数字图像修补算法工程实现的研究

【作者】 沈妮；

【导师】 彭启琮；

【作者基本信息】 电子科技大学 ， 信号与信息处理， 2007， 硕士

【摘要】 图像修复是图像复原研究中的一个重要内容,也是当前图像处理和计算机视觉领域中的一个研究热点。图像修复是对图像上信息缺损区域进行信息填充的过程,其目的是恢复有信息缺损的图像,并使观察者无法察觉图像曾经缺损或已被修复。该项技术在文物保护、影视特技制作、老照片的修复、图像中文本的去除、障碍物的去除以及视频错误隐藏等方面,有着很高的应用价值。该领域的研究,国外正在蓬勃发展,国内尚处于起步阶段。本文介绍了一种新型的图像修补方法,将目前先进的图像修复算法与当前运用广泛的数字信号处理芯片相结合,使图像修复过程变得简单易掌握。本论文采用了两种图像修补算法,一种是基于p-Laplace算子的图像修复算法,用于修复图像中的划痕等较小尺度的损伤;第二种是结合自然图像的纹理走向及纹理匹配块的自适应选取,利用纹理合成技术,对大块信息丢失的图像进行修复。本文的主要贡献如下:一、分别研究了基于p-Laplace算子和基于纹理合成的图像修复算法的原理,在Matlab下进行了仿真理解,并与同类算法做比较,了解了这两种算法的先进性。二、在基于TMS320DM642芯片的硬件平台上实现了以上两个算法,首先是系统方案设计,然后搭建了演示平台,在EVM开发了两套图像修复软件,并对软件进行了优化、调试和折中处理。为显示处理后的图像数据,设计了PC端的界面软件并进行了联机调试,最后演示了系统的工作过程。三、通过上述工作,验证了两种修复算法的工程可实现性,证明了在DSP上针对不同种类的数字图像做修复,可达到良好的效果,然后将本文提出的图像修复方法与已有的修复方法进行对比,讨论了在DSP上实现图像修复和特殊处理的优越性,指出了目前系统还存在的不足,并提出了进一步的工作。更多还原

【Abstract】 Image inpainting is an important research topic in the area of image restoration. Currently, this kind of techniques has found broad applications in image processing, vision analysis, digital restoration of ancient paintings for conservation purposes, text removal and objects removal in images for special effects, restoration of old photographs or films with scratches or missing patches, disocclusion in computer vision, errors conceal in videos, and so on. However, everything is just at the beginning. Based on understanding and utilization of the previous achievements, this dissertation focuses on developing innovative digital image inpainting methods.This paper proposed a new digital image inpaiting method, which combining two advanced algorithms and DSP processors. It makes image inpainting quick and easy to operate. The main jobs of this theme are followed:1、The principles of the two algorithms mentioned above are researched and analyzed respectively. Compared the tow with other inpainting algorithms by simulating them using Matlab, results showed that the ones used by this paper have obvious advantages.2、This theme focuses on the implementation of the algorithms on the hardware platform based on TMS320DM642. First, the design of whole system is introduced, as the DM642 EVM the main hardware platform, DSP software in terms of the two algorithms are developed on it. To increase its efficiency, several ways of code optimization are introduced. What is more, the development of an interface software on PC for displaying the digital images is also showed with the whole displaying process.3、The testing results show that the image inpainting on DSP is feasible and it is proved that this method has more advantages than traditional ways. Several drawbacks, as well as their causes, have been discussed, and the corresponding plan of improvement has been proposed.更多还原