【作者】 袁明；

【导师】 石光明；

【作者基本信息】 西安电子科技大学 ， 电路与系统， 2006， 硕士

【摘要】 由于多媒体技术的发展,图像压缩是目前图像处理学科中的一个研究热点。分析表明,原始图像数据在相邻像素之间存在较强的相关性,通过将从空间域映射到变换域,在变换域进行处理,可以有效去除相关性,获得更好的压缩效果。实践表明,将块截断编码与子带编码相结合(即子带块截断编码算法),能去除前者所带来的严重的块效应,使重建图像质量明显提高。尤其是在HDTV和高分辨率视觉通信的模拟和累进编码的实时传输上表现出某些特殊的优点。本文主要分析了完全重构子带滤波器组的图像分解与合成,以及块截断编码的基本原理,进而将两者结合,实现子带块截断编码,并介绍了子带块截断编码算法(SAMBTC)的优缺点及其实现过程。在此基础上,本文提出了一种新颖的基于MMSE的按能量排序的子带块截断编码算法。该方法有效的克服了以往SAMBTC算法所带来的图像质量不理想,压缩比不高的弊端,并保持了原有的运算简捷性。通过对灰度图像的仿真实验发现,该算法在那些细节较少的灰度图像上显示出了很好的实时传输性和重建效果。同时,考虑到该算法的简洁性,本文也将该算法应用于彩色图像上。虽然重建图像存在一定的色差,但是通过改进,仍具有很大的潜力。更多还原

【Abstract】 With the development of multimedia technology, the image compression hasbecome more and more important in the field of image processing. It’s analyzed thatthere are better correlations between neighbor pixels inside source image data. Bettercompression ratios could be reached through transformations from space field tofrequencyfield.Some experiments have proved that the combination of Block Truncation Codingtechnique and Subband Coding can get the reconstructed picture of a little raggedness ofsharp edges, which makes use of the masking property in human vision. SubbandAMBTC not only outperforms full-band AMBTC by a great margin but also presentscertain intrinsic merits for real-time and flexible implementations, especially tofacsimile and progressive coding of future HDTV and high-resolution visualcommunications.In this paper, it focuses on the research of the decomposition and synthesis ofimage based on the perfect-reconstruction filter banks and Absolute Moment BlockTruncation Coding. Then we combine the Absolute Moment Block Trncation Codingand Subband Coding,which is called Subband Aboslute Moment Block TrncationCoding (SAMBTC). After we introduce the relative merits and implementationprocedure of SAMBTC, a novel algorithm of SAMBTC based on MMSE is presentedin this paper, which is not only to effectively eliminate the blocking effect even with atwo-tap QMF bank, but also to improve the compression ratio and PSNR of thereconstructed image. Experiment results illustrate that the novel algorithm of SAMBTCbased on MMSE is successfully applied to the gray images which has little details withgood real-time transmission and achieving better imaging quality. This algorithm is alsoused in color images because of its simple and fast computation with some chromaticaberration.更多还原