SIFT-Based Robotic Stereo Visual Navigation Optimized by Local Segmentation and Maximum-Uncertainty Comparability Measurement

摘要：

Scale-Invariant Feature Transform (SIFT) is the most common feature detection and matching algorithm in binocular images for robotic Simultaneous Localization and Mapping (SLAM). This paper develops a SIFT-based robotic stereo visual navigation. Besides, this paper proposes a local segmentation method to increase a substantial number of feasible features, and adopts maximum-uncertainty comparability measurement to reduce the significant computation time of searching candidate matching features. Experimental results indicate the proposed methods can improve original SIFT-based robotic stereo visual navigation by over 400％ increment of feature amount, at the cost of 200％ increment of computatton time.

关键词： SIFT robot visual navigation uncertainty

作者: Chian C. Ho Chung-Lin Li Hsuan T. Chang Ching-Lung Chang

作者单位: Embedded SoC Laboratory, Department of Electrical Engineering, National Yunlin University of Science and Technology,Douliou, Yunlin County 64002, Taiwan, China

会议类型: 国际会议

会议名称: The Tenth International Conference on Information and Management Sciences(IMS)(第十届信息与管理科学国际会议)

会议地点: 拉萨

会议语种:英文

页码: 229-236

在线出版日期: 2011-08-06（万方平台首次上网日期，不代表论文的发表时间）