项目名称: 基于图像序列的动态场景三维结构和运动恢复的鲁棒性算法
项目编号: No.61273282
项目类型: 面上项目
立项/批准年度: 2013
项目学科: 自动化技术、计算机技术
项目作者: 刘继忠
作者单位: 南昌大学
项目金额: 80万元
中文摘要: 本项目旨在研究由未标定图像序列恢复动态场景的三维结构及运动信息。项目的主要研究内容包括鲁棒性重建算法、动态场景的三维结构和运动估计、以及探索新的复杂动态场景的建模方法。其研究成果不仅具有重要的理论意义,而且可以广泛应用于机器人导航、视觉监控、人机交互、虚拟现实等领域。 项目的特色及创新之处在于:一是提出具有强鲁棒性的关于非刚体和动态场景的重建算法,可以较好地解决数据存在较大噪声、错误匹配以及特征点丢失等问题。这是一个极具挑战性的课题,文献中已有方法主要是针对刚体,而且计算量很大;二是研究关于复杂动态场景三维结构和运动的恢复算法,动态场景中可能同时含有非刚体、铰接物体、运动的非刚体等。这是本领域的一个未解难题,已有方法只能适用于单一物体和场景。
中文关键词: 计算机视觉;三维重建;鲁棒算法;运动分析;视觉重构
英文摘要: The project aims to recover three-dimensional structure and motion parameters of dynamic scenes from uncalibrated image sequences. The main research topics include robust 3D reconstruction algorithm, structure and motion estimation of dynamic scenes, and new modeling method for highly dynamic scenarios. The study is not only academically significant, it is also significant to many applications such as robot navigation, environment modeling, visual surveillance, human-machine interaction, and virtual reality. The novelty of this project lies in two aspects. First, the research will propose robust 3D reconstruction algorithms for challenging nonrigid objects and dynamic scenes in presence of significant noise, outliers, and missing features. Previous algorithms can only work for rigid objects or static scenes. Second, the research will establish new approaches to recovering the structure and motion parameters of highly dynamic scenes that may contain nonrigid objects, articulated objects, and moving objects. The study of these open problems is more challenging due to its complexity and the lack of reports in the literature.
英文关键词: computer vision;three-dimensional recovery;robust algorithm;motion analysis;vision reconstruction