View : 699 Download: 0

Full metadata record

DC Field Value Language
dc.contributor.advisor윤정호-
dc.contributor.author최보수-
dc.creator최보수-
dc.date.accessioned2016-08-26T12:08:02Z-
dc.date.available2016-08-26T12:08:02Z-
dc.date.issued2012-
dc.identifier.otherOAK-000000072464-
dc.identifier.urihttps://dspace.ewha.ac.kr/handle/2015.oak/189371-
dc.identifier.urihttp://dcollection.ewha.ac.kr/jsp/common/DcLoOrgPer.jsp?sItemId=000000072464-
dc.description.abstractWe propose a method for image denoising which is based on the well-known non-local means(NLM) algorithm. However, it introduces a fundamental modification. The suggested method first approximates gradient vectors by using the moving least squares (MLS) methods to find patches which are similar to the pixel to be repaired. If there is higher similarity between the evaluation point and reference value in their patches, larger weight is given to that reference value. In addition, we adapt an auxiliary function to adjust the given data before the multiplication. This method is iteratively applied for effective denoising. Experimental results are presented and compared to those of the classical non-local means and other recently developed nonlinear methods: total variation and non-local total variation. The comparison is made visually and numerically by using two quantitative fidelity measures: PSNR and Mean SSIM. The results demonstrate the new algorithm’s ability to denoise an image while preserving edge features.;이 논문에서는 수정된 non-local means 방법을 반복적으로 적용하는 denoising 방법을 소개한다. 기존의 방법은 새롭게 정의하고자 하는 픽셀을 그 주변의 픽셀들의 비슷한 정도를 측정하여 kernel 값을 구하지만 필자는 moving least squares 방법을 이용하여 구한 gradient vector의 비슷한 정도도 고려해서 kernel 값을 결정한다. 이러한 방법은 측정하고자하는 픽셀과 비슷한 구조를 갖는 픽셀들을 더 정확히 찾아준다. 또한 Gradient 값은 비슷하지만 이미지 값이 다른 경우를 위해 보정 함수도 새롭게 도입한다. 결과적으로 노이즈도 잘 제거하지만 세세한 패턴도 기존 방법보다 잘 유지시키는 결과를 얻을 수 있다. 여러 가지 이미지에 적용한 시각적인 결과와 수치적인 결과는 이 논문에서 제안하는 방법의 우수성을 보여준다.-
dc.description.tableofcontents1 Introduction 1 2 A Novel Iterative Non-Local Means Method 6 2.1 The Classical Non-Local Means Method 6 2.2 A New NLM Method with Gradient Similarity Measure 8 2.3 Iterative Non-Local Means 11 2.4 The Approximation to the Gradients 13 3 Experimental Results 18 4 Conclusion 26 References 27 논문초록 29-
dc.formatapplication/pdf-
dc.format.extent2829011 bytes-
dc.languageeng-
dc.publisher이화여자대학교 대학원-
dc.subject.ddc500-
dc.titleA New Iterative Non-Local Means for Image Denoising-
dc.typeMaster's Thesis-
dc.format.pageii, 29 p.-
dc.identifier.thesisdegreeMaster-
dc.identifier.major대학원 수학과-
dc.date.awarded2012. 8-
Appears in Collections:
일반대학원 > 수학과 > Theses_Master
Files in This Item:
There are no files associated with this item.
Export
RIS (EndNote)
XLS (Excel)
XML


qrcode

BROWSE