Digital Image Halftoning Based on Extended Dot Diffusion

• Main Authors: Jun-Chun Gao, 高俊淳
• Other Authors: Wen-Liang Hsue
• Format: Others
• Language: zh-TW
• Published: 2017
• Online Access: http://ndltd.ncl.edu.tw/handle/xspnx9

碩士 === 中原大學 === 通訊工程碩士學位學程 === 105 === In the past, traditional digital image halftoning techniques, such as order-dither, error diffusion, dot diffusion methods, transform continue-tone images into images with only two binary gray levels 0 and 255. It uses the human visual system property that makes halftone images look like continue-tone images. As to the digital image inverse halftoning technique, it converts halftone images to continue-tone images. It has been widely used for image processing such as edge detection, sharpening, and compression. This research focuses on digital image halftoning based on extended dot diffusion and inverse halftoning based on K-means clustering. Digital image halftoning based on extended dot diffusion combines order-dither and three kinds of dot diffusions as Knuth, Mese and Guo’s halftoning techniques. It utilizes the threshold matrix rather than one fixed threshold to improve the efficiency of processing. The method of measuring the quality of image is PSNR(Peak Signal to Noise Ratio). Digital image halftoning based on extended dot diffusion gets better image quality than the Knuth’s dot diffusion and better processing efficiency than each of them. Inverse halftoning based on K-means clustering takes the advantage of the easy algorithm and clustering, which brings it better classification results. This research compares the inverse halftoning based on K-means clustering with the Sobel and Laplacian data edge classifications, as well as the LMS and LS inverse halftoning techniques, and then the results are analyzed.