Publications
Detailed Information
Multi-image blind deconvolution using low-rank representation : 저차수 표현을 이용한 다영상 블라인드 디콘볼루션
DC Field | Value | Language |
---|---|---|
dc.contributor.advisor | 유석인 | - |
dc.contributor.author | 강덕영 | - |
dc.date.accessioned | 2017-07-13T07:22:09Z | - |
dc.date.available | 2017-07-13T07:22:09Z | - |
dc.date.issued | 2017-02 | - |
dc.identifier.other | 000000142342 | - |
dc.identifier.uri | https://hdl.handle.net/10371/119297 | - |
dc.description | 학위논문 (박사)-- 서울대학교 대학원 : 전기·컴퓨터공학부, 2017. 2. 유석인. | - |
dc.description.abstract | Blind deconvolution is a restoration process of an image which is blurred by an unknown point spread function (PSF) (a.k.a. blur kernel). Some previous works show that multiple blurred captures of the same scene can improve the quality of blind deconvolution result. However, the previous multi-image blind deconvolution methods are prone to inconsistencies between observations such as moving objects, illumination changes or mis-alignment. In this paper, we present a new multi-image blind deconvolution algorithm which natually combines low-rank approximation, moving object detection and incremental alignment with blind deconvolution in a unified framework. Our framework alternatively solves a Schatten-0 norm low-rank approximation and MAP-based L0 norm blind deconvolution for finding the true latent images and their corresponding PSFs and transformation parameters. The experimental results show that our approach can recover high quality images in the presence of possible corruptions on both static and moving scenes and outperforms the state of the art results. | - |
dc.description.tableofcontents | Chapter 1 Introduction 1
Chapter 2 Related Work 6 2.1 Single-image Blind Deconvolution 6 2.2 Multi-image Blind Deconvolution 8 2.3 Image Alignment 11 2.3.1 Image Transformation 11 2.4 Low-rank Representation 13 Chapter 3 Preliminary: Optimization 18 3.1 Conjugate Gradient 18 3.2 Iteratively Reweighted Least Squares 22 3.3 Generalized Soft Thresholding 23 3.4 Singular Value Thresholding 25 3.5 Inexact Augmented Lagrangian Method 26 3.6 Lucas-Kanade Alignment Algorithm 29 Chapter 4 Model and Algorithm 32 4.1 Formulation 32 4.2 Optimization 35 4.2.1 Low-rank Approximation 37 4.2.2 Blur Kernel Estimation 40 4.2.3 Latent Image Estimation 41 Chapter 5 Implementation Detail 43 5.1 Reducing Interpolation Error 43 5.2 Boundary Handling Method 44 5.3 Multi-scale Approach 47 5.4 Point Spread Function Centering and Sub-pixel Alignment 49 Chapter 6 Experimental Result 52 6.1 Evaluation on Static Scene Data 52 6.2 Evaluation on Moving Scene Data 71 Chapter 7 Conclusion 89 Bibliography 91 요약 104 | - |
dc.format | application/pdf | - |
dc.format.extent | 13348049 bytes | - |
dc.format.medium | application/pdf | - |
dc.language.iso | en | - |
dc.publisher | 서울대학교 대학원 | - |
dc.subject | multi-image blind deconvolution | - |
dc.subject | low-rank representation | - |
dc.subject | moving object detection | - |
dc.subject | image alignment | - |
dc.subject.ddc | 621 | - |
dc.title | Multi-image blind deconvolution using low-rank representation | - |
dc.title.alternative | 저차수 표현을 이용한 다영상 블라인드 디콘볼루션 | - |
dc.type | Thesis | - |
dc.contributor.AlternativeAuthor | Deokyoung Kang | - |
dc.description.degree | Doctor | - |
dc.citation.pages | 104 | - |
dc.contributor.affiliation | 공과대학 전기·컴퓨터공학부 | - |
dc.date.awarded | 2017-02 | - |
- Appears in Collections:
- Files in This Item:
Item View & Download Count
Items in S-Space are protected by copyright, with all rights reserved, unless otherwise indicated.