以物件辨識為基礎之抗縮放紋理影像分類器

An Object-based Classification Method for Rotating and Scaling Texture

余亞昀(Ya-yun Yu)

指導教授 ： 姚志佳

朝陽科技大學/資訊學院/資訊工程系/碩士(2012年)

摘要

本篇論文中提出了一個有效克服旋轉及縮放問題的混合型分類器，針對紋理影像經過放大或縮小後且經過裁切的影像進行辨識，此種影像常出現在照相機使用不同焦距來拍同一個景象時所出現的現象。這個混合分類器中包含了物件比對演算法及改良型賈波濾波器。物件比對演算法是以物件比對為基礎，找尋不同倍數的影像中相同內容的物件以進行物件相似性的匹配，並利用七個抗縮放的參數進行辨識。本篇論文所提出的另一種方法為改良型賈波濾波器，是根據賈波本身的多角度頻域優點，針對區域物件做訊號的轉換，並以區域物件的縮放比例，設定賈波頻率參數。此兩種分類器均採用SVM來當作分類模組，經實驗結果證明在抗旋轉及抗縮放辨識效果比其他方法來的好。

關鍵字

縮放 ； 賈波濾波器 ； 物件比對 ； 旋轉

並列摘要

This thesis developed a hybrid classifier for correcting texture variation resulting from scale magnification, narrowing caused by cutting into the original size, or spatial rotation. These variations usually occur in images captured by a camera using different focal distances. The hybrid classifier contains an object matching algorithm and an improved Gabor filter. In object matching algorithm the classification between two textures is based on the comparisons of a set of similar objects which are extracted by using JSEG method. By adopting seven invariant scaled parameters the similar objects can be identified. Besides, an improved Gabor filter is proposed. In the improved Gabor filter the setting of scale parameter is based on the scale of object and the scanning region is located within the object. Under this modification the Gabor filter is more precise and more effective. Finally, SVMs are used as a classification model. Experimental results show that our proposed method outperforms existing design algorithms.

並列關鍵字

Gabor filter ； Object matching algorithm ； Rotation ； Scale

參考文獻

[1] T.-S. Lee, “Image representation using 2D Gabor wavelets,” IEEE Transactions on Pattern Analysis and Machine Intelligence, 1996, Vol.18 , No.10, pp. 959-971.

[2] L.-L. Shen, J.-I. Zhen, ”Gabor wavelet selection and SVM classification for object recognition,” Acta Automatica Sinica, 2009, Vol.35, No.4, pp. 350-355.

[3] C. Koley, P. Purkait, S. Chakravorti ,”Wavelet-aided SVM tool for impulse fault identification in Transformers,” IEEE Transactions on Power Delivery,2006,Vol.21, No.3, pp. 1283-1290.

[4] S. Arivazhagan, L. Ganesan, ”Texture classification using wavelet transform,” Pattern Recognition Letters,2003,Vol.24, pp.1513-1521.

[5] C.-S. Lu, P.-C. Chung, C.-F. Chen, ”Unsupervised texture segmentation via wavelet transform,” Pattern Recognition, 1997,Vol.30, Issue 5, pp.729-742.