项目名称: 图像破损结构的三维重建理论研究
项目编号: No.61271326
项目类型: 面上项目
立项/批准年度: 2013
项目学科: 无线电电子学、电信技术
项目作者: 何凯
作者单位: 天津大学
项目金额: 70万元
中文摘要: 大区域复杂背景图像的自动修复是当前多媒体图像处理领域的研究热点,而图像破损结构信息的重建则是解决这一问题的关键。由于缺乏有效的三维建模方法,当前结构重建算法大都是在二维空间进行,对结构信息缺乏完整的描述,难以实现其完整修复和自动传播。课题拟利用图像分解技术,将近年来出现的Kansa三维建模方法引入图像处理领域,开展对图像结构信息三维重建理论的基础性研究,以实现图像破损结构信息的自动传播和完整修复。课题主要研究内容包括:根据结构图像三维建模的需要,研究能够反映图像结构特征的建模中心点自适应选取方法,并在此基础上,确定结构信息三维重建的数学模型和具体方法;研究图像分解参数的最优化选取方法;根据图像结构的三维模型,分析和改进已有纹理合成方法,实现二者的有机结合;改进传统纹理合成方法,研究能够适应不同图像结构的样本块大小自适应选取方法,为大区域复杂背景图像的自动修复提供新的方法和理论依据。
中文关键词: 图像修复;结构传播;3维建模;纹理合成;
英文摘要: The image inpainting in large region with complex background is the hot research topic in the field of multimedia image processing, while the reconstruction of damaged image's structure information is the key of solving this problem. Due to the lack of effective 3D modeling method, current structure reconstruction algorithms are mostly realized in 2D space. It is difficult to achieve complete restoration and automatic propagation of structure information with current methods for lacking of a complete description of structure information. This project intends to introduce the Kansa 3D modeling method to the image processing field with image decomposition technique, and develop basic research of image structure's 3D reconstruction theory to realize the automatic propagation and complete restoration of image's structure information. The main research topics of this project include: developing an adaptive test points selection method according to the requirment of image structure 3D modeling; determining the mathematical model and realizable method for 3D reconstruction of image structure information based on the selected test points; researching the optimal selection method of image decomposition parameters; analyzing and improving the exist texture synthesis methods to achieve their organic combination according t
英文关键词: Image Inpainting;Structure Propagation;3D Modeling;Texture Synthesis;