一种呈现清晰纹理的快速误差扩散算法  

Structure-aware and High-speed Error Diffusion Algorithm

作　　者：章芳[1] 庞明勇[1] 

机构地区：[1]南京师范大学教育技术系,江苏南京210097

出　　处：《小型微型计算机系统》2010年第8期1609-1612,共4页Journal of Chinese Computer Systems

基　　金：江苏省高校自然科学基础研究项目(07KJD460108)资助;南京师范大学高层次人才科研启动基金项目资助

摘　　要：清晰地呈现图像的细节纹理结构是现代数字半色调技术研究的重要方面.提出一种简单有效、且能保持原图像纹理结构的半色调误差扩散算法.该算法以Ostromoukhov算法为基础,根据图像灰度空间的局部分布特征,通过计算误差扩散像素点的局部视觉纹理反馈信息,对原固定阈值进行调节.实验结果表明,改进后的算法生成的半色调图像再现的纹理结构清晰自然,且具有蓝噪声特性,在PSNR、UIQI和SSIM三项指标评测方面均有较好表现.How to clearly show details of edge and texture structure in halftone images is an important and key aspect for modern halftone techniques.This paper proposes a simple and effective error diffusion halftone algorithm,which can keep structure aware in halftoned images.The method is based on the well-known Ostromoukhov algorithm and some improved steps are added to it.According to local area gray distribution in images,our algorithm calculates local visual information of texture feedback called ITF for short around each pixel,and then the ITF is used to adjust a variational threshold,which is often used as a fixed value in traditional halftoning algorithms.The experiment results show that achieved halftone images by our method have clarity and natural texture.The halftoned images not only represent ideal blue noise spectrum,but also show a better performance in evaluation index of PSNR、UIQI and SSIM.

关 键 词：半色调 误差扩散 蓝噪声 纹理结构 

分 类 号：TP301[自动化与计算机技术—计算机系统结构]