Optics and Precision Engineering, Volume. 30, Issue 22, 2923(2022)
No-reference video quality objective assessment method based on the content visual perception and transmission distortion
Figures & Tables(11)
Fig. 1. Flow chart of building video quality evaluation model based on four kinds of delay distortion
Fig. 2. Analyzing the correlation between the scores of evaluating videos in two databases by the proposed VQA model integrating multiple delay effects and the subjective MOS
Fig. 3. Analyzing the correlation between the scores of evaluating videos built two databases by the MCPBD model and the subjective MOS
Fig. 4. Analyzing correlation between the subjective VQA scores and the scores of evaluating videos in three open source databases by the proposed MCPBD method
Fig. 5. Comparing the operation time of the proposed model MCPBD with that of 9 VQA models
Table 1. Duration setting of video initial delay and middle single buffer delay
View tableView in ArticleTable 1. Duration setting of video initial delay and middle single buffer delay
Databases Parameters Parameter values LIVEour （25 fps） Initial delay 0.12 0.32 0.52 1.0 1.52 2.0 3.0 5.0 10.0 Middle buffer 0.12 0.32 0.52 1.0 1.52 2.0 3.0 5.0 10.0 VIPSLour （30 fps） Initial delay 0.10 0.30 0.50 1.0 1.50 2.0 3.0 5.0 10.0 Middle buffer 0.10 0.30 0.50 1.0 1.50 2.0 3.0 5.0 10.0
Table 2. Average duration setting of multiple buffer delay
View tableView in ArticleTable 2. Average duration setting of multiple buffer delay
Databases Number of buffer N Average buffer delay/s LIVEour（25 fps） 3/6/10 0.12 0.32 0.52 1 2 3 5 10 - - VIPSLour（30 fps） 2/3/4 0.1 0.3 0.5 0.8 1 1.5 2 3 4 5
Table 3. Buffer frequency （number） setting
View tableView in ArticleTable 3. Buffer frequency （number） setting
Databases Parameters Parameter values LIVEour（25 fps） Number of buffer N 1 2 3 5 8 10 — frequency of buffer/（N·min-1） 6 12 18 30 48 60 — Average buffer delay/s 0.1/1/5 VIPSLour（30 fps） Number of buffer N 1 2 3 4 5 6 8 frequency of buffer/（N·min-1） 6 12 18 24 30 24 48 Average buffer delay/s 0.3/1/2
Table 4. Comparing the accuracy PLCC and SROCC values of the proposed model MCPBD with those of 6 existing FR-VQA models
View tableView in ArticleTable 4. Comparing the accuracy PLCC and SROCC values of the proposed model MCPBD with those of 6 existing FR-VQA models
Databases LIVE Database LIVE Stall Ⅱ Our Database LIVEour VIPSLour VAQ models PSNR MS-SSIM VSNR VQM MOVIE ST-MAD MCPBD MCPBD MCPBD PLCC 0.596 2 0.700 1 0.656 7 0.723 6 0.818 1 0.826 7 0.884 6 0.926 0.914 8 SROCC 0.575 5 0.704 3 0.642 4 0.702 6 0.800 5 0.809 2 0.833 6 0.920 2 0.907 7
Table 5. Comparing the accuracy PLCC and SROCC values of the proposed model MCPBD with those of 11 existing NR-VQA models
View tableView in ArticleTable 5. Comparing the accuracy PLCC and SROCC values of the proposed model MCPBD with those of 11 existing NR-VQA models
Databases LIVE Database LIVE Qualcomm LIVE Stall Ⅱ Our Database LIVEour VIPSLour VAQ models NVSM 3D-DCT V-CORNIA V-BLIINDS COME VQAUCA TRR-QoE C-VQA NR-DCT BRVPVC QAIWV MCPBD MCPBD MCPBD PLCC 0.732 0.891 0.768 0.759 0.852 0.88 0.839 0.7927 0.7855 0.8547 0.732 0.884 6 0.926 0.914 8 SROCC 0.703 0.782 0.74 0.881 0.821 0.90 0.8221 0.772 0.8031 0.826 0.737 0.833 6 0.920 2 0.907 7
Table 6. Comparing the accuracy PLCC and SROCC values of the proposed model with those of seven existing models in four open source databases
View tableView in ArticleTable 6. Comparing the accuracy PLCC and SROCC values of the proposed model with those of seven existing models in four open source databases
Databases VQA models VQM MOVIE ST-MAD 3D-DCT V-BLIINDS VQAUCA BRVPVC MCPBD LIVE PLCC 0.723 6 0.811 6 0.829 9 0.891 0 0.759 0 0.88 0.854 7 0.884 6 SROCC 0.702 6 0.789 0 0.824 2 0.782 0 0.881 0 0.90 0.826 0 0.833 6 VQEG PLCC 0.746 0 0.858 0 0.857 7 0.424 0 0.567 0 — 0.895 1 0.877 3 SROCC 0.726 0 0.781 0 0.850 0 0.437 0 0.561 0 — 0.886 7 0.875 IRCCyN PLCC 0.753 7 0.753 7 — — — — 0.838 9 0.902 9 SROCC 0.782 5 0.782 5 — — — — 0.835 6 0.885 1 Lisbon PLCC 0.86 — 0.83 — — 0.75 — — SROCC 0.87 — 0.87 — — 0.79 — —
Tools
Get Citation
Copy Citation Text
Juncai YAO, Haowei TANG, Jing SHEN. No-reference video quality objective assessment method based on the content visual perception and transmission distortion[J]. Optics and Precision Engineering, 2022, 30(22): 2923
Paper Information
Category: Information Sciences
Received: Mar. 21, 2022
Accepted: --
Published Online: Nov. 28, 2022
The Author Email: YAO Juncai (yjc4782@163.com)
© Copyright 2018-2021 | Chinese Laser Press.
All Rights Reserved 沪ICP备15018463号-20