Image feature tracker for SLAM with monocular vision

In this paper, an image feature tracking algorithm is proposed for improving the data association in robot visual Simultaneous Localization and Mapping (SLAM). The detection of speeded-up robust features (SURF), a scale-invariant feature, is employed to provide a robust description for image features. However, to match the high-dimensional data sets created for SURF, the conventional nearest-neighbor (NN) method does not seem to provide a robust tool in dynamic environment. An algorithm based on Shi-Tomasi tracker is utilized to overcome the problem of unstable feature tracking. Experiments are carried out on a hand-held camera to verify the proposed algorithm and the results show that the performance of the feature tracking algorithm is efficient for dealing with data association problem in visual SLAM.