCS, pp. 162–173). Springer Verlag. https://doi.org/10.1007/978-3-030-27202-9_14
Mendeley helps you to discover research relevant for your work.