透過您的圖書館登入
IP:18.224.33.107
  • 學位論文

結合生物特徵與主成分分析法的人臉影像辨識

Face Recognition Method by Integrating the Techniques of Biometrics and Principal Component Analysis

指導教授 : 陳榮昌
若您是本文的作者,可授權文章由華藝線上圖書館中協助推廣。

摘要


本研究提出一個結合生物特徵與主成分分析法的人臉影像識別方法,並以此為基礎建立一個具有多種生物特徵的人臉影像識別系統來實驗所提出的識別方法。在此系統中,我們使用了雙色彩空間的模型來加強處理膚色分離以擷取人臉,並從所有候選人臉影像擷取出各個人臉的生物特徵,計算出各個候選人臉的人臉特徵差異向量,接著透過主成分分析法來找到特徵的權重向量,然後將這些人臉特徵差異向量及權重向量存到資料庫,當輸入一個欲辨識的人臉影像時,只要擷取出該人臉的生物特徵,計算出人臉特徵差異向量,就可以利用這些權重向量,依照我們提出的人臉識別流程,循序漸進的加權比對,找出最接近的人臉。 最後,我們將上述流程與方法以實驗進行實驗,不斷測試與校正流程後獲得初步的識別成功率,驗證以主成分分析法結合生物特徵確實可以對人臉進行識別,且因為是取用幾個具有識別力的生物特徵作為人臉識別依據,所以與過去其他研究先進所提之方法,以整張人臉像素為識別依據相比可以節省下更多的資料量與計算量。雖然現階段只提出了少數幾個生物特徵做為識別依據,但相信未來以此作為基礎,繼續參考其他學者針對生物特徵的相關研究來增加選用的生物特徵多樣性,應可持續提高識別的成功率,使整個人臉識別系統既快速又準確。

並列摘要


This study proposes a face recognition method by integrating the techniques of biometrics and principal component analysis (PCA). Based on this method, we construct a multi-face recognition system which is based on fourteen biometric features. In this system, improve the detection process by using two color space models to extract face regions from the picture. We capture the biometric features from every candidate of face image, calculate the difference of facial feature vector (DFFV) and find weights of feature vector by PCA. Then, these data are stored in facial database for face recognition. When a new face image is coming, we capture the biometric features can be captured from the coming face image, calculate DFFV, and compare them with DFFVs in database by progressively use the weights which obtained from the PCA to find the closest face. Finally, we continued test and regulate the experimental procedure, we obtain initial recognition success rate and confirmed our face recognition method which used PCA and biometrics can be used. And because this method use only some biometrics features to detection face, therefore, we can save more calculation time and amount of data. In the future, we need to study more the other research on biometrics and improve amount of features, we think it can be improve success rate make the face recognition both fast and accurately.

參考文獻


[1] 吳明芳、李振興、王炳聰、詹慧珊、黃建邦(2011),「多人臉影像視覺辨識技術」,Journal of Kun Shan University,No. 8,第75-95頁。
[5] 顏于嘉(2006),生物特徵與資訊隱私權—從國家利用個人生物特徵辨識人民身分談起,碩士論文,台灣大學法律系,台北。
[6] 薛傑仁(2010),生物辨識之人臉辨識的方法,碩士論文,亞洲大學生物與醫學資訊系碩士學位論文,台中。
[8] M. M. Aznaveh, H. Mirzaei, E. Roshan, and M. Saraee(2008), “A New Color Based Method for Skin Detection Using RGB Vector Space," Human System Interactions Conference, pp. 932-935.
[9] M. A. Berbar, H. M. Kelash, and A. A. Kandeel(1993), “Faces and Facial Features Detection in Color Images,” Geometric Modeling and Imaging, pp. 209-214.

延伸閱讀