Real-Time Feature Acquisition and Integration for Vision-Based Mobile Robots

In this paper we propose a new system for real-time feature acquisition and integration based on high-resolution stereo images that is suitable for mobile robot platforms with limited resources. We combine a fast feature detection stage with a stable scale-invariant feature description method utilizing optimized spatial matching. Putative image feature matches are used to determine 3D coordinates of feature points and to estimate corresponding view transformations. Experimental results show the advantages of our system in terms of performance and accuracy compared to the standard methods used in the area. Key words: Feature acquisition, feature integration, vision, mobile robot

• Type: Conference
• Year: 2009
• Where: ISVC
• Authors: Thomas Hübner, Renato Pajarola
• Keywords: Applied Computing, Fast Feature Detection, Feature Acquisition, ISVC 2009, Real-time Feature Acquisition