臺灣博碩士論文加值系統

彩色紋路分析及分割在影像處理應用上是一個重要的主題，它普遍應用在不同領域中例如醫學診斷，影像的索引以及影像搜尋等。因此在本篇論文我們提出了植基於紋路特徵編碼法的彩色影像分割及紋路分析的相關技術。紋路特徵編碼是轉換原始灰階影像到紋路特徵影像的編碼法則。其中紋路特徵影像中的像素值，稱之為紋路特徵數，可以視為原始影像對應像素的梯度值。
本論文包含了三個部份，彩色影像分割，紋路影像的週期性及方向性檢定。第一部份我們結合模糊彩色統計圖及紋路特徵數統計圖成為一個特徵分佈，利用此一特徵分佈去獲得在影像分割過程中不同區域間的相似性量測並據以分割不同的紋路區域。在此所使用的影像分割方法稱之為階層式分裂及合併演算法；它包含了階層式分裂、凝聚式合併以及逐點分類。在第二部份主要是探討如何獲得紋理影像的週期性，在這裡我們藉由紋路特徵數的自相關函數去計算總體週期性。在最後一個部份，我們使用紋路特徵影像的複雜曲線去得到紋理影像的方向性。從不同種類的彩色紋理影像的實驗結果顯示我們所提出的方法可優於其他傳統的方法。

The color texture analysis and segmentation are important topics of image processing and applications. It is widely used in different areas such as medical diagnosis, image index and retrieval. Therefore we propose a few techniques of the color image segmentation and texture analysis based on texture feature coding method in this thesis. The texture feature coding method is coding scheme that transform the original gray-level image into the texture feature image whose value of pixel, called the texture feature number, can be regarded as gradient of corresponding pixel of original image.
The contents of the thesis consist of three parts that are color image segmentation, the detections of periodicity and directionality of the texture image. In the first part we conjoin the two feature distributions that are the fuzzy color histogram and the texture feature number histogram to derive homogeneity measure for separating the different regions of color image under the segmentation process. The segmentation method used herein is the hierarchical splitting and merging algorithm that includes the hierarchical splitting, agglomerative merging and the pixel-wise classification. The issue of second part is how to detect the periodicity of textured image. The auto-correlation of the corresponding the texture feature number of texture feature image is computed to obtain the overall periodicity of image. In the final part we explore the detection of the directionality of the texture image by using the complexity curve of the texture feature image. Experimental results of conducting with different kinds of color texture images reveal that the proposed methods are superior to the ones of other traditional methods.

第一章、緒論
第一節、前言．．．．．．．．．．．．．．．．．．．．．．1
第二節、研究動機與目的．．．．．．．．．．．．．．．．．4
第三節、研究方法概論．．．．．．．．．．．．．．．．．．5
第四節、章節概述．．．．．．．．．．．．．．．．．．．．6
第二章、相關理論與背景知識
第一節、前言．．．．．．．．．．．．．．．．．．．．．．8
第二節、顏色空間．．．．．．．．．．．．．．．．．．．．8
第三節、模糊顏色統計圖．．．．．．．．．．．．．．．．．15
第四節、使用特徵分佈的彩色紋路分割．．．．．．．．．．．18
第五節、紋路週期性．．．．．．．．．．．．．．．．．．．28
第六節、紋路方向性．．．．．．．．．．．．．．．．．．．35
第三章、以紋路特徵編碼法為基礎之影像處理
第一節、前言．．．．．．．．．．．．．．．．．．．．．．42
第二節、研究流程．．．．．．．．．．．．．．．．．．．．42
第三節、紋路特徵編碼法．．．．．．．．．．．．．．．．．44
第四節、以紋路特徵編碼法為基礎的影像分割法．．．．．．．55
第五節、以紋路特徵編碼法為基礎的紋路週期性判定．．．．．62
第六節、以紋路特徵編碼法為基礎的紋路方向性判定．．．．．67
第四章、實驗
第一節、前言．．．．．．．．．．．．．．．．．．．．．．73
第二節、以紋路特徵編碼法為基礎的影像分割法實驗．．．．．73
第三節、以紋路特徵編碼法為基礎的紋路週期性判定實驗．．．86
第四節、以紋路特徵編碼法為基礎的紋路方向性判定實驗．．．89
第五章、結論．．．．．．．．．．．．．．．．．．．．．．．94

[01]Ming-Huwi Horng,Yung-Nien Sun,Xi-Zhang Lin,”Texture feature coding method for classification of liver sonography”,Computerized Medical Imaging and Graphics 26,33-42,2002.
[02]C.-C. Wong, C.-C. Chen and S.-L. Yeh, "K-means-based fuzzy classifier design," Proc. IEEE Int. Conf. Fuzzy Systems, vol. 1, 48-49, 2000.
[03]M.Stricker and M.Orengo,”Similarity of Color Images”,SPIE vol.2420, No.1, 1995
[04]Ju Han, Kai-Kuang Ma, Fuzzy color histogram and its use in color image retrieval, IEEE Transactions on Image Processing, Aug. (2002), Vol. 45, No.1, p19-p23.
[05]Kan-Min Chen, Shu-Yuan Chen, Color texture segmentation using feature distributions, Pattern Recognition Letters 23 (2002), p755-p771.
[06]Jia-Guu Leu, On indexing the periodicity of image textures, Image and Vision Computing 19 (2001), p987-p1000.
[07]S.Baheerathan, F.Albregtsen, H.E.Danielsen, New texture features based on the complexity curve, Pattern Recognition 32 (1999), p.605-p.618.
[08]Dunn,J.C., A fuzzy relative of the ISODATA process and its use in detecting compact, well separated clusters,J.Cybernet 3, p32-p57.
[09]J. Bezdek. Pattern Recognition with Fuzzy Objective Function Algorithms. Plenum Press, New York, 1981.
[10]C.S. T. Choy and W.C. Siu,“A class of competitive learning models which avoids neuron underutilization problem,” IEEE Trans. Neural Networks, vol. 9, no. 6, pp. 1258-1269, 1998.
[11]W.C. Chu and N.K. Bose,“Vector quantization of neural networks,” IEEE Trans. Neural Networks, vol. 9, no. 6, pp. 1235-1245, 1998.
[12]D. Lee, S. Baek and K. Sung,“Modified k-means algorithm for vector quantizer design,” IEEE Signal Processing Letters, vol. 4, no. 1, pp. 2-4, 1997.
[13]P. Scheunders, “A comparison of clustering algorithms applied to color image quantization,” Pattern Recognition Letters, vol. 18, pp. 1379-1384, 1997.
[14]P. Scheunders, “A genetic c-means clustering algorithm applied to color image quantization,” Pattern Recognition, vol. 30, no. 6, pp. 859-866, 1997.
[15]T. Uchiyama and M.A. Arbib, “Color image segmentation using competitive learning,” IEEE Trans. Pattern Analysis and Machine Intelligence, vol. 12, no. 16, pp. 1197-1206, 1994.
[16]Bajcsy, R., and Lieberman, L., Texture gradient as a depth cue, Computer Graphics and Image Processing, Vol.5, No.1, pp52-67.
[17]Witkin, A.P., Recovering surface shape and orientation from texture, Artificial Intelligence, Vol.17, No.13, pp.17-45.
[18]Kender, J.R., Shape from texture: An aggregation transform, that maps a class of textures into surface orientation, IJCAI-79, Proceedings of the Int. Joint. Conf. On Artificial Intelligence, Tokyo, Aug:20-23, 1979, pp.475-480.
[19]PATRICK C. CHEN,THEODOSIOS PAVLIDIS, Segmentation by Texture Using Correlation, IEEE Transactions on pattern analysis and machine intelligence, vol. PAMI-5, No1, 1983, p64-p69.