SLAM Localization System Based on Sub-Pixelation of Selectable Feature Points

In order to compensate for the lack of accuracy of the results of feature point-based Simultaneous Localization and Mapping (SLAM) algorithms due to the problem of feature point accuracy, this paper proposes an optimization method for selectable sub-pixel feature points. The method first determines the number of feature points extracted from the image, and then subpixelates the feature points that satisfy the conditions. Such processing improves the accuracy of the algorithm while avoiding the long running time of the algorithm due to the overall subpixelation of feature points, making the feature point-based SLAM localization algorithm achieve a balance between running speed and accuracy. The method is experimented using the EuRoC public dataset and verified to significantly improve the accuracy of the SLAM algorithm by comparing the experimental results with those of A Robust and Versatile Monocular Visual-Inertial State Estimator (VINS-mono).