基于结构光方法的全向视觉测量与控制研究

项目名称： 基于结构光方法的全向视觉测量与控制研究

项目编号： No.61472037

项目类型： 面上项目

立项/批准年度： 2015

项目学科： 计算机科学学科

项目作者： 李原

作者单位： 北京理工大学

项目金额： 80万元

中文摘要： 全向视觉在场景感知和机器人视觉导航中有着广泛的应用，是当前计算机视觉的一个热点分支。目前全向视觉研究一般是基于单一视点约束下进行系统模型分析、图像信息处理和场景特征定位。本项目将工业领域广泛应用的结构光视觉引入到全向视觉系统中，利用结构光视觉中照明光模态可控的约束条件，建立全向结构光视觉系统中景物点和图像点之间的映射关系，拟合和估计视觉系统成像参数，实现视觉系统标定；结合室内环境景物模型的先验知识，研究非单一视点约束条件的室内全景图像的展开复原算法；分析混合图像中对象的成像模型，图像特征空间分布和结构光约束下运动规律，研究复杂图像特征提取算法。通过本项目的工作，将丰富结构光视觉和全向视觉理论，并结合移动机器人平台研究全向结构光视觉的场景感知测量与导航控制方法。

中文关键词： 结构光视觉；全向视觉；标定；视觉测量；特征提取

英文摘要： Omnidirectional vision has been widely used in robot vision navigation, is currently a hot branch of computer vision. Currently the study on omnidirectional vision is generally carried on the analysis and information processing system model under the constraint of a single viewpoint. The project introduces structured light vision to the visual system, which is successfully applied in industry. Using of structured light vision constraint, the whole mapping between structure of the light and the image point of the scene point visual mapping between the can be obtained, fitting and estimation of system parameters are achieved for vision system calibration. Combine indoor environment prior knowledge, panoramic image restoration algorithms are investigated based on non-single viewpoint constraint; with the analysis of model mix images of different objectives, and the spatial distribution of the image features and structure of light constraints, complex image feature extraction algorithms are developed. The work of this project will enrich the structured light vision and omni-directional vision theory, and omnidirectional structured light vision measurement and control methods are proposed with the mobile robot platform.

英文关键词： structured light vision;omnidirectional vision;calibration;vision measurement;feature extraction