基于分形編碼和奇異值分解的混合人臉識別方法
A Hybrid Method for Face Recognition Based on Fractal Coding and Singular Value Decomposition
-
摘要: 該文提出了一種新的基于分形編碼的人臉識別方法。在分形近鄰距離的基礎(chǔ)上，提出了分形奇異值近鄰距離，并把分形編碼和局部奇異值分解結(jié)合起來，提高了識別率。實驗結(jié)果表明，與僅僅使用分形近鄰距離相比，該算法對光照變化、表情和姿態(tài)變化具有更大的容忍度，而且訓(xùn)練時間短，識別率高。
-
關(guān)鍵詞:
- 人臉識別;分形編碼;局部奇異值分解;分形奇異值近鄰距離
Abstract: In this paper, a new face recognition method based on fractal coding is proposed. Fractal singular value neighbor distance is brought forward based on fractal neighbor distance. Fractal coding and local singular value decomposition are used to improve the recognition rate. The experimental results show that the method can keep a good robustness to the variation of illumination, pose and expression, compared with fractal neighbor distances. Furthermore, the method shows that the training time is short and recognition rate is high. -
Barnsley M F. Fractal Everywhere, 2nd edition[M]. New York:Academic Press, 1995:1 - 60.[2]Keller J M, Chen S, Crownover R M. Texture description and segmentation through fractal geometry[J].Computer Vision,Graphics andImageProcessing.1989, 45(2):150-[3]Jacquin A E. Image coding based on a fractal theory of iterated contractive image transformations[J].IEEE Trans. on Image Processing.1992, 1(1):18-[4]Fisher Y. Fractal Image Compression: Theory and Application[M].New York: Springer-Verlag, 1995:55 - 77.[5]Teewoon T, Hong Y. The fractal neighbor distance measure[J].Pattern Recognition.2002, 35(6):1371-[6]Ebrahimpour-Komleh H, Chandran V, Sridharan S. Face recognition using fractal codes[A]. IEEE International Conference on Image Processing[C], Thessaloniki, Greece, 2001:58 - 61.[7]Kouzani A, He F, Sammut K. Fractal face representation and recognition[A]. IEEE International Conference on Systems, Man and Cybernetics[C], Orlando, USA, 1997:1609 - 1613.[8]劉小軍,王東峰,鄒謀炎.一種基于奇異值分解和隱馬爾可夫模型的人臉識別方法[J].計算機(jī)學(xué)報,2003,26(3):340-344.[9]Hong Z Q. Algebraic feature extraction of image recognition[J].Pattern Recognition.1991,24(3):211- -
計量
- 文章訪問數(shù): 2190
- HTML全文瀏覽量: 97
- PDF下載量: 632
- 被引次數(shù): 0
下載:
