本論文有兩大貢獻，分別是改良式點擴散半色調技術與分類式特徵逆半色調技術，簡述如下:
數位半色調技術(digital halftoning)是指將連續色調影像以較少的色調表現出來，會產生此技術的原因是裝置輸出顯示的色階有所限制。以印表機為例，印表機只能控制打點(黑)與不打點(白)。如何應用這兩種色調，表現出連續色調影像的特性，這就是半色調的研究精神。首先本論文提出使用六角形模板(hexagonal grid)改進數位半色調技術中的點擴散法(dot diffusion)。傳統的點擴散半色調技術會有一組擴散矩陣(diffused matrix)與順序矩陣(class matrix)，其特點是具有平行處理(parallel processing)的能力，但其建立的半色調影像會有區塊效應(blocking effect)，人眼視覺感官上是不舒服的。本論文提出六角形順序矩陣取代原始的矩形順序矩陣，經過本論文實驗的證明，它可以大幅的減少區塊效應。本論文的成本函數(cost function)結合了人類視覺系統(Human Visual System, HVS)，並使這成本函數最小化，使其半色調影像品質更佳。在訓練過程中，本論文引入了模擬退火法(Simulated Annealing, SA)，使得訓練過程中，不易落入區域解，並最佳化本論文的順序矩陣與擴散係數。實驗證明，本論文的結果比一些常用的錯誤擴散法有更好的影像品質。
數位逆半色調技術(digital inverse halftoning)是將半色調影像還原成近似原始影像的連續色調影像。許多的影像處理技術都是為了灰階影像設計，例如：放大、縮小、明暗度與解析度調整或其他幾何變化的影像處理技術。這些影像處理技術通常無法直接使用在半色調影像上，通常先會使用數位逆半色調技術得到連續色調影像，在對其連續色調影像進行影像處理，處理完後在使用數位半色調技術得到影像處理過後的半色調影像。本論文第二部分提出分類式特徵逆半色調技術(Feature-Classified Inverse Halftoning, FCIH)，本論文提出三種特徵去執行逆半色調，分別是樣本變異數(pattern variance)、點型態(dot type)與區域平均灰階度(local average grayscale)。應用這三個特徵，搭配使用最小均方演算法(Least Mean Square, LMS)去訓練出適合各種狀況的濾波器。為了決定各特徵濾波器的數目，本論文提出一個最大濾波器差異導向(Maximum Filters Difference Guidance, MFDG)的方式決定濾波器的數目。在實驗結果裡，本論文提出的逆半色調方法不管在影像品質、記憶體消耗與計算複雜度都優於先前的方法。

Digital halftoning is a technique that can translate a continuous-tone image to a bi-tone image. It is broadly applied for displaying and printing, etc. The main challenge in this research topic is to provide high image quality while maintain low computational complexity and memory consumption. Two contributions are addressed in this thesis. In the first half, an improved dot diffusion is proposed using hexagonal grid. Currently, dot diffusion is not employed in commercial market for its rather low image quality. The dot diffusion can provide parallel processing advantage with the aid of class matrix and diffused matrix. Yet, the main deficiency of dot diffusion is the annoying blocking effect. For this, the former rectangle shape class matrix is replaced with hexagonal shape for improving image quality. As documented in the experimental results, the blocking effect is significantly reduced. The image quality is even better than some commercial error diffusion schemes. In this work, a set of class matrices and diffused matrices are provided, and the Human Visual System (HVS) is also involved for image quality evaluation to minimize the cost function. Moreover, the Simulated Annealing (SA) is employed for training process to yield the optimized class matrix and diffusion matrix.
In the second half, a Feature-Classified Inverse Halftoning (FCIH) is proposed. Inverse halftoning is a technique that can translate a two tone image to a continuous-tone image, which can be used for halftone image compression, such as facsimile transmission. Herein, the pattern variance, dot type, and local average grayscale are considered as significant features. The three features are used to classified adaptive filters for inverse halftoning using Least Mean Square (LMS). Herein, the Maximum Filters Difference Guidance (MFDG) is developed to determine the number of classified filters. As documented in the experimental results, the proposed inverse halftoning shows excellent performance in image quality, memory consumption, and complexity compared with former approaches.

[1] J. Jarvis and C.Roberts, “A new technique for displaying continuous tone images on bilevel display,” IEEE Trans. Communication .24(8), pp. 891-898, Arg. 1976.
[2] R. Ulichney, Digital Halftoning. Cambridge, MA. MIT Press, 1987.
[3] Per-Erik Axelson, "Quality Measures of Halftoned Images (A Review)," Linkoping. Department of Science and Technology, LITHITN-MT-EX--03/027, 2003.
[4] R. Nasanen, “Visibility of halftone dot textures,” IEEE Trans. Syst. Man, Cybern, SMC-14(6), pp.920-924, Dec. 1984.
[5] F. W. Cambell, R. H. S. Carpenter, and J. Z. Levinson, “Visibility of aperiodic patterns compared with that of sinusoidal gratings,” J. Physiol., 204(2), pp. 283-298, 1969.
[6] J. L. Mannos and D. J. Sakrison, “The effects of a visual fidelity criter on the encoding of images,” IEEE Trans. Information Theory, IT-20(4), July 1974.
[7] S. Daly. “Subroutine for generation of a two dimensional human visual contrast sensitivity function,” Eastman Kodak, Tech. Rep. 233203y, 1987.
[8] R. Ulichney, “The void-and-cluster method for dither array generation,” Proc. SPIE 1913, pp. 223-343, 1993.
[9] R. W. Floyd and L. Steinberg, “An adaptive algorithm for spatial gray scale,” in Proc. SID 75 Digest. Society for information Display, pp. 36–37, 1975.
[10] J. F. Jarvis, C. N. Judice, and W. H. Ninke, “A survey of techniques for the display of continuous-tone pictures on bilevel displays,” in Comp. Graph. Image Proc., vol. 5, pp. 13–40, 1976.
[11] P. Stucki, “MECCA-A Multiple-Error Correcting Computation Algorithm for Bilevel Image Hardcopy Reproduction,” IBM Res. Lab., Zurich, Switzerland, Res. Rep. RZ1060, 1981.
[12] J. N. Shiau and Z. Fan, “A set of easily implementable coefficients in error diffusion with reduced worm artifacts,” SPIE, 2658: 222-225, 1996.
[13] P. Li and J. P. Allebach, “Block interlaced pinwheel error diffusion,” JEI, vol. 14, Apr-Jun. 2005.
[14] V. Ostromoukhov, “A simple and efficient error-diffusion algorithm,” Computer Graphics (Proceedings of SIGGRAPH 2001), pp. 567-572, 2001.
[15] P. Li and J. P. Allebach, “Tone-dependent error diffusion,” IEEE trans. on Image Processing, vol. 13, pp. 201-215, Feb. 2004.
[16] T. C. Chang and J. P. Allebach, “Memory efficient error diffusion,” IEEE Trans. on Image Processing, 18 vol. 12, pp. 1352-1366, Nov. 2003.
[17] D. E. Knuth, “Digital halftones by dot diffusion,” ACM Trans. Graph., vol. 6, no. 4, Oct. 1987.
[18] M. Mese and P. P. Vaidyanathan, “Optimized halftoning using dot diffusion and methods for inverse halftoning,” IEEE Trans. Image Processing, vol. 9, pp. 691-709, Apr. 2000.
[19] J. P. Allebach, “FM screen design using DBS algorithm,” in Proc. ICIP, vol. 1, Lausanne, Switzerland, pp. 549-552, 1996.
[20] J. M. Guo and Y. F. Liu, “Improved Dot Diffusion by Diffused Matrix and Class Matrix Co-Optimization,” IEEE Trans. Image Processing, vol. 18, no. 8, pp. 1804-1816, Aug. 2009.
[21] M. Mese and P. P. Vaidyanathan, “Look up table (LUT) method for image halftoning,” in Proc. ICIP, vol. 3, pp. 993-996, 2000.
[22] M. Analoui and J. P. Allebach, “Model based halftoning using direct binary search,” in Proc. SPIE, vol. 1666, pp. 96–108, Feb. 1992.
[23] D. J. Lieberman and J. P. Allebach, “A Dual Interpretation for Direct Binary Search and Its Implications for Tone Reproduction and Texture Quality,” IEEE Trans. on image processing, vol. 9, no. 11, Nov. 2000.
[24] M. Mese and P. P. Vaidyanathan, “Look-Up Table (LUT) Method for Inverse Halftoning,” IEEE Trans. on Image Processing, vol. 10, no. 10, pp. 1566-1578, Oct. 2001.
[25] K. L. Chung and S. T. Wu, “Inverse halftoning algorithm using edge-based lookup table approach,” IEEE Transactions on Image Processing, vol. 14, no. 10, pp. 1583-1589, Oct. 2005.
[26] J. Canny, “A computational approach to edge detection,” IEEE Trans. Pattern Anal. Mach. Intell., vol. PAMI-8, no. 11, pp. 679–698, Nov. 1986.
[27] 張斐章、張麗秋，類神經網路，東華書局，臺北，第79-88頁(2005)。
[28] Widrow B. and Hoff M. E. Jr., “Adaptive Switch Circuits,” IRE Western Electric Show and Convention Recode, part 4, pp. 96-104, 1960.
[29] P. C. Chang, C. S. Yu and T. H Lee, “Hybrid LMS-MMSE inverse halftoning technique,” IEEE Trans. on Image Processing, vol. 10, no. 1, pp. 95-103, Jan. 2001.
[30] 李世炳、鄒忠毅，「簡介導引模擬退火法及其應用」，中央研究院物理研究所物理雙月刊，第二十四卷，第二期，第307-319頁(2002)。
[31] S. Kirkpatrick et.al., “Optimization by simulated annealing”, Science 220, pp. 671-680, 1983.
[32] Daniel L. Lau and Robert Ulichney, “Blue-Noise Halftoning for Hexagonal Grids,” Hewlett-Packard Company, 2004.
[33] http://en.wikipedia.org/wiki/Peak_signal-to-noise_ratio
[34] J. M. Guo and C. H. Chang, “Prediction-Based Watermarking Schemes Using Ahead/Post AC Prediction,” Signal Processing, vol. 90, no. 8, pp. 2552-2566, August, 2010.
[35] S. Hein and A. Zakhor, “Halftone to continuous-tone conversion of error-diffusion coded images,” IEEE Trans. Image Processing, vol. 4, pp. 208-216, Feb. 1995.
[36] S. M. Schweizer and R. L. Stevenson, “A Bayesian approach to inverse halftoning,” in Proc. SPIE Conf. Human Vision, Visual Processing, Digital Display IV, vol. 1913, San Jose, CA, pp. 282-292, Feb. 1993.
[37] P. W. Wong, “Inverse halftoning and kernel estimation for error diffusion,” IEEE Trans. Image Processing, vol. 4, pp. 486-498, April 1995.
[38] Z. Xiong, M. T. Orchard, and K. Ramchandran, “Inverse halftoning using wavelets,” in Proc. IEEE Int. Conf. Image Processing, vol. 1, Lausanne, Switzerland, pp. 569-572, Sep. 16-19, 1996.
[39] T. D. Kite, N. Damera-Venkata, B. L. Evans, and A. C. Bovik, “A fast, high-quality inverse halftoning algorithm for error diffused halftones, ” IEEE Trans. Image Processing, vol. 9, no. 9, Sep. 2000.
[40] N. Damera-Venkata, T. D. Kite, M. Venkataraman, and B. L. Evans, “Fast blind inverse halftoning,” in Proc. IEEE Int. Conf. Image Processing, vol. 2, Chicago, pp. 64-68, Oct. 4-7, 1998.
[41] U. F. Siddiqi and S. M. Sait, “Algorithm for parallel inverse halftoning using partitioning of Look-Up Table (LUT),” IEEE International Symposium on Circuits and Systems, pp. 3554 - 3557, May 2008.
[42] N. Atamena, Z. Tifedjadjine, Z. Dibi and A. Bouridane, “A Fast Inverse Halftoning Algorithm using LUT Approach for Grey-level Images,” Information and Communication Technologies: From Theory to Applications, pp. 1-4, April 2008.
[43] 劉雲夫，「應用最小均方演算法與天真貝式分類器之分類式逆半色調技術」，碩士論文，長庚大學電機工程研究所，桃園(2009)。