臺灣博碩士論文加值系統

骨架擷取技術在許多應用領域中被廣泛的採用。然而過去所經常使用的細線化(thinning)所產生的骨架，往往會多出許多不必要的小分支或在叉點處產生變形，造成在特徵擷取上的不穩定。因此，動態式自我組織神經網路被提出以擷取出更為完整及精確的骨架。但在實際的應用上可以發現，動態式自我組織神經網路的執行速度過於緩慢，且其在處理具有叉點及迴圈點時容易產生骨架變形。因此，本研究提出利用k-curvature演算法來估測特徵點，及利用區域劃分來處理骨架點的連結以加速骨架擷取的速度。特徵點的估測，可以讓輸出神經元一開始便落在適當的骨架點上，使得網路在學習時不必設定太大的鄰域範圍；區域劃分則可以讓已形成的骨架點在連結時減少錯誤。在實驗部份，本研究以工具、阿拉伯數字、英文字母及漢字等圖樣作為實驗對象。從實驗結果可以發現，本研究所提出的方法可以有效的加快動態式神經網路在骨架擷取的速度，且其所擷取出的骨架較不易產生叉點變形的問題。

Skeleton shape extraction technique was widely adopted in many application such as object modeling、character recognition、machine vision and computer animation .The thinning process always be used in skeleton extraction, but it often distorts the local shapes of an observed pattern. This is an inherent defect for all thinning algorithms. Therefore, a Dynamic Self-Organizing Map(DSOM) was proposed to extract skeleton precisely. But the process speed of DSOM is too slow. In order to speed up the process speed, the methods of feature-points estimation and area segmentation were proposed in this paper. Feature-points estimation process estimates the position of output neurons instead of random decision. The area segmentation process avoids the fault when DSOM used in crossing or loop pattern. In the experimental results, the tools、alphanumeric and Chinese characters are used as test patterns. The results show that the proposed methods improve the process speed of DSOM and it wouldn’t distort the local shapes of an observed pattern.

摘要 iv
英文摘要 v
誌謝 vi
目次 vii
表目錄 viii
圖目錄 x
第一章 緒論 1
1.1 研究動機 1
1.2 平行式細線化演算法介紹(Zhang&Suen) 1
1.3 研究方法 4
1.4 論文章節 4
第二章 系統架構 5
第三章 自我組織特徵映射圖網路 7
3.1 相似性量測 7
3.2 競爭式學習網路 9
3.3 自我組織特徵映射圖網路架構 11
3.4 自我組織特徵映射圖網路演算法 14
第四章 前處理 16
4.1 輸出神經元的估測方法 16
4.1.1 隨機估測 17
4.1.2 特徵點估測 18
4.2 邊緣偵測 20
4.3 影像分割 22
4.4 k-curvature演算法 24
第五章 動態式自我組織神經網路 27
5.1 區域劃分 27
5.2 動態式自我組織神經網路演算法 29
第六章 實驗結果 33
6.1 針對工具的實驗 33
6.2 針對阿拉伯數字的實驗 35
6.3 針對英文字母的實驗 36
6.4 針對漢字的實驗 39
6.5 與其它演算法的比較 44
6.5.1.1 擷取速度的比較(與HMSOM) 44
6.5.1.2 骨架品質的比較(與HMSOM) 45
6.5.2.1 擷取速度的比較(與Zhang&Suen演算法) 49
6.5.2.2 骨架品質的比較(與Zhang&Suen演算法) 49
第七章 結論與未來展望 51
參考文獻 52

1.Q. Z. Ye and P. E. Danielsson, “Inspection of Printed Circuit Boards by Connectivity Preserving Shrinking,” IEEE Trans. on Pattern Analysis and Machine Intelligence, Vol. 10, no. 5, Sept. 1988, pp. 737-742.
2.H. Blum and R. N. Nagel, “Shape Description Using Weighted Symmetric Axis Features ,” Pattern Recognition, Vol. 10, 1967, pp.167-180.
3.M. M. Altuwaijri and M. A. Bayoumi, “A Thinning Algorithm for Arabic Characters Using ART2 Neural Network,” IEEE Trans. on Circuits and Systems-II：Analog and Digital Signal Processing, Vol. 45, No. 2, Feb. 1998, pp. 260-264.
4.T. Y. Zhang and C. Y. Suen, “A Fast Parallel Algorithm for Thinning Digital Pictures,” Communications of the ACM archive, Vol. 27, No. 3, Mar. 1984, pp.236-239.
5.Y. Y. Zhang and P. S. P. Wang, “A Modified Parallel Thinning Algorithm,” IEEE 9th Int. Conf. on Pattern Recognition, Vol. 2, Nov. 1988, pp.1023-1025.
6.L. Lam, S. W. Lee and C. Y. Suen, “Thinning Methodologies-a Comprehensive Survey,” IEEE Trans. on Pattern Analysis and Machine Intelligence, Vol. 14, No. 9, Sept. 1992, pp.869-885
7.T. Kohonen, “The Self-Organizing Map,” Proceedings of the IEEE, Vol. 78, No. 9, 1990, pp.1464-1480.
8.J. A. Kangas, T. Kohonen and J. T. Laaksonen “Variants of Self-Organizing Maps,” IEEE Trans. on Neural Network, Vol. 1, No. 1, Oct. 1991, pp.93-99.
9.B. Fritzke, “Let It Grow-Self-Organizing Feature Maps with Problem Dependent Cell Structure,” Artificial Neural Networks, North-Holland, Amsterdam, Vol. 1, pp. 403-408.
10.A. Datta and S. K. Parui, “Skeletons From Dot Patterns：A Neural Network Approach,” Pattern Recognition Letters, Vol. 18, 1997, pp.335-342.
11.A. Datta, S. K. Parui and B. B. Chaudhuri, “Skeletal Shape Extraction From Dot Patterns by Self-Organizing Maps,” in Proc. IEEE 13th Int. Conf. on Pattern Recognition, Vol. 4, Aug. 1996, pp.80-84.
12.J. F. Canny, “A Computational Approach to Edge Detection ,” IEEE Trans. Pattern Anal. Machine Intell., Vol. PAMI-8, Aug. 1986, pp.679-698.
13.V. S. Nalwa and T. O. Binford, “On Detecting Edges,” IEEE Trans. Pattern Anal. Machine Intell., Vol. PAMI-8, June. 1986, pp.699-714.
14.T. Bayer, U. Kressel and M. Hammelsbeck, “Segmenting Merged Characters,” Proc. 11th Intll. Conf. Pattern Recogniton, Vol. 2, conf. B：Pattern Recognition, Methodology and Systems, 1992, pp.346-349.
15.R. G. Casey and E. Lecolinet “A Suvey of Methods and Strategies in Character Segmentation,” IEEE Trans. on Pattern Analysis and Machine Intll., Vol. 18, No. 7, July. 1996, pp.690-706.
16.K. Dunkelberger and O. Mitchell “Contour Tracing for Precision Measurement,” IEEE Inter. Conf. on Robotic and Automation, Vol. 2, Mar. 1985, pp.22-27.
17.J. E. Byun and T. Nagata, “Determination of 3-D pose of a flexible object by stereo matching of curvature representations,” Proceeding of IEEE on Intell. Robots and Systems, Vol. 3, Sept. 1994, pp.1992-1999.
18.R. Singh, V. Cherkassky and N. Papanikolopoulos, “Self-Organizing Maps for the Skeletonization of Sparse Shapes,” IEEE Trans. on Neural Network, Vol.11, No. 1, January. 2000, pp.241-248.
19.林昇甫、洪成安，神經網路入門與圖樣辨識，全華科技圖書股份有限公司，民85年。
20.謝明勳、M. N. Shirazi、張鴻德，應用階層式修正型自我組織特徵映射圖網路(HMSOM)於骨架擷取，南台科技大學電子工程研究所碩士論文，民93年。