Research of tracking robot based on SURF features

Abstract

A tracking algorithm using Kalman Filter (KF) based on Speed-up Robust Features (SURF) is proposed. In the tracking process, the SURF features in the frames are matched by RANdom SAmple Consensus (RANSAC) with objective template, and the Fuzzy C-Means (FCM) clustering analysis is used to eliminate the mismatching cases. Then the precise position of the object is determined, and the KF prediction of the object position is calibrated. The experiment results verify the robustness and real-time of the proposed method.