Please use this identifier to cite or link to this item: http://hdl.handle.net/11455/4855
標題: 改善傅立葉分析在人臉識別上做陰影補償的應用
Improving the Shadow Compensation Using Fourier Analysis With Application to Face Recognition
作者: 蕭淑美
Shiau, Shu-Mei
關鍵字: Face recognition;人臉識別;Fourier transforms;illumination variation;illumination direction;shadow compensation;傅立葉轉換;照明變化;光照方向;陰影補償
出版社: 通訊工程研究所
引用: [1]R. C.Gonzalez. , R. E.Woods and S. L.Eddins , 繆紹綱 譯,Digital Image Processing Using MATLAB,1st Edition數位影像處理-運用MATLAB,台灣培生教育出版,初版一刷,2008。 [2]Gonzalez,Woods , 繆紹綱 譯,Digital Image Processing 3/e, 數位影像處理台灣培生教育出版,初版三刷,2009。 [3]何承諭,“三維離散X射線轉換實現之改良 ”,國立中興大學碩士論文,2011。 [4]連國珍 著作,數位影像處理,儒林圖書,五版一刷,2008。 [5]盧俊良,“基於光線與臉部表情變化之下人臉辨識”,國立中央大學資訊工程研究所,2009。 [6]S.I. Choi , and G.M. Jeong,“Shadow Compensation Using Fourier Analysis With Application to Face Recognition,”IEEE Signal Process. Lett.,vol. 18, pp.23-26,2011. [7]X. Zou, J. Kittler, and K. Messer, “Illumination invariant face recognition: Asurvey,” in IEEE Int. Conf. on Biometrics: Theory, Applications, and Systems, pp. 1–8, 2007. [8]R. Javier and Q. Julio, “Illumination compensation and normalization in eigenspace-based face recognition: A comparative study of different pre-processing approaches,” Pattern Recognit. Lett. , vol. 29, pp.1966–1979, 2008. [9]T. Zhang, Y. Y. Tang, B. Fang, Z. Shang, and X. Liu, “Face recognition under varying illumination using gradient faces,” IEEE Trans. Image Process., vol. 18, pp. 2599–2606, 2009. [10]A.S. Georghiades and P. N. Belhumeur, “Frome few to many: Illumination cone models for face recognition under variable lighting and pose,” IEEE Trans. Pattern Anal. Mach. Intell., vol. 23, pp.643–660, 2001. [11]R..Ramamoorthi, “Analytic PCA construction for theoretical analysis of lighting variability in images of a Lambertian object,”IEEE Trans.Pattern Anal. Mach. Intell., vol. 24, pp. 1322–1333, 2002. [12]S.-I. Choi and C.-H. Choi, “An effective face recognition under illumination and pose variations,” in Proc. Int. Joint Conf. Neural Networks, pp. 914–919,2007. [13]H.Wang, S. Li, and Y.Wang, “Face recognition under varying lighting conditions using self quotient image,” in Proc. IEEE Int. Conf.Automatic Face and Gesture Recognition, pp. 819–824, 2004. [14]X. Xie and K.-M. Lam, “An efficient illumination normalization method for face recognition, ”Pattern Recognit . Lett., vol. 27, pp.609–617, 2006. [15]T. Ahonen , A. Hadid , and M. Pietikainen , “Face description with localbinary patterns: Application to face recognition,” IEEE Trans. Pattern Anal. Mach. Intell. , vol. 28, pp. 2037–2041, 2006. [16]A.V. Oppenheim, A. S.Willsky , and S. H. Nawab , Signal and Systems,2nd ed. Upper Saddle River, NJ: Prentice-Hall, pp. 303–304, 1996. [17]R. C. Gonzales and R. E.Woods , Digital Image Processing, 2nd ed. Upper Saddle River, NJ: Prentice-Hall, 2002.
摘要: 
由於光照的變化,使得人臉產生陰影,改變了臉部的外觀,並且降低臉部識別的特性。因此,我們提出一個新的陰影補償方式,是以傅立葉分析為基礎來處理光照對於改變人臉辨識度的影響。而我們採用改良由傅立葉分析在人臉識別上做陰影補償的方式。
首先,我們提出先將原始影像根據光照的方向進行分類:光線從左邊照射、均勻照射和右邊照射,再依照人臉上陰影覆蓋區域的多寡做第二次分類:1/2與2/3區域被陰影遮蓋。在傅立葉轉換後,藉由將原始影像的幅度與輔助的頻譜幅度作一個比例的相加,再和原先保留的相位結合,來重建被陰影所惡化的人臉圖像。而每一個類別都有各自一組輔助頻譜幅度與影像原本幅度的相加比例,並依照類別使用不同相加比例,作為補償陰影的方式。
在光照的變動之下,由於不同的光照方向與人臉上被陰影覆蓋的比例不定,都會影響影像恢復的成效。因此,藉由先分類光照方向再由輔助的頻譜幅度補償傅立葉轉換後的扭曲頻譜幅度。而我們所提出的方法不僅可提高人臉辨識度與改善覆蓋陰影的區塊還能改善原始的方法。然而,實驗結果與分析明顯呈現我們改良的方法,在光照變化之下,比原始方法更能大幅提升識別的性能。

Shadows that occur on face images due to illumination variation can change the appearance of a face and degrade face recognition performance. In this thesis, we implement and improve a new shadow compensation method based on the Fourier analysis for handling illumination variation.

First, we classify the original images into three categories according to its illumination direction: (1) illumination from left, (2) uniform illumination, and (3) illumination from right. The classified images are then further classified into two sub-categories according to the area on the face that is covered by shadow: (1) 1/2 area and (2) 2/3 area.

Second, after Fourier transform, we adjust the proportion between the auxiliary magnitude and the magnitude of the original image according to the classified category. The magnitude spectrum of the restored image is the sum of the auxiliary magnitude and the magnitude of the original image and its phase is the original phase components.

The proposed method improved the shadow compensation as compared to the previous approach because of its adaptability on illumination direction. The experimental results and analysis show that the performance of our proposed method is indeed better than the original method significantly.
URI: http://hdl.handle.net/11455/4855
其他識別: U0005-0207201211411100
Appears in Collections:通訊工程研究所

Show full item record
 

Google ScholarTM

Check


Items in DSpace are protected by copyright, with all rights reserved, unless otherwise indicated.