图像分割的误差分散半调算法  

Error diffusion halftone algorithm based on image segmentation

作　　者：万波[1] 王泉[1] 高有行[1] 

机构地区：[1]西安电子科技大学计算机外部设备研究所,陕西西安710071

出　　处：《西安电子科技大学学报》2009年第3期496-501,546,共7页Journal of Xidian University

基　　金：国家部委基金资助

摘　　要：为了减少数字图像半调过程中传统误差分散法产生的可见人工纹理,提出了一种基于图像分割技术的误差分散法.该方法使用图像分割技术将图像分割为同灰度区域和不同灰度区域,在半调过程中只对同灰度区域使用阈值调制,并使用与输入灰度相关的具有4个方向的误差分散系数处理半调误差.实验结果表明该方法能够有效地减少半调图像中的人工纹理,而且在峰值信噪比、加权信噪比和统一图像质量指标标准上优于Floyd-Steinberg的误差分散法、Ostromoukhov的误差分散法和Bingfeng Zhou的误差分散法.To reduce the visual artifacts produced by the conventional error diffusion algorithm in the digital halftone process, an error diffusion algorithm based on image segmentation is presented. In this algorithm the image segmentation technology is used to segment an image into two types of region. In one region, pixels have the same gray level while in the other they have a different gray level. In the halftone process, threshold modulation is only applied to the region which has the same gray level and at the same time the input dependent error diffusion coefficients which have four directions are used to process the halftone error. Experimental results show that our algorithm can reduce visual artifacts effectively and outperforms Floyd-Steinberg＇s error diffusion, Ostromoukhov＇s error diffusion and Bingfeng Zhou＇s error diffusion in the PSNR （Peak Signal Noise Ratio）, WSNR （Weighted Signal Noise Ratio） and UIQI （Universal Image Quality Index） criteria.

关 键 词：图像分割 误差分散 数字半调 

分 类 号：TP391.41[自动化与计算机技术—计算机应用技术]