| Author: | Wang, Xiuyuan |
| Title: | Stereoscopic image reflection removal based on Wasserstein Generative Adversarial Network |
| Advisors: | Lun, P. K. Daniel (EIE) |
| Degree: | M.Sc. |
| Year: | 2020 |
| Subject: | Image processing -- Digital techniques Hong Kong Polytechnic University -- Dissertations |
| Department: | Department of Electronic and Information Engineering |
| Pages: | 41 pages : color illustrations |
| Language: | English |
| Abstract: | Reflection removal is a long-standing problem in computer vision. Although much progress has been made in single-image solutions, the limitations are also obvious due to the challenging nature of this problem. In this study, we propose to use stereoscopic image pairs of a scene for reflection removal. In comparison to the previous work which is on the basis of 5 views, our proposed approach shows higher flexibility and more advantages especially when the reflection is strong. Besides, it can give a more natural perceptual effect. For the proposed approach, we first propose a Block Matching-based method for disparity estimation, given two views of one scene to calculate relative motions. Then, we separate the background and reflection edges using the K-Means algorithm, because it is assumed that the motions of reflection are always less than that of the background. Given the powerful reconstruction ability of the Wasserstein Generative Adversarial Network (WGAN), we reconstruct the background edge map from the initial estimate using the proposed edge reconstruction network (ERN). Finally, the whole background is reconstructed by another WGAN, called the background reconstruction network (BRN). We compare the performance of the proposed approach with the state-of-the-art reflection removal methods. Results show that our approach performs better especially when the reflection in the image is strong. |
| Rights: | All rights reserved |
| Access: | restricted access |
Files in This Item:
| File | Description | Size | Format | |
|---|---|---|---|---|
| 5171.pdf | For All Users (off-campus access for PolyU Staff & Students only) | 1.33 MB | Adobe PDF | View/Open |
Copyright Undertaking
As a bona fide Library user, I declare that:
- I will abide by the rules and legal ordinances governing copyright regarding the use of the Database.
- I will use the Database for the purpose of my research or private study only and not for circulation or further reproduction or any other purpose.
- I agree to indemnify and hold the University harmless from and against any loss, damage, cost, liability or expenses arising from copyright infringement or unauthorized usage.
By downloading any item(s) listed above, you acknowledge that you have read and understood the copyright undertaking as stated above, and agree to be bound by all of its terms.
Please use this identifier to cite or link to this item:
https://theses.lib.polyu.edu.hk/handle/200/10769