Vision-based recognition for robot localization

Local scale-invariant features are used as natural landmarks in unstructured and unmodified environment. As autonomous robots, possessing visual acquisition capability is very crucial to explore unknown environments reliably, SIFT (Scale Invariant Feature Transform) key points are powerful in detecting objects under various imaging conditions, robot can use the recognized object as landmarks to navigate and localize itself. This paper presents a method to reduce the size, complexity and matching time of SIFT features in robot SLAM context. Experimental results demonstrate the effectiveness of the proposed algorithm.