基于区域的三叉树图象编码新方法

A New Region-based Image Coding Method Using Binary Tree

本文提出了一种基于区域的图象编码新技术，它根据图象区域灰度分布特点，以灰度误差最小平方和为准则，采用四向递归二分法逐渐将图象表面划分为若干凸多边形，使之逼近原始图象。软件模拟实验表明，当头肩灰度图象压缩比超过２０：１时，重建图象主观质量仍然较好。文中介绍了四向递归二分法的基本算法和二叉树图象编码方法，同时给出了若干实验结果。

new region-based image coding method is proposed in this paper. According to the gray level distribution of the image surface, the compressed image is partitioned into many convex polygons to appoximate the original image. Encoding and decoding use the criterion of the least-squares gray error and the approach of recursive 4-way bipartitioning. The experiments show that reconstruction images will high gaaliy can be obtained even when the compression ratio is more than 20: 1. The region-based and surface description coding theory and an efficient algotithem of recursive 4-way bipartitioning are described Some experimental results are also provided.