Laser & Optoelectronics Progress, Volume. 59, Issue 8, 0811001(2022)
No-Reference Image Quality Assessment of DIBR-Synthesized Images Based on Statistical Characteristics
Figures & Tables(11)
Fig. 2. Texture distortion blocks of DIBR image and corresponding Benford’s law feature values. (a) Reference block and black hole block; (b) geometric distortion block; (c) blurry block; (d) incorrect texture block; (e) corresponding feature value
Fig. 3. Traditional images with different distortion levels and corresponding Benford’s law feature values. (a) First-level Gaussian blur; (b) second-level Gaussian blur; (c) third-level Gaussian blur; (d) fourth-level Gaussian blur; (e) feature values extracted from six traditional images with different distortion levels
Fig. 4. The first digit probability of DCT coefficients of images with different quality
Fig. 5. DCT coefficients divided along the three directions of 45°, 90°, and 135°
Fig. 8. Results of different distortion types and different sequences. (a) Results of six conventional distorted images; (b) results of nine sequences of images
Table 1. Information of databases used in the experiments
View tableTable 1. Information of databases used in the experiments
Dataset Number of sequences Number of DIBR algorithms Number of other distortions Number of images IVC 3 7 None 84 IETR 10 7 None 140 MCL-3D 9 4 6 693
Table 2. Comparison of different algorithms on three datasets
View tableTable 2. Comparison of different algorithms on three datasets
Method Type IVC dataset IETR dataset MCL-3D dataset PLCC SRCC RMSE PLCC SRCC RMSE PLCC SRCC RMSE Bosc R 0.5841 0.4903 0.5408 0.4536 0.4330 2.2980 3DSwIM 0.6864 0.6125 0.4842 0.6519 0.5683 1.9729 ST-SIAQ 0.6914 0.6746 0.4812 0.3345 0.4232 0.2336 0.7133 0.7034 1.8233 MP-PSNR 0.6729 0.6272 0.4925 0.5753 0.5507 0.2032 0.7831 0.7899 1.6179 MW-PSNR 0.6200 0.5739 0.5224 0.5301 0.4845 0.2106 0.7654 0.7721 1.6743 SC-IQA 0.8496 0.7640 0.3511 0.6856 0.6423 0.1805 0.8194 0.8247 1.4913 LOGS 0.8256 0.7812 0.3601 0.6687 0.6683 0.1845 0.7614 0.7579 1.6873 APT NR 0.7307 0.7157 0.4546 0.4225 0.4187 0.2252 0.6433 0.6200 1.9870 MNSS 0.7700 0.7850 0.4120 0.3387 0.2281 0.2333 0.3766 0.3531 2.4101 NIQSV 0.6346 0.6167 0.5146 0.1759 0.1473 0.2446 0.6460 0.5792 1.9820 NIQSV+ 0.7114 0.6668 0.4679 0.2095 0.2190 0.2429 0.6138 0.6213 2.0375 Proposed method 0.8416 0.7768 0.4802 0.5185 0.4132 0.2073 0.9093 0.8907 1.1043
Table 3. Separate experiment of three features
View tableTable 3. Separate experiment of three features
Feature IVC IETR MCL-3D PLCC SRCC RMSE PLCC SRCC RMSE PLCC SRCC RMSE Benford’s law 0.7927 0.5951 0.5435 0.5400 0.3842 0.2000 0.7156 0.7137 1.8287 DCT coefficients of variation 0.2829 -0.1096 0.7402 0.1451 -0.0257 0.2334 0.7302 0.6678 1.7579 GGD 0.7135 -0.4372 0.6230 0.3314 0.1319 0.2341 0.8488 0.8301 1.4001
Tools
Get Citation
Copy Citation Text
Yanli Li, Ruofeng Xu. No-Reference Image Quality Assessment of DIBR-Synthesized Images Based on Statistical Characteristics[J]. Laser & Optoelectronics Progress, 2022, 59(8): 0811001
Paper Information
Category: Imaging Systems
Received: Feb. 26, 2021
Accepted: Apr. 14, 2021
Published Online: Apr. 11, 2022
The Author Email: Xu Ruofeng (5189@cumt.edu.cn)
© Copyright 2018-2021 | Chinese Laser Press.
All Rights Reserved 沪ICP备15018463号-20