Image features and seasons revisited

We present an evaluation of standard image features in the context of long-term visual teach-and-repeat mobile robot navigation, where the environment exhibits significant changes in appearance caused by seasonal weather variations and daily illumination changes. We argue that in the given long-term scenario, the viewpoint, scale and rotation invariance of the standard feature extractors is less important than their robustness to the mid- and long-term environment appearance changes. Therefore, we focus our evaluation on the robustness of image registration to variable lighting and naturally-occurring seasonal changes. We evaluate the image feature extractors on three datasets collected by mobile robots in two different outdoor environments over the course of one year. Based on this analysis, we propose a novel feature descriptor based on a combination of evolutionary algorithms and Binary Robust Independent Elementary Features, which we call GRIEF (Generated BRIEF). In terms of robustness to seasonal changes, the GRIEF feature descriptor outperforms the other ones while being computationally more efficient.