本論文提出一個新穎的布種紋理與印花織物的自動化色彩分析系統，此系統能對布種紋理和印花織物影像進行自動化分色及自動化辨識。本系統是以掃描器進行影像的取得，所使用到的彩色格式有光色與亮度(LAB)及色調飽和度亮度(HSI)兩種。
在自動化辨識布種紋理系統分析上，由於布的織物組織會在織物的表面形成紗線起伏的週期性結構，分析織物影像時，利用光線反射不一在布疋表面上形成紋理，藉以分析織物的紋理組織，因此利用機器視覺系統作為線上品管檢測之工具，不但可減少人工成本的支出，免除長久工作下來人為的疏失，更可在短時間內取得正確之檢驗數據，進行生產及檢驗之快速修改，提高產品品質，因此本研究是利用非監督式類神經網路達成自動化分辨織物的布種及其主要織物組織類型，首先利用LAB 色彩模型取得特徵值及小波轉換顯示出織物影像的紋理結構,再以共生矩陣求出角二次矩、熵、對比、相關值等紋理結構特徵值, 最後採用自組織映射圖網路為分類器。
在自動化印花織物分色系統分析上，由於印花布不像梭織布能以色紗排列加上織物組織來精確地描述彩色圖案，所以印花布只能以顏色數、顏色值及圖案來描述印花布的彩色圖案，加上印花布圖案的變化多樣，使得印花布的彩色圖案很難用文字或較簡單的方法來加以描述，這使得印花布在分析、辨識與比對上都得以人工方式進行，這造成自動化上的困難，因此本研究是以兩種方式分析，第一種方式利用小波轉換來縮小織物影像減少分色的運算負荷，並完整保留原始影像的印花結構和色彩的分佈，再以LAB 色彩模型 取得色彩特徵值，最後採用自組織映射圖網路為分色演算；另一方式利用將RGB彩色影像經過中值濾波的影像處理，可降低因為織布表面的織造結構引起的光線反射不一所造成的顏色變化，再將RGB 色彩模型轉換 HIS 色彩空間 ，使得色彩的分析與表達更能配合人類的色彩感覺及使用習慣，接著利用基因演算法搜尋出與印花織物原影像相同的色彩分佈，此子影像的面積是原影像的9.06%，不但可以減少分色的運算負荷，並完整保留原始影像的印花結構和色彩的分佈，再以HIS色彩空間取得色彩特徵值，最後採用自組織映射圖網路為分色演算。
實驗結果顯示本研究能自動化準確分類出織物的布種類(包含梭織物, 針織物及非織物)及其主要的織物組織類型(平紋、斜紋、緞紋、 單針織物、雙針織物、不織布)及自動化精確且快速分色印花織物。

This dissertation proposes a novel automatic color analysis system for printed fabrics and texture fabric natures that can automatically recognizing color texture and automatically make color separation. This system uses a scanner to obtain color images. There are two types of color models, LAB (light, color, lightness) and HSI (hue, saturation, intensity).
Automatic recognizing system for color texture fabric nature, the fabric texture will evolve into the periodical structure indicating the yarn variation on the fabric surface. In analyzing the fabric image, the texture will be formed on the fabric surface as a result of the variation in light reflection. So the texture of the fabric can be analyzed to gain an understanding of the method of weaving fabric and the tissue. Therefore Machine Vision System can not only reduce the labor cost but also avoid the careless mistake committed by the workers who have been in service for a long time. Additionally, it can ensure the acquisition of correct detection data within a short period of time, and the swift modification to the production and inspection programs so as to enhance the product quality. The method of recognizing color texture brought forth in the present study is to employ unsupervised learning network to automatically recognize the fabric type and the main texture types. Firstly, CIE-LAB Color Model is taken to obtain the feature value and wavelet transform is utilized to display the texture of the fabric image. Then co-occurrence matrices is employed to figure out the feature values of the texture structure such as Angular second moment, Entropy, Homogeneity, Contrast. Finally, Self-Organizing Map Network (SOMN) is used as the classifier.
Automatic Color Separating System for Printed Fabric, we can only use color numbers, color values and design to describe the color pattern of printed fabrics, which is different from woven fabrics with yarn disposition and texture as pattern determinants. Since most printed fabrics contain many different patterns nowadays, we need more than words and simple methods to describe the color patterns. The complication in pattern identification has made the analysis and comparison difficult and will have to be conducted manually; this system can be operated by two ways. One is conducted by the system first uses wavelet transformation to minify the fabric image to reduce the calculation load of color separation and also reserve the printing structure and color distribution of the original image. It also uses LAB color model to acquire characteristic value of the colors and the SOMN to conduct color separation; another the system first uses a color scanner to record RGB color images of the printed fabrics and uses median filter processing to reduce color changes due to uneven light reflections arising from the fabric surface weaving texture. Then RGB Color Space is transformed to HSI Color Space so that color analysis can match human color sense and use customary procedures. Next, the GA is employed to search for color distributions that are the same as the original image of printed fabrics. The area of each sub-image is 9.06% of the original image, not only reducing color segmentation operation time, but completely reserving the print structure and color distribution of the original image. Afterwards, color eigenvalues are obtained in HSI Color Space. Finally SOMN is adopted for the color segmentation operation.
According to our experimental results, this system automatically and accurately classify the fabric types (including shuttle-woven fabric, knitted fabric and non-woven fabric) and main texture type of the fabric(such as plain weave, twill weave, satin weave, single jersey, double jersey and non-woven fabric )and rapidly and automatically complete color separation and identify repeating patterns for printed fabrics’ images.

1. Lung, H. J., Application and Development of Automatic Color Separating System for Print, Journal of the China Textile Institute, 2(4), 24-31 (1992).
2. Lung, H. J., Development of Automatic Color separating System for Print, Journal of the China Textile Institute, 4(2), 111-115 (1994).
3. Huang, C. C., Liu, S. C., and Yu, W. H., Woven Fabric Analysis by Image Processing Part Ⅰ: Identification of Weave Patterns, Textile Res. J. 70(6), 481-485(2000).
4. Huang, C. C., Liu, S. C., and Yu, W. H., Woven Fabric Analysis by Image Processing Part Ⅱ: Identification of Weave Patterns, Textile Res. J. 70(6), 481-485(2001).
5. Jeon, B. S., Bae, j. H., and Suh, M. W., Automatic Recognition of Woven Fabric Patterns by an Artificical Neural Network, Textile Res. J. 73(7), 645-650(2003).
6. Kang, T. J., Choi, S. H., and Kin, S. M., Automatic Recognition of Fabric Weave Patterns by Digital Image Analysis, Textile Res. J. 69(2), 77-83(1999).
7. Ravandi, S. A. H., and Toriumi, K., Fourier Transform Analysis of Plain Weave Fabric Appearance, Textile Res. J. 65(11), 676-683(1995).
8. Salari, E., and Ling, Z., Texture Segmentation using hierarchical wavelet transform, Industrial Electronic, 1, 216-220(1995).
9. Jeffrey Kuo, C. F., and Tsai, C. C., Automatic Recognition of Fabric Nature by Using the Approach of Texture Analysis, Textile Research Journal, 76(5), 375-382 (2006).
10. Plataniotis, K. N., and Venetsanopoulos, A. N., Color Image Processing and Applications, Springer, New York, 239-246 (2000).
11. Liu, J., and Yang, Y. H., Multiresolution Color Image Segmentation, IEEE Trans. Pattern Anal. Mach. Intell., 16, 689-700 (1994).
12. More, J.-M., and Solimini, S., Variational Methods in Image Segmentation: With Seven image Processing Experiments, Birkhäuser, Boston, 21-34 (1995).
13. Gauch, J., and Hsia, C., A Comparison of Three Color Image Segmentation Algorithms in Four Color Spaces, Visual Communications and Image Processing, 1818, 1168-1181 (1992).
14. Fan, J., Aref, W. G.., Hacid, M. S., and Elmagarmid, A. K., An improved automatic isotropic color edge detection technique, Pattern Recognition Letters, 22, 1419-1429 (2001).
15. Kuo, C.-F. J., Su, T. L., and Huang, Y.-J., Computerized Color Separation System for Printed Fabrics by Using Backward-Propagation Neural Network, Fibers and Polymers, 8(5), 529- 536 (2007).
16. Shih, C. Y., Hsu, M., S., and Hsu, K., L., Color Pattern Analysis of Printed Fabric by Fuzzy Clustering Method, Journal of the China Textile Institute, 21(1), 69-80 (2003).
17. Kuo, C.-F. J., Shih, C. Y., Kao, C. Y., and Lee, J. Y., Color and Pattern Analysis for Printed Fabric by Unsupervised Clustering Method, Textile Res. J., 75(1), 9-12 (2005).
18. Kuo, C.-F. J., Shih, C. Y., and Lee, J. Y., Repeat Pattern Segmentation of Printed Fabrics by the Hough Transform Method, Textile Res. J., 75(11), 779-783 (2005).
19. Kuo, C.-F. J., Shih, C. Y., and Lee, J. Y., Automatic Recognition of Fabric Weave Patterns by Fuzzy C-Means Clustering Method, Textile Res. J., 74(2), 107-111 (2004).
20. Gonzalez, R. C., and Woods, R. E., Digial Image Processing, 2nd Edition, Prentice-Hall, New York, 2002.
21. Nixon, M. S., and Aguado, A. S., Feature Extraction and Image Processing, Newnes, Oxford, 89-95, (2002).
22. Plataniotis, K .N., and Venetsanopoulos, A. N., Color Image Processing and Applications, Springer, New York, (2000).
23. Swain, M. J., and Ballard, D. H., Color Indexing, International Journal of Computer, 7(1), 11-32 (1991).
24. Nievergelt, Y., “Wavelets Made Easy”, Birkauser, Boston, 1999.
25. Kuo, C.-F. J., Su, T. L., Gray Relational Analysis for Recognizing Fabric Defects, Textile Res. J. 73(5), 461-465(2003).
26. Antonini, M., Barlaud M., Mathieu, P., and Daubechies, I., Imgae coding usilng weavlet transform, IEEE trans. On Image Processing, 4(2), 205-220(1992).
27. Holland , J., Adaptation in Natural and Artificial Systems, MIT press, Cambridge, (1992).
28. Mitsuo, G.., and Runwei, C., Genetic Algorithms & Engineering Optimization, Wiley-Interscience, New York, (2000).
29. Huang, Y.-P. and Huang, C.-H., Real-valued genetic algorithms for fuzzy grey prediction system, Fuzzy Sets and Systems, 87, 265-276(1997).
30. Kuo, C.-F. J., Kao, C. Y., Self-organizing Map Network for Automatically Recognizing Color Texture Fabric Nature, Fibers and Polymers, 8(2), 174-180 (2007)
31. Cohen, F. S., Fan, Z., and Attali, S., Automated inspection of textile fabrics using textural models, IEEE Transactions on PAMI, 13, 803-808(1991).
32. Haralick, R.M., Shanmugam, K., and Dinstein, I., Textural features for image classification, IEEE Trans. Systems Man and Cybernetics, 3, 610-621(1973).
33. Kohonen, T., Self-Organizing Maps, 3rd ed., New York, (2001).
34. Ong, S. H., Yeo, N. C., Lee, K. H., Venkatesh, Y. V., Cao, D.M., Segmentation of color images using a two-stage self-organizing network, Image and Vision Computing, 20, 279-289(2002).
35. Kraaijveld, M. A., Mao, J., and Jain, A. K., A nonlinear projection method based on Kohonen's topology preserving maps, IEEETrans. On Neural Networks, 6, 548-559 (1995).
36. Vesanto, J., and Alhoniemi, E., Clustering of the self-organizing map, IEEE trans. on Pattern Analysis and Machine Intelligence, 11(3), 586-600(2000).
37. Addison-Wesley P. C., The Delphi method: techniques and applications, Advanced Book Program, 1975.