一個修改的架構樹集合分割編碼演算法

Abstract

由於網際網路以及各種多媒體應用的普及化，數位影像已經成為一種重要的資料形式。一般來說，數位影像需要大量的儲存空間，對於儲存及傳輸效率而言，數位影像的壓縮技術是不可或缺的。近年來應用小波轉換諸多優異特性的壓縮演算法相繼被提出，其中架構樹集合分割編碼演算法即是相當廣為人知的一個編碼演算法。本篇論文提出了兩種方法以改善架構樹集合分割編碼演算法的效率。第一種方法去除此演算法中可預知的編碼；第二種方法則利用了不同頻段的係數強度大小的分佈不同的特性以節省此演算法中不必要的編碼。文中並提供一些實驗結果以說明所提出的兩個方法改進了此演算法的效率。

Owing to the popularity of the Internet and various multimedia applications, digital images have become an important source of information. For the sake of storage and transmission efficiency, image compression techniques are essential for various applications. In recent years, many compression algorithms utilizing the excellences of the wavelet transform were proposed. The SPIHT algorithm is one of the well-known image coding algorithms. In this thesis, two methods were proposed to improve the coding efficiency of the SPIHT algorithm. The first method removes the predictable coding redundancies in the algorithm; the second method takes advantage of the characteristic that different subbands have different magnitude distributions. Some experimental results are also given to show the effectiveness of the algorithm.