Journal Search Engine
Search Advanced Search Adode Reader(link)
Download PDF Export Citaion korean bibliography PMC previewer translate in English
ISSN : 1975-6291(Print)
ISSN : 2287-3961(Online)
Journal of Korea Robotics Society Vol.4 No.3 pp.185-191

Affine Category Shape Model을 이용한 형태 기반 범주 물체 인식 기법

김 동 환1, 최 유 경1, 박 성 기1†

A New Shape-Based Object Category Recognition Technique using Affine Category Shape Model

Sung-Kee Park1†, Dong Hwan Kim1, Yukyung Choi1


This paper presents a new shape-based algorithm using affine category shape modelfor object category recognition and model learning. Affine category shape model is a graph ofinterconnected nodes whose geometric interactions are modeled using pairwise potentials. In itslearning phase, it can efficiently handle large pose variations of objects in training images byestimating 2-D homography transformation between the model and the training images. Sincethe pairwise potentials are defined on only relative geometric relationship betweenfeatures, theproposed matching algorithm is translation and in-plane rotation invariant and robust to affinetransformation. We apply spectral matching algorithm to find feature correspondences, whichare then used as initial correspondences for RANSAC algorithm. The 2-D homographytransformation and the inlier correspondences which are consistent with this estimate can beefficiently estimated through RANSAC, and new correspondences also can be detected byusing the estimated 2-D homography transformation. Experimental results on object categorydatabase show that the proposed algorithm is robust to pose variation of objects and providesgood recognition performance.


오늘하루 팝업창 안보기 닫기