期刊刊名：工程應用技術學刊 　 卷期：1卷1期

篇名出版日期：2012年2月1日

作者：郭振輝,賴岦俊,Chen-Hui Kuo,Li-Chun Lai

語言：English

關鍵字：Face recognition, Support vector machine, Feature selection, Hidden Markov model, Discrete cosine transform, Discrete wavelet transform, 人臉辨識,支援向量機,特徵選取,隱馬可夫模型,離散餘旋轉換,離散小波轉換

被點閱次數：4次

閱讀時間：23sec

摘要：
In this paper, we propose a comparison of DCT and DWT methods for feature selection, also comparison of SVM and HMM methods for facial classifiers. We used two scan methods, top-bottom and raster scan, on purpose for data scan and then to compare the performance of two scan method. The DCT and DWT are used to frequency domain analysis to reduce the feature dimension. Nevertheless, we prove that they are the same result if apply our extract feature method. SVM was originally designed for linear binary classification. Our experiments use one-against-one method for multi-class classification because it’s more suitable for practical use than the method of one-against-all. HMM has been applied with success in speech recognition. This paper reports on a comparison of the two classifiers in facial recognition. We have tested two classifiers on the ORL facial database.

　　　　　[全文下載]