A DEFORMABLE MATCHING OF 2-D BOUNDARY SHAPES USING WAVELET TRANSFORM

Accession number;07A0187180
Title;A DEFORMABLE MATCHING OF 2-D BOUNDARY SHAPES USING WAVELET TRANSFORM
Author; HATA TAKASHI (Iwate Univ., Iwate, Jpn) KIMURA AKIO (Iwate Univ., Iwate, Jpn) WATANABE TAKASHI (Iwate Univ., Iwate, Jpn)
Journal Title;IEIC Technical Report (Institute of Electronics, Information and Communication Engineers)
Journal Code:S0532B
ISSN:0913-5685
VOL.106;NO.448(IE2006 138-184);PAGE.35-40(2007)
Figure&Table&Reference;
Pub. Country;Japan
Language;English
Abstract;We present a wavelet transform-based template matching method to be able to deal with local boundary deformations of a two-dimensional shape. Template matching is one of the most important research subjects in image processing or image understanding. Particularly, in many industrial applications, a fast and flexible method is required to realize a practical matching system. In order to establish such a robust matching, in this paper, we describe both of a template and a target shape as Mallat's wavelet zero-crossings representations, and measure the dissimilarity between them by comparing amplitudes and intervals of these zero-crossing points. Since this algorithm adopts Mallat's dyadic wavelet transform, it can disregard local boundary deformations at the higher scales. And also, since the two-dimensional shape's information is translated into a one-dimensional description adequately, processing is very fast. Experimental results showed that our method performs well to both artificial and real images. (author abst.)