项目名称: 数字图像自动修补理论与算法研究
项目编号: No.60802040
项目类型: 青年科学基金项目
立项/批准年度: 2009
项目学科: 电工技术
项目作者: 张红英
作者单位: 西南科技大学
项目金额: 20万元
中文摘要: 图像修补技术是当前图像处理和计算机视觉领域中的一个研究热点。目前存在的图像修补技术大都需要专业技术人员指定修补区域和选择修补算法,因而该技术不利于进行大批量的图像修补处理,难以推广到视频修复的应用。因此,迫切希望有一种自动、快速、简单的修补方法。本项目主要针对破损区域的自动获取方法、根据破损区域的大小选取不同的修补算法以及修补算法的快速实现等方面展开研究,在垂直划痕自动检测与修复、任意划痕自动检测、视频字幕自动提取与修复、噪声滤除、破损区域大小的判定与选取不同修补算法、修补算法的最优实现以及修补算法的应用等方面取得了一定的研究成果,为后续的研究提供了基础性工作。
中文关键词: 图像修补;偏微分方程;纹理合成;破损检测
英文摘要: Image inpainting is a hot research topic in image processing and computer vision field.The existing image inpainting technology mostly requires professional and technical personnel to specify the inpainted regions and select the inpainting algorithm, so the technology is not conducive to inpaint large quantities of images which make it be difficult to be extended to video inpainting applications. Therefore, it is eager to have an automatic, fast and simple way to inpaint images. Our project aimed at automatically detect the damaged areas, select a different inpainting algorithms based on the size of the damaged areas, as well as quickly implement of inpainting algorithms.We have made some progress in the automatic detection and restoration of vertical scratches, the automatic detection of any scratches, the automatic extraction and inpainting of video captions, noise filtering, the judgment of the size of the damaged area and selection a different inpainting algorithms to inpaint the different damaged areas, and the optimal implementation of algorithms, and so on. All these researches provid a basic work for the follow-up study.
英文关键词: image inpainting;Partial Differential Equations; texture synthesis; damaged area detection