Reconstruction of super-resolution images using a new adaptive kernel function and multiframe image fusion새로운 적응 커널 함수와 다중 영상 융합 기법을 이용한 초해상도 영상 복원에 관한 연구

Cited 0 time in webofscience Cited 0 time in scopus
  • Hit : 366
  • Download : 0
To increase the image resolution, various approaches have been proposed over past several decades. Super-resolution is one of the very active research area of this purpose and indicates a lot of different approaches to enhance the image resolution from the frequency domain to the spatial domain super-resolution reconstruction algorithms. In conventional super-resolution algorithms, a reference frame for the super-resolution image reconstruction is selected arbitrarily between given input frames. This thesis deals with the problem of the reference frame creation. The quality of the reconstructed high resolution image depends strongly on the quality of the reference frame of the given input frames. In order to improve the quality of super-resolution image reconstruction, a good reference frame should be selected and this is an open problem because the subjective visual quality is hard to be measured numerically. This thesis shows that the problem of finding the good reference frame can be changed to the problem of creating the good reference frame which is also robust to the outlier frame contamination. The quality of the created reference frame is independent of the quality of each input frame and very robust to the outlier contamination so that it can be recommended for the super-resolution image reconstruction. In this thesis, the data-adaptive super-resolution algorithm is also considered because the conventional super-resolution algorithms usually use a single scheme for a whole image, regardless of the regional characteristics. Since an image consists of various regions having different characteristics, these algorithms may not provide equally good performance for all regions. The steering kernel regression framework for super-resolution image reconstruction is one of data-adaptive algorithms in order to consider these regional characteristics using variable shapes of kernel functions to the different characteristic regions. For the point of maintaining ...
Advisors
Park, Dong-Joresearcher박동조researcher
Description
한국과학기술원 : 전기 및 전자공학과,
Publisher
한국과학기술원
Issue Date
2011
Identifier
467853/325007  / 020093250
Language
eng
Description

학위논문(석사) - 한국과학기술원 : 전기 및 전자공학과, 2011.2, [ vii, 88 p. ]

Keywords

super-resolution; reference frame creation; multiframe image fusion; data-adaptivity; 초해상도; 기준 영상 생성; 다중 영상 융합; 영상 특성에 대한 적응성; 커널 회귀법; kernel regression

URI
http://hdl.handle.net/10203/180755
Link
http://library.kaist.ac.kr/search/detail/view.do?bibCtrlNo=467853&flag=dissertation
Appears in Collection
EE-Theses_Master(석사논문)
Files in This Item
There are no files associated with this item.

qr_code

  • mendeley

    citeulike


rss_1.0 rss_2.0 atom_1.0