2018 Fiscal Year Research-status Report
Modeling the Perceptual Underpinnings for Quality Assessment of Restored Textures
| Project/Area Number |
17K00232
|
|---|
| Research Institution | Shizuoka University |
|---|
Principal Investigator |
|
|---|
| Co-Investigator(Kenkyū-buntansha) |
大橋 剛介 静岡大学, 工学部, 教授 (80293603)
|
|---|
| Project Period (FY) |
2017-04-01 – 2020-03-31
|
| Keywords | quality assessment / texture quality / visual perception / visual masking / image restoration |
|---|
| Outline of Annual Research Achievements |
The objective of this research project is to investigate the perceptual foundations of restored textures, to research and develop associated computational models, and to research and develop practical QA algorithms for restoration applications. In Year 2 of this project, the objective was to create computational models that could predict the factors that underly restored-texture quality assessment (RTQA), and then use these models in an RTQA algorithm. We have developed two such models: (1) a model based on contrast detection thresholds, and (2) a model based on color and texture similarity. A basic RTQA algorithm was created. The features used for (2) were also used to develop a blind QA algorithm for multiply and singly distorted images.
|
| Current Status of Research Progress |
Current Status of Research Progress
3: Progress in research has been slightly delayed.
Reason
Our main objective in Year 2 was to create and refine independent models that underlie the quality judgement process, and then to combine these models into an RTQA algorithm. One major factor which we found to underlie QA is contrast detectability. We showed via experimental data that a strong relationship exists between contrast detection thresholds and optimal contrast scaling factors. We have successfully applied and refined a multichannel feature-based masking model to predict these data. As an added benefit, these features were also used to develop a blind QA algorithm for multiply and singly distorted images. The other major factors which we have researched is joint color and pattern similarity. This model is still under development.
|
| Strategy for Future Research Activity |
The research is largely on-schedule, except we have not yet finished the color and pattern similarity models. The reason for this delay is that we are trying to develop a model that can not only be used for QA, but also for segmentation during texture restoration. If we are successful, we will have a much more powerful algorithm that can not only assess quality, but which can also guide restoration. In Year 3, we will also begin to address RTQA of video textures, starting with a database. Analysis of the database will allow us to investigate key differences between image-based texture quality and video-based texture quality. We plan to apply our image-based models to the video data, first on a frame-by-frame basis, and later using spatiotemporal slices.
|
| Causes of Carryover |
There are unspent funds for three main reasons: (1) travel expenses were cheaper than planned; (2) we were able to save money by building computers; and (3) students in our research labs performed the experiments, thus minimizing the need to pay external subjects. We plan to use these savings to primarily pay for paper publication charges and travel in 2019.
|
Research Products
(4 results)
-
-
-
[Journal Article] Perspectives on the definition of visually lossless quality for mobile and large format displays2018
Author(s)
R. Allison, K. Brunnstrom, D. M. Chandler, H. Colett, P. Corriveau, S. Daly, J. Goel, J. Long, L. Wilcox, Y. Yaacob, S. Yang, Y. Zhang
-
Journal Title
Journal of Electronic Imaging
Volume: 27
Pages: 1-23
DOI
Peer Reviewed / Int'l Joint Research
-
