导　　师： 李学庆

学科专业： 081202

授予学位： 博士

作　　者： ;

机构地区： 山东大学

摘　　要： 视觉是人类从世界中获取信息和经验的主要方式。多年以来,科学家和工程师们一直都致力于提高图像质量。近来数码相机的普及提升了图像获取的质量,同时也产生了大量的图像和照片。但是,在某些情况下获取高质量的图像即使对于有经验的摄影人员也是比较困难的事情。另外,图像在打印、传输和扫描等过程中也会由于半色调处理和信道噪声造成图像退化降质的问题。当使用扫描仪对打印图像进行扫描后,由于打印图像时半色调处理带来的网屏效应及扫描处理噪声,往往后造成类似于波纹的图像瑕疵或噪声。实际应用中,需要对以上图像进行有效的反半色调及其去噪处理。因此,在图像恢复的各项技术中,反半色调技术/(Inverse Halftoning/)、去网屏/(Image descreening/)技术和图像去噪/(Image Denoising/)技术是各项应用的基础、核心技术,许多图像处理应用都建立在反半色调和图像去噪等图像恢复技术的基础之上。 由于缺少输入数据的足够信息,图像恢复中的反半色调和去噪问题得到的结果往往不是唯一的。该问题属于不适定问题,因此缺少统一的模型进行图像的恢复。结合基于图像内容分析的方法,针对不同种类的半色调图像包括二值图像和扫描半色调图像本文提出了统一的解决方案,本文的主要工作和创新点包括： 1.提出一种针对二值半色调图像的基于Voronoi划分的图像特征分析方法。该方法充分利用了图像内容的几何分析方法,将Voronoi图引入到半色调的特征分析中。通过Voronoi划分获取了图像的局部特征包括灰度值、邻域约束和图像梯度等。 2.提出了基于离散Sibson插值的反半色调算法。该算法充分利用了以上基于Voronoi图的特征分析方法,通过Voronoi图获取图像中种子点的灰度值估计,然后通过离散Sibson插值计算获取最终的恢复图像。有效地提� The visual is one of the important forms of the human experience. Over the ages scientists and engineers have strived to capture precious and high quality images. Recently, the popularity of digital cameras improved the quality of the image acquisition and generated a lot of images and photos. However, it is relatively difficult to obtain high quality images even for the experienced photographer in some cases. In addition, images can be degraded during the process of print, transmission and scanning due to the halftoning process and channel noise. When using a scanner to scan the printed image, there are a lot of screen patterns and noises in the scanned images. In the practice application, we should adopt inverse halftoning and denoising algorithm to recover images. Therefore, the method of image restoration for inverse halftoning and denoising, is the basis and core technique of image processing. Many image processing applications are built on the quality of image restoration. Image restoration is a challenge problem. Inverse halftoning and image denoising problem are ill-posed, where the number of unknown values outweights the number of observations. As a result, it is necessary to adopt image content to improve the restoration process. Recent Research focus on using content based adaptive method to improve image restoration quality. In this work, we introduced image content analysis approach to address the image quality restoration. The contributions of the paper mainly include: 1. A halftone image analysis approach based on discrete Voronoi diagram is proposed. With the image geometry presentation, Voronoi diagram was introduced to halftone image local features analysis. The image feature including tone values, neighborhood constraints and gradient can be calculated from above analysis approach. 2. We proposed an inverse halftoning algorithm based on discrete Sibson interpolation. Firstly, the discrete Voronoi diagram is introduced to estimate halftone dot's gray value. Then, an improved discrete

关 键 词： 反半色调 离散 图 离散 插值 分布 非局部去噪

领　　域： [自动化与计算机技术] [自动化与计算机技术]