摘要
针对基于模型的姿态估计问题提出了一种新颖的深度估计算法,该算法能根据特征点对而不是单个的特征点与相机的几何关系,直接计算出相机运动前后特征点的深度.根据该深度估计技术,进一步提出了一种简单的可得到闭式解的运动参数估计方法.该姿态估计方法实时性高、便于实际应用,其方法的有效性得到了实验验证.
This paper first presents a novel model-based algorithm for the pose depth estimation. Depending on the geometric relationship between a camera and feature point pairs instead of an individual point, the algorithm directly calculates the depth of the feature point for the moment prior-and-post camera motion. Based on the depth estimation, also proposed is a simple motion parameter estimation method by which the close-formed solution can be obtained. The proposed approach is characterized in being able to obtain the close-formed solution based on the feature points, and real time applicable. Adequacy of the presented algorithm is validated by the experimental results, as presented in this research.
出处
《宁波大学学报(理工版)》
CAS
2008年第1期1-6,共6页
Journal of Ningbo University:Natural Science and Engineering Edition
基金
国家科技部创新基金(06C26213311210)
宁波市自然科学基金(2007A610013)
浙江省教育厅高校科研计划项目(20061653)
关键词
姿态估计
深度估计
运动估计
相机标定
pose estimation
depth estimation
motion estimation
camera calibration