即時二元影像轉換系統之演算法則及架構設計

Abstract

本篇論文敘述一個二元影像轉換系統的演算法則及架構設計. 此系統主要 分為兩部份. 第一部份是一個用來消除影像中雜訊的中值濾波器, 我們利 用一種新開發出的架構---最佳化插入與消去 Optimized Delete-and -Insert(ODI), 來做為中值濾波器之中值尋用. 第二部份將介紹一個為了 能達到即時操作而新發展出的二元影像轉換法則----適應性數位轉換法 則 Adaptive Binary Thresholding (ABT).在演算法則設計上, 主要是一 個將灰階影像轉換為二元影像的即時法則, 其考慮的因素包含整體影像的 分析---Global Histogram Analysis,及圖素間相鄰影像環境的分析-- Local Image Environment Analysis, 以決定要將灰階之圖素以壹或零表 示.在架構設計上, 主要是一個以移位暫存器為主要架構之高速排序器,以 作為中值濾波器中之中值找尋用, 此排序器原理是利用新開發出的排序法 則----插入排序器, 且藉由國科會晶片設計製作中心的協助, 已製成一 68支接腳的高速排序IC, 經驗證此排序法則為目前現有之排序法中最有效 率者, 若同時考慮速度及硬體複雜度. 此法則已申請專利. In this thesis, an algorithm and an ASIC architecture for bi- nary image coversion system are presented. This system theoreti- cally contains two functional blocks. The first block is a median filter which is used for eliminat- ing the burst noise of the input image, and a newly developed median- filtering structure, 'Optimized Delete-and-Insert' or ODI method, is implemented. The second block of this system also uses a newly developed alrogithm called 'Adaptive Binary Thre- sholding' or ABT method for real time operation. In algorithm design, it is an adaptive real-time algorithm considering both the global histogram analysis and local image enviroment analysis for converting, or thresholding, the gray- scale images into binary ones. In structure design, a 68-pins ASIC for sorting has been implemented through the MPC support from Chip Implementation Center (CIC) of NSC. This sorter based on ODI method uses shift register arrays as its basic skeleton, and is a new sorting algorithm by the insertion sorting method, it has been proved to be a very powerful and effective method for data sorting so far, and it is also applied as a patent in both ROC and USA.