基于视觉感知显著性的三维模型理解和绘制技术研究

项目名称： 基于视觉感知显著性的三维模型理解和绘制技术研究

项目编号： No.61070126

项目类型： 面上项目

立项/批准年度： 2011

项目学科： 金属学与金属工艺

项目作者： 缪永伟

作者单位： 浙江工业大学

项目金额： 10万元

中文摘要： 基于视觉感知理论和三维数字模型的视觉显著性刻画，项目组围绕数字模型视觉显著性几何描述的建立，数字模型可视化增强技术和表意性绘制等方面开展了创新性研究工作，对其关键技术问题提出了若干新的解决思路和新的算法。项目组提出了复杂数字模型显著性度量的新方法；根据模型的显著性度量提出了模型显著域上的形状调控和处理方法以实现模型的显著特征增强；利用模型的显著性度量指导模型顶点法向的动态调整以实现模型的表意性绘制；基于模型显著性信息熵定义实现了三维模型的最优观察视点选择；同时项目组还对曲线曲面几何进行了理论分析，实现了曲线曲面的特征分析和特征提取等。

中文关键词： 视觉显著性；几何描述；表意性绘制；可视化增强

英文摘要： Based on the viusal perception theory and the definition of visual saliency, this project has investigated some kernel research area for understanding and rendering 3D models,such as the geometric descriptions of visual saliency, the visualization enhancement and expressive rendering techniques of 3D models. Some new ideas and key algorithms are proposed. First, due to the visual perception theory, a novel relief saliency definiton for 3D complex models is presented. Second, owing to the visual saliency measure and saliency-domain shape modeling techniques, a new shape manipulation scheme is presented to enhance the visually salient features of 3D models. Third, due to the visual saliency measure of 3D shape, a new expressive rendering technique is proposed to adjust the illumination and shading of the underlying model by dynamically perturbing the surface normals. Fourth, to automatically select the best viewpoint of 3D models, a novel relief saliency entropy driven view descriptor is proposed, which is defined by incorporating the relief saliency measure of complex models into the entropy analysis of viewpoint mutual information. Furthermore, the project also investigated some key problems for the feature analysis and feature extraction of surfaces and curves.

英文关键词： Visual Saliency; Geometric Description; Expressive Rendering; Visualization Enhancement