지원사업
학술연구/단체지원/교육 등 연구자 활동을 지속하도록 DBpia가 지원하고 있어요.
커뮤니티
연구자들이 자신의 연구와 전문성을 널리 알리고, 새로운 협력의 기회를 만들 수 있는 네트워킹 공간이에요.
논문 기본 정보
- 자료유형
- 학술저널
- 저자정보
- 저널정보
- 대한의용생체공학회 Biomedical Engineering Letters (BMEL) Biomedical Engineering Letters (BMEL) Vol.4 No.1
- 발행연도
- 2014.1
- 수록면
- 55 - 67 (13page)
이용수
초록· 키워드
오류제보하기
Purpose In the image processing community total variation(TV) is widely acknowledged as a popular and state-of-thearttechnique for noise reduction because of its edgepreservingproperty. This attractive feature of TV is dependenton optimal selection of regularization parameter. Contributionsin literature on TV focus on applications, properties and thedifferent numerical solution methods. Few contributionswhich address the problem of regularization parameterselection are based on regression methods which pre-existintroduction of TV. They are generic and elegantly formulated,and their operation is in series with TV framework. For thesereasons they render TV computationally inefficient and thereis significant manual tuning when they are deployed inspecific applications.
Methods This paper describes a non-regression approach forselection of regularization parameter. It is based on a newconcept, the Variational-Bayesian (VB) cycle. Within thecontext of VB cycle we derive two important results. First,we confirm the notion held for a long time by researchers,within image processing and computer vision community, that variational and Bayesian techniques are equivalent.
Second, the value of regularization parameter is equal tonoise variance, and is determined, at no computational costto TV denoising algorithm, from a mathematical model thatdescribes relationship between Markov random field energyand noise level in magnetic resonance images (MRI) ofbrain. The second result is similar to one reported in [1] inwhich the authors, for special choice of regularizationoperator in different regression methods, derive value ofregularization parameter as equal to noise variance.
Results Our proposal was evaluated on brain MRI imageswith different acquisition protocols from two clinical trialsstudy management centers. It was based on visual quality,computation time, convergence and optimality.
Conclusions The result shows that our proposal is suitable inapplications where high level of automation is demandedfrom image processing software.
목차
등록된 정보가 없습니다.
참고문헌 (59)
참고문헌 신청
Galatsanos NP / 1992 / Methods for choosing the regularization parameter and estimating the noise variance in image restoration and their relation / IEEE T Image Process 1(3):322~336
Zhang Y / 2001 / Segmentation of brain mr images through a hidden markov random field model and the expectation-maximization algorithm / IEEE T Med Imaging 20(1):45~57
Osadebey M. / 2009 / Simulation of realistic head geometry using radial vector representation of magnetic resonance image data
Gold R / 2012 / Placebo-controlled phase 3 study of oral bg-12 for relapsing multiple sclerosis / New Engl J Med 367(12):1098~1107
Thulborn KR / 2000 / Volumetry and topography of the human brain by magnetic resonance / Int J Imag Syst Tech 11(3):198~208
Zhang Y / 2001 / Segmentation of brain mr images through a hidden markov random field model and the expectation-maximization algorithm / IEEE T Med Imaging 20(1):45~57
Osadebey M. / 2009 / Simulation of realistic head geometry using radial vector representation of magnetic resonance image data
Gold R / 2012 / Placebo-controlled phase 3 study of oral bg-12 for relapsing multiple sclerosis / New Engl J Med 367(12):1098~1107
Thulborn KR / 2000 / Volumetry and topography of the human brain by magnetic resonance / Int J Imag Syst Tech 11(3):198~208
이 논문의 저자 정보
최근 본 자료
댓글(0)
0