多媒體的資料盛行的今日，如何有效傳輸成為重要課題。除了擴大頻寬外，另一途徑就是壓縮檔案，因此不同的壓縮技術紛紛出籠。我們的工作重心放置在靜態影像的壓縮技術上，運用的方法稱為tree structure VQ(Vector quantization) ，這是屬於VQ中的一種方法，目的是提高壓縮的速度。首先，我們會利用一個以distortion為基礎的adaptive resonance theory 2（ART2）方法建構一個樹狀的編碼簿，接著我們運用較有效率的搜尋方法，利用這個編碼簿將原有的圖片取代成一連串的編號，接收到檔案的使用者，就可以利用編碼簿還有編號將原來的圖片還原。在我們的方法中，我們可以較有效率的建構一個樹狀的編碼簿，並利用這個編碼簿節省編碼所需的時間，達到較快的壓縮速率和較佳的壓縮效果。

In this paper, we propose a novel approach of vector quantization using a merge-based hierarchical neural network. Vector quantization（VQ）is known as a very useful technique for lossy data compression. Recently, Neural network（NN）algorithms have been used for VQ. Vlajic and Card proposed a modified adaptive resonance theory (modified ART2）[1] which is a constructing tree structure clustering method. However, modified ART2 has disadvantages of slow construction rate and constructing many redundant levels. Therefore, we propose a more efficient approach for constructing the tree in this paper. Our method establishes only those required levels without losing the fidelity of a compressed image.

摘要 i
Abstract ii
第一章簡介 1
1.1 VQ 的基本介紹 2
1.2 code-book 的區別 4
1.3 壓縮品質的評量 6
第二章VQ 的方法介紹 9
2.1 Enhanced LBG（ELBG） 10
2.2 Hierarchical SOM（HSOM） 14
2.3 Modified ART2 17
第三章我們的方法 21
3.1 樹狀code-book 的建立 22
3.2 單層的學習（one level learning） 24
3.3 編碼的方法（algorithm for encoding） 31
第四章模擬與分析 35
第五章總結 54
參考文獻 56

[1] Chang, C. and Shiue, C., “Tree Structured Vector Quantization with Dynamic Path Search,” in Proceedings of International Workshops on Parallel Processing, Sept. 1999

[2] Russo, M. and Patane, G., “Improving the LBG Algorithm,” in Lecture Notes in Computer Science. New York: springer-Verlag, pp.621-630, 1999.

[3] Russo, M. and Patane, G., “The enhanced LBG Algorithm,” Neural Networks, pp.1219-1237, 2001.

[4] Vlajic, N. and Card, C., “Vector quantization of images using modified adaptive resonance algorithm for hierarchical clustering,” in IEEE Transactions on Neural Networks, Vol.12, Sept. 2001.

[5] Gersho, A., and Gray, R. “Vector quantization and signal compression,” Boston: Kluwer, 1992.

[6] Sayood, K., “Introduction to Data Compression,” CA: Morgan Kaufmann, 2nd ed., 2000.

[7] Linde, Y., Buzo, A. and Gray, R., “An algorithm for vector quantization design,” in IEEE Transactions on Communications, pp.84-94, 1980.

[8] Lloyd, S., “Least squares quantization in PCMs ” , Murray Hill: Bell Telephone Laboratories paper.

[9] Balakrishnan, M., Pearlman, W. A. and Lu, L. ,“Variable Rate Tree-structure Vector Quantization,” in IEEE Transactions on Information Theory, pp.917-930, July 1995.

[10] Barbalho, M., Duarte, A., Neto, D., Costa, F. and Netto, A., “Hierarchical SOM applied to image compression,” in Proceedings of International Joint Conference on Neural Networks, pp.442 -447, July 2001.

[11] Karayiannis B., Pai P. and Zervos H., “Image compression based on fuzzy algorithms for learning vector quantization and wavelet image decomposition,” in IEEE Transactions on Image Processing, Vol.7, Aug. 1998.
[12] MacQueen J., “Some methods for classification and analysis of multivariate observations,” in proceedings of the fifth Berkeley symposium on Math. Stat. and prob., pp. 281-296, 1967.

[13] Forgey E., “Cluster analysis of multivariate data: efficiency vs. inter-pretability of classification.” Biometrics, 1965.

[14] Mukherjee D. and Mitra K., “Vector set partitioning with classified successive refinement VQ for embedded wavelet image coding,” in Proceedings of IEEE International Symposium on circuits and Systems, pp. 25-28, June 1998.

[15] Bayazit U. and Pearlman A., “Variable length constrained storage tree structure vector quantization,” in IEEE Transactions on Image Processing, Vol.8, March 1999.

[16]Koikkalainen P. and Oja E., “Self-organizing hierarchical feature maps,” in proceedings of International Joint Conference on Neural Networks, pp. 279 -284, June 1990.

[17] Vlajic N., Card C. and Kunz T., “Image-compression for wireless World Wide Web browsing: a neural network approach,” in proceeding of IEEE International Joint Conference on Neural Networks, Vol. 1, July 2000.

[18] Lee H. and Lee D., “A gain-shape vector quantizer for image coding,” in proceedings of IEEE International Conference on Acoustics, Speech, and Signal Processing, pp. 141-144, April 1986.

[19] Dong-Chul Park and Young-June Woo, “Weighted centroid neural network for edge preserving image compression” in IEEE Transactions on Neural Networks, Vol. 12, Sept. 2001.

[20] Ho Y.-S. and Gersho A., “Variable-rate multi-stage vector quantization for image coding,” in proceedings of International Conference on Acoustics, Speech, and Signal Processing, pp.1156-1159, 1988.

[21] Kusumoputro B., Widyanto R., Fanany I. and Budiarto H., “Improvement of artificial odor discrimination system using fuzzy-LVQ neural network,” in Proceeding of International Conference on Computational Intelligence and Multimedia Applications, pp. 474-478, Sept. 1999.

[22] Hammer B., Brandt A. and Schielein M., “Hierarchical encoding of image sequences using multistage vector quantization,” in proceedings of IEEE International Conference on Acoustics, Speech, and Signal Processing, pp.1055 -1058, Apr. 1987.

[23] Hung Yuen, Kam-Chi Li and Hanzo L., “Efficient variable rate vector quantization using quadtree segmentation,” in proceedings of IEEE International Symposium on Circuits and Systems, pp.1636-1639, 1995.