隨著網際網路的蓬勃發展，網路的安全機制也日趨重要，而如何建立一套有效且方便的個人身份辨識系統，成了當今一重要的課題。在過去，有許多的掌形辨識系統被提出，雖然利用掌形來辨識個人可以達到不錯的辨識率，但是卻有的使用上的不便性，主要的原因是，幾乎所有的掌形機都需要使用者將手掌放在固定的支架上，這造成了使用者的不方便，針對此一問題，我們利用一般的掃描器來取得手掌影像，使用者只需將手掌自然的張開擺放在平台上，完全沒有任何的固定支架，而依然可以作良好的身份確認。
在辨識的方法上，我們採用三種不同的方法，第一種為基本的「Template Matching」方法，再利用線性相關係數，來作為相似度比較的依據，利用這一種方式，我們可以達到90％的辨識率。第二種方法，我們採用「Back-propagation」倒傳式類神經網路來辨識掌紋，利用此一方法，我們可以達到94％的辨識率。雖然利用倒傳式演算法可以達到不錯的辨識率，但是確有收斂速度過慢的缺點，針對此一缺點，我們採用了「Scaled Conjugate Gradient」的演算法來改善此一現象，利用此一演算法，我們可以達到幾乎99.5％的辨識率。

In this thesis, we present a novel approach to identify people using their palm prints. We use the common scanner to capture the palm print images and don''t need any fixed pegs to fix the palm. In palm shape recognition machine, users have to put their hand on test plane with some fixed pegs, and this is unsuitable for some users like children. We solve this problem by our approach and make users easier to use the personal identification device.
We tried the template matching and neural network methods to verify the palm print images. In template matching method, we adopt the linear correlation function to measure the similarity between different palm print images. In this approach, we can achieve about 90% accuracy rate. In neural network approach, we use the standard backpropagation and the scaled conjugate gradient algorithms to train the network. In the backpropagation neural network experiment, we have above 94% accuracy rate. Although the backpropagation neural network has desirable performance the slow convergence is the fatal drawback. In order to produce a significant improvement in the convergence performance of a multilayer perceptron, we have to use high-order information in the training process. So we use the scaled conjugate gradient method on the multilayer perceptron network, and we achieve almost 99.5% accuracy rate.

[1] R. Clarke, "Human identification in information systems: Management challenges and public policy issues" Information Technology & People, Vol. 7, No. 4, pp. 6-37, 1994
[2] A. K. Jain, R. Bolle and S. Pankanti, BIOMETRICS Personal identification in network society, Kluwer Academic Publishers, 1999.
[3] R. P. Miller, "Finger dimension comparison identification system", US Patent, No.3576538, 1971.
[4] D. P. Sidlauskas, ŗD hand profile identification apparatus", US Patent, No. 4736203, 1988.
[5] A. K. Jain and N. Duta, "Deformable matching of hand shapes for verification", http://www.cse.msu.edu/~dutanico/
[6] C. C. Han, P. C. Chang and C. C. Hsu, "Personal identification using hand geometry and palm-print", in Proc. of Fourth Asian Conference on Computer Vision (ACCV), pp. 747-752, 2000.
[7] S. Y. Kung, S. H. Lin and M. Fang, "A neural network approach to face/palm recognition", in Proc. of 1995 International Conference on Neural Network, pp. 323-332, 1995.
[8] D. G. Joshi, Y. V. Rao, S. Kar, V. Kumar and R. Kumar, "Computer-vision-based approach to personal identification using finger crease pattern ", Pattern Recognition, Vol. 31, pp. 15-22, 1998.
[9] D. Zhang and W. Shu, "Two novel characteristics in palm print verification: datum point invariance and line feature matching", Pattern Recognition, Vol. 32, pp. 691-702, 1999.
[10] M. Sonka, V. Hlavac and R. Boyle, Image Processing, Analysis, and Machine Vision, PWS Publisher, 1999.
[11] S. Haykin, Neural Network, Prentice-Hall Publisher, 1999.
[12] M. F. Moller, "A Scaled Conjugate Gradient Algorithm for Fast Supervised Learning", Neural Networks, Vol. 6, pp. 525-533, 1993.