摘要
理论和实验证明:把图象的高频和低频分量分别编码可获得较大的数据压缩比和良好的图象质量。高低频编码的关键技术是高低频分量的分解与综合。本文研究了利用有序抖动(Ordered Dither)矩阵匹配法检测边缘,从而把图象分成边缘子块和均匀子块,较好地解决了上述问题。本文具体介绍了边缘子块的检测算法和一个数据压缩方案,它对边缘子块采用BTC编码,对均匀子块用其子块灰度均值表示。对于有复杂背景的人头象,取得了0.701 bit/pel的结果。
It has been proved by theory and experiments that high data compression ratio and good image quality can be obtained by coding the high and low frequency component of the image respectively. The key of this image coding technique is the decomposition of the image into high and low frequency component and the image reconstruction from them. This paper presents a new way of image decomposition-Ordered Dither Matrix Matching that seperates the image into edge blocks and flat blocks. The edge block is processed by BTC (Block Truncation Coding), while the flat block is represented by its average greyscale. Computer simulation results show that the bit rate for a head-and-shoulder image is 0.701 bit/pel with subjective quality neatly similar to that encoded by BCT at the bit rate of 2 bit/pel.
出处
《通信学报》
EI
CSCD
北大核心
1991年第2期77-81,共5页
Journal on Communications
