291 0

앙상블 희소 코딩 모형을 이용한 객체와 배경이 모호한 이미지 분류

Title
앙상블 희소 코딩 모형을 이용한 객체와 배경이 모호한 이미지 분류
Other Titles
Ensemble patch sparse coding for ambiguous edge image classification
Author
이재환
Alternative Author(s)
Lee, Jae Hwan
Advisor(s)
이기천
Issue Date
2017-02
Publisher
한양대학교 일반대학원
Degree
Master
Abstract
Sparse coding methods with a learned dictionary have been successful in several image classification problems. However, sparse representations from a unit dictionary may not contain full information as they are affected by environmental factors of images such as light, shadow, background, and forth. In addition, sparse features formed by one dictionary can fall into a trap of singularity of training image data. To handle these problems, especially in images with ambiguous edges, we propose a new sparse coding method based on ensembles of image patches. Thus, we transform such images into overlapped patches for better classification performance. Then, we assign patch-wise weights and learn optimal weights not by single sparse representation but by ensemble learning. In learning optimal weights, we propose a two-step update scheme. We collectively update the weights of all patches in a misclassified image. Furthermore we propagated the weights of misclassified patches to those of other patches in the image. Experimental results on the Northeastern University surface defect data sets and close-up skin data sets show better classification accuracy than some existing methods and demonstrate the potential advantage of the proposed method in ambiguous-edge image classification.
URI
https://repository.hanyang.ac.kr/handle/20.500.11754/124936http://hanyang.dcollection.net/common/orgView/200000430270
Appears in Collections:
GRADUATE SCHOOL[S](대학원) > INDUSTRIAL ENGINEERING(산업공학과) > Theses (Master)
Files in This Item:
There are no files associated with this item.
Export
RIS (EndNote)
XLS (Excel)
XML


qrcode

Items in DSpace are protected by copyright, with all rights reserved, unless otherwise indicated.

BROWSE