论文部分内容阅读
分块截断编码(Block Truncation Coding 缩写为 BTC)是一种新的图象压缩法。它是将图象分割成适当大小的子块,利用子块内各邻近象素的相关性,灰度级的相互近似性,利用两个等级(1bit)的非参量量化器,使图象的局部特性达到自适应,局部样本矩保持相等,并能得到1.5 bit/pel 的压缩图象。虽然压缩比不很高,但处理时不要求大的数据存储器,而且计算量小。本文简要叙述了分块截断编码的信息压缩原理,提供了程序框图。利用我们现有的图象处理系统,按基本的 BTC 法,最小均方差(MSE)逼真准则和最小平均绝对误差(MAE)逼真准则分别对同一幅静止图象进行了实验,同时对加噪声图象进行了 BTC 处理,最后对各种结果进行了分析与比较。
Block truncation coding (Block Truncation Coding abbreviated as BTC) is a new image compression method. It divides the image into sub-blocks of the appropriate size. By using the correlation of adjacent pixels in the sub-blocks and the mutual approximation of gray levels, two levels (1 bit) non-parametric quantizer are used to make the The local characteristics are adaptive, the local sample moments remain equal, and 1.5-bit / pel compressed images are obtained. Although the compression ratio is not high, but processing does not require large data memory, and a small amount of computation. This article briefly describes the block truncation coding information compression principle, provides a block diagram. Using our existing image processing system, the same static image was tested based on the basic BTC method, the minimum mean square error (MSE) criterion and the minimum mean absolute error (MAE) criterion respectively. At the same time, Like the BTC processing, the final analysis of various results were compared.