Object Class Detection using Shape-Based Feature Descriptors

Title
Object Class Detection using Shape-Based Feature Descriptors
Authors
김동환박성기
Keywords
object class detection; object categorization; affine category shape model; spectral matching; local shape histogram
Issue Date
2010-11
Publisher
The 7th International Conference on Ubiquitous And Ambient Intelligence
Abstract
This paper presents a new object class detection technique using shape-based feature descriptors. We develop our algorithm based on the affine category shape model, which is a graph model consists of interconnected nodes of which geometric interactions are modeled with pairwise potentials between nodes. Since colors or texture components of objects can be different one another in an object class, we have a focus on shape information of objects for our feature descriptors. Especially, we employ relative geometric relationship of features, which makes our algorithm invariant to translation and in-plane rotation and robust to affine transformation. Experimental results on object category database show that the proposed algorithm is robust to pose variation of objects and provides good detection performance.
URI
http://pubs.kist.re.kr/handle/201004/39106
Appears in Collections:
KIST Publication > Conference Paper
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