Illumination change compensation and extraction of corner feature orientation for upward-looking camera-based SLAM

In this paper, the improved upward-looking camera based SLAM is proposed. With the corner features that are highly robust to the illumination change and contain their orientation information, SLAM performance can improve effectively. The illumination properties from the image can be rejected using the retinex algorithm to make the input image less sensitive to the illumination change. In addition, to deal with the correspondence problem between indistinguishable adjacent corner features, the orientation of the corner feature is additionally taken into account. These robust features are used as a landmark for the extended Kalman filter EKF-based SLAM to evaluate the robustness. A series of experiments show the improved results using the proposed scheme.