Large parallax image stitching and its quantitative assessment
- Title
- Large parallax image stitching and its quantitative assessment
- Translated Title
- 큰 시차 이미지 스티칭 및 정량적 평가 방법
- Authors
- Kyunghwa Jung
- DGIST Authors
- Kyunghwa Jung; Jaesung Hong; Hyunki Lee
- Advisor(s)
- 홍재성
- Co-Advisor(s)
- Hyunki Lee
- Issue Date
- 2021
- Available Date
- 2022-07-07
- Degree Date
- 2021/02
- Type
- Thesis
- Table Of Contents
- Ⅰ. Introduction 1 1.1 Narrow field-of-view of arthroscopic image 1 1.2 Panoramic image stitching 2 1.3 Assessment of image stitching methods 3 1.4 Thesis contributions 5 Ⅱ. Background and related works 6 2.1 Parallax 6 2.1.1 Planar parallax in computer vision 6 2.1.2 Parallax level caused by camera configuration 8 2.2 Parallax-tolerant image stitching 11 2.2.1 Feature correspondence 11 2.2.2 Image warp for alignment 11 2.3 Quality assessment of stitched images 13 Ⅲ. Method 15 3.1 Proposed object panorama method 15 3.1.1 Active feature matching and outlier rejection 15 3.1.2 Image morphing using a curved interpolation line 16 3.1.3 Seamless composition 20 3.2 Assessment of parallax-tolerant image stitching method 21 3.2.1 Planar parallax estimation 22 3.2.2 Completion of the image dataset 25 3.2.3 Grouping of parallax images 25 Ⅳ. Experiments and results 27 4.1 Proposed object panorama method 27 4.1.1 Comparison in terms of the visible region 27 4.1.2 Panorama according the difference in viewpoints 28 4.1.3 Comparison with other panorama algorithms 29 4.1.3.1 Results using AutoStitch, Photoshop, and ICE 30 4.1.3.2 Results using APAP, SPHP, ANAP, ELA, and the proposed method 30 4.1.4 Similarity with the middle region 32 4.2 Assessment of parallax-tolerant image stitching method 35 4.2.1 Reliability of parallax estimation metric 35 4.2.2 Grouping of parallax image 38 4.2.3 Performance assessment on grouped image 40 4.3 Assessment of the proposed object panorama method 45 4.3.1 Parallax level of input image 45 4.3.2 Assessment about parallax improvement 45 Ⅴ. Discussion 47 Ⅵ. Conclusion 52
- Degree
- Doctor
- Department
- Robotics Engineering
- University
- DGIST
- Related Researcher
-
-
Lee, HyunKi Camera Culture Group
- Research Interests Machine Vision, Intelligent Robot, Design of Optical Systems
-
- Files:
There are no files associated with this item.
Items in DSpace are protected by copyright, with all rights reserved, unless otherwise indicated.