doi:10.1016/S0167-8655(01)00099-X
Copyright © 2002 Elsevier Science B.V. All rights reserved.
Rotation-invariant pattern matching using wavelet decomposition
References and further reading may be available for this article. To view references and further reading you must
purchase this article.
Du-Ming Tsai
, 
and Cheng-Huei Chiang
Department of Industrial Engineering and Management, Yuan-Ze University, 135 Yuan-Tung Road, Nei-Li, Tao-Yuan 32026, Taiwan, ROC
Received 5 October 2000;
Revised 12 March 2001.
Available online 27 November 2001.
Abstract
In this paper, we propose a wavelet decomposition approach for rotation-invariant template matching. In the matching process, we first decompose an input image into different multi-resolution levels in the wavelet-transformed domain, and use only the pixels with high wavelet coefficients in the decomposed detail subimage at a lower resolution level to compute the normalized correlation between two compared patterns. To make the matching invariant to rotation, we further use the ring-projection transform, which is invariant to object orientation, to represent an object pattern in the detail subimage. The proposed method significantly reduces the computational burden of the traditional pixel-by-pixel matching. Experimental results on a variety of real images have shown the efficacy of the proposed method.
Author Keywords: Template matching; Object detection; Wavelet decomposition; Ring projection; Rotation-invariant
Fig. 1. One stage in a multi-resolution image decomposition.
Fig. 2. A head image in two different orientations: (a), (b) the original images; (c), (d) the corresponding ring-projection plots of (a) and (b), respectively.
Fig. 3. The effect of changes in wavelet support length: (a) the template image; (b)–(e) the detection results from wavelet bases Haar, D4, D8 and D12, respectively; (f)–(i) the composite detail subimages at resolution level 1 for respective (b)–(e).
Fig. 4. The effect of changes in image rotation for a Chinese “one ” image: (a) the template marked with a square; (b), (c) the rotated images in 45° and 90°, respectively (the detected objects are marked with circles); (d), (e) the corresponding composite detail subimages at resolution level 1.
Fig. 5. Detecting an IC component on the PCB: (a) the original PCB image; (b) the detected component marked with a circle; (c) the composite detail subimage at resolution level 2.
Fig. 6. Detecting a bar-coded strip: (a) the original image; (b) the detected strip marked with a circle; (c) the composite detail subimage at resolution level 1.
Fig. 7. Detecting license plates on cars: (a) the plate used as the template; (b), (c) the detected plates marked with circles; (d), (e) the corresponding composite detail subimages at resolution level 1.
Corresponding author. Tel.: +886-3-463-8800; fax: +886-3-463-8907
Abstract
Purchase PDF (509 K)
E-mail Article
Add to my Quick Links
Cited By in Scopus (20)
Purchase PDF (1374 K)
