基於渾沌門檻值調變與邊緣增強誤差擴散法之FPGA設計與實現

Abstract

數位半色調是一種將連續色調影像轉換成有限色調影像的技術，此技術的主要特點為使用有限的色調去呈現出連續色調的影像。此特點被廣泛應用於顯示器與印表機裝置上。然而，傳統的數位半色調演算法仍有一些缺點，因此在本論文中我們提出一個新的數位半色調演算法。有別於傳統演算法使用固定的門檻值，我們的演算法所使用的是渾沌門檻值調變，藉由渾沌門檻值調變去減少圖片中不自然的紋理。為了更進一步的增加圖片的視覺品質，我們加入邊緣增強之量化方式以增強物體之邊緣。最後為了增加執行速度以符合即時處理之需求，我們設計了一個硬體電路來實現我們的演算法。實驗結果證明，我們所提出的新演算法的確能提高視覺品質，得到較佳的圖片，並且我們設計的硬體電路的執行速度能比軟體的執行速度快超過10倍。

Digital halftoning is the technique of transforming continuous-tone images into a limited number of gray level ones. The goal is to create the perception of a continuous-tone image using limited gray levels. Hence, it is broadly applied to display and printing devices. However, there are still some problems in traditional digital halftoning methods. In this thesis, we present an effective digital halftoning algorithm to solve these problems. Unlike traditional algorithms using the fixed threshold, the new algorithm uses chaotic threshold modulation to reduce the visual artifacts. Furthermore, for edge enhancement, we also incorporate a novel quantization method into our algorithm. To speed up the execution time to meet the requirement for real time processing, we design a useful hardware architecture to implement the proposed algorithm. In our experiment, the proposed algorithm achieves more satisfying performance than existing algorithms and the execution time of hardware can be sped up more than 10 times than software.