M.S. Farid, M. Lucenteforte, M. Grangetto, "PERCEPTUAL QUALITY ASSESSMENT OF 3D SYNTHESIZED IMAGES," in Proc. IEEE International Conference on Multimedia and Expo (ICME), Hong Kong, 2017, pp. 505-510.
Abstract: Multiview video plus depth (MVD) is the most popular 3D video format where the texture images contain the color information and the depth maps represent the geometry of the scene. The depth maps are exploited to obtain intermediate views to enable 3D-TV and free-viewpoint applications using the depth image based rendering (DIBR) techniques. DIBR is used to get an estimate of the intermediate views but has to cope with depth errors, occlusions, imprecise camera parameters, re-interpolation, to mention a few issues. Therefore, being able to evaluate the true perceptual quality of synthesized images is of paramount importance for a high quality 3D experience. In this paper, we present a novel algorithm to assess the quality of the synthesized images in the absence of the corresponding references. The algorithm uses the original views from which the virtual image is generated to estimate the distortion induced by the DIBR process. In particular, a block-based perceptual feature matching based on signal phase congruency metric is devised to estimate the synthesis distortion. The experiments worked out on standard DIBR synthesized database show that the proposed algorithm achieves high correlation with the subjective ratings and outperforms the existing 3D quality assessment algorithms.
Download
Experimental Evaluation:
Database used for evaluation: IRCCyN/IVC DIBR Images database
Table 1. Overall performance of DSQM and other 3D-IQA algorithms on IRCCyN/IVC DIBR Images database.
Metric | 3DSwIM | Tsai | PQM | Gorley | Dragana | MW-PSNR | SIQE | DSQM |
PLCC | 0.6420 | 0.4830 | 0.4709 | 0.3183 | 0.7191 | 0.7378 | 0.6058 | 0.7895 |
SROCC | 0.5613 | 0.4740 | 0.4869 | 0.3000 | 0.6846 | 0.7070 | 0.4347 | 0.7151 |
RMSE | 0.5105 | 0.5975 | 0.5877 | 0.6312 | 0.4627 | 0.4494 | 0.5298 | 0.4086 |
MAE | 0.4038 | 0.5051 | 0.4771 | 0.4876 | 0.3370 | 0.3184 | 0.4022 | 0.3067 |
Table 2. Overall performance of DSQM and other 2D-IQA algorithms on IRCCyN/IVC DIBR Images database.
Metric | SSIM | MS-SSIM | PSNR | VSNR | DSQM |
PLCC | 0.5639 | 0.5489 | 0.4283 | 0.5145 | 0.7895 |
SROCC | 0.4687 | 0.5324 | 0.4628 | 0.5051 | 0.7151 |
RMSE | 0.5499 | 0.5566 | 0.6017 | 0.5709 | 0.4086 |
Table 3. Performance of DSQM with different block sizes (m,n). Results of the best performing block size are in bold font.
mxn | PLCC | SROCC | RMSE |
8x8 | 0.5062 | 0.5301 | 0.5742 |
16x16 | 0.6665 | 0.5998 | 0.4964 |
32x32 | 0.6886 | 0.6083 | 0.4828 |
64x64 | 0.6962 | 0.604 | 0.4779 |
128x128 | 0.7895 | 0.7151 | 0.4086 |
256x256 | 0.7529 | 0.6749 | 0.4382 |
Full Image | 0.7261 | 0.6183 | 0.4578 |
Last updated: March 20, 2017