No-reference artifacts measurements based video quality metric

• Published in: Signal processing. Image communication 2019-10, Vol.78, p.345-358
• Main Authors: Vranješ, Mario, Bajčinovci, Viliams, Grbić, Ratko, Vajak, Denis
• Format: Article
• Language: English
• Subjects: Algorithms; AMB-VQM; Compression tests; Freezing; Graphical user interface; Masking; Multimedia; No-reference; Performance prediction; Quality; Quality assessment; Signal quality; User satisfaction; Video artifacts; Video compression; Video quality assessment; Video quality database; Video signals
• ISSN: 0923-5965; 1879-2677
• DOI: 10.1016/j.image.2019.07.015

In multimedia delivery, perceived quality of video signal has a significant role in overall users Quality of Experience (QoE). Therefore, multimedia service providers must constantly measure and monitor perceived video quality, which is usually performed using no-reference (NR) video quality metrics. In this paper, a novel NR objective video quality metric named Artifacts Measurements Based Video Quality Metric (AMB-VQM) is proposed. In addition to artifacts measures computed by blocking (BL), packet-loss (PL), and freezing (FZ) artifacts detection algorithms, the metric incorporates artifacts masking based on spatial and temporal video content complexity. Furthermore, a newly created FERIT-RTRK-2 video quality database, which contains 486 Full HD test video sequences impaired by video compression (MPEG-2, H.264/AVC, and H.265), packet-loss, frame freezing, and combinations of these procedures, is presented in this paper. FERIT-RTRK-2 database is publicly available for scientific community at http://www.rt-rk.com/other/VideoDBReadme.html. Additionally, newly created user-friendly subjective video quality assessment tool with graphical user interface, which can be used for conducting of subjective video quality experiments, is presented. In experimental part the performance of the proposed AMB-VQM is compared to this of 10 other objective video quality metrics (PSNR, SSIM, VSNR, PSNRHVS, PSNRHVSM, VIFP, ViS3, ST-MAD, BRISQUE, VIIDEO) using distorted video sequences from four different video quality databases: CSIQ, LIVE, FERIT-RTRK and FERIT-RTRK-2. The results show that AMB-VQM achieves high performance when predicting the video quality for videos distorted in different manners. AMB-VQM results outperform the results of most of the analyzed popular and very often used video quality metrics. •We made new no-reference VQA metric called AMB-VQM, which achieves high performance.•We made FERIT-RTRK-2 video database with 486 signals and their subjective scores.•We made new user-friendly subjective VQA tool with graphical user interface.•We performed the comparison for 9 objective quality metrics on 4 different databases.•AMB-VQM outperforms most of the analyzed popular and often used video quality metrics.