摘要(英) |
In this thesis, three algorithms based on Nearest Feature Line Embedding (NFLE) including Fuzzy NFLE (FNFLE), Kernel NFLE (KNFLE),and the proposed Kernel Fuzzy NFLE (KFNFLE) methods are implemented to demonstrate their effectiveness applying on face recognition.
The motivation of this study relies main by on the fact that both Kernel method and Fuzzy model can enhance the transformed feature space. Therefore, Kernel method and Fuzzy model are both considered in the proposed KFNFLE to further improve the performance of original NFLE. Firstly, the training faces are transformed by applying PCA method. Then, the transformed matrixes based on FNFLE, KNFLE,
and KFNFLE are obtained, respectively. Next, the prototype and testing samples are projected onto the feature space via the obtained transformed matrixes. Last, nearest neighbor method is applied for matching.
In the experiments, the CMU face database, and our real-case face database are utilized to evaluate the performance of the proposed KFNFLE method. Experimental results demonstrate that the performance of the proposed KFNFLE is superior to FNFLE, KNFLE, and original NFLE when the sample size increases. |
參考文獻 |
[1] Window Home Security. Home security systems for a peaceful life. 取自http://www.windowhomesecurity.com/?p=30, 2013.
[2] Bu Linda. House alarm systems tips & guide. 取自http:
//homesecuritysystemsreview-s.com/house-alarm-systems,
2011.
[3] Yoav Freund and Robert E. Schapire. ”a decision-theoretic generalization
of on-line learning and an application to boosting”. In
European Conference on Computational Learning Theory, pages 23–37, 1995.
[4] H. C. Wang and L. M. Zhang. ”a novel fast training algorithm for
adaboost”. Journal of Fudan University (Natural Science), 43(1):27–33, 2004.
[5] 鄭文昌、詹定懋.「多種特徵抽取方法比較與結合之adaboost 行人
偵測」. Proceedings of the Conference on the Information Education and Technological Applications, 2011.
[6] K. Pearson. ”on lines and planes of closest fit to systems of points in space”. Philosophical Magazine, 2(6):559–572, 1901.
[7] S. Z. Li. ”nearest feature line”. Scholarpedia, 3(3):4357, 1999.
[8] Li. Zhao, Wei. Qi, S. Z. Li, S. Q. Yang, and H. J. Zhang. ”contentbased retrieval of video shot using the improved nearest feature line method”. Chinese Journal of Software, 13(4):586–590, April 2002.
[9] Y. N. Chen. ”Face Recognition Using Nearest Feature Space Embedding”.PhD thesis, National Central University, 2011.
[10] L. A. Zadeh. ”fuzzy sets”. Information And Control, 8:338–353, 1965.
[11] M. Aizermann, E. Braverman, and L. Ronzonoer. ”theoretical foundations of the potential function method in pattern recognition
learning”. Automatika i Telemekhanika, 25:147–169, 1964.
[12] Bernhard E. Boser, Isabelle M. Guyon, and Vladimir N. Vapnik. ”a training algorithm for optimal margin classifiers”. In Proceedings of the 5th Annual ACM Workshop on Computational Learning Theory, pages 144–152. ACM Press, 1992.
[13] J. Mercer. ”functions of positive and negative type, and their connection with the theory of integral equations”. Philosophical Transactions of the Royal Society, London, 209:415–446, 1909.
[14] César R. Souza. Kernel functions for machine learning
applications. 取自http://crsouza.blogspot.com/2010/03/
kernel-functions-for-machine-learning.html, 2010.
[15] Davis Nello Cristianini University of California. Kernel methods for general pattern analysis. 取自http://www.kernel-methods.net/tutorials/KMtalk.pdf.
[16] 陳士杰. 支持向量機基礎. 取自http://sjchen.im.nuu.edu.tw/
MachineLearning/final/CLS_SVM.pdf, 2005.
[17] Adam Jówik. ”a learning scheme for a fuzzy k-nn rule”. Pattern Recogn. Lett., 1(5-6):287–289, July 1983.
[18] J. M. Keller, M. R. Gray, and Jr. ”a fuzzy k-nearest neighbor algorithm”. IEEE Transactions on Systems, Man, and Cybernetics, 15:580–585, 1985.
[19] CMU Face Group. Frontal and profile face databases. 取自http://vasc.ri.cmu.edu/idb/html/face/facial_expression, 2009. |