Rethinking the Fourier-Mellin Transform: Multiple Depths in the Camera’s View

Remote sensing and robotics often rely on visual odometry (VO) for localization. Many standard approaches VO use feature detection. However, these methods will meet challenges if the environments are feature-deprived or highly repetitive. Fourier-Mellin Transform (FMT) is an alternative approach that has been shown to show superior performance in scenarios used remote sensing. One limitation of FMT it requires environment equidistant camera, i.e., single-depth. To extend applications multi-depth environments, this paper presents extended (eFMT), which maintains advantages with respect scenarios. robustness accuracy eFMT, we implement eFMT-based framework test toy examples a large-scale drone dataset. All experiments performed data collected challenging scenarios, such as, trees, wooden boards featureless roofs. The results eFMT performs better than settings. Moreover, also outperforms state-of-the-art algorithms, as ORB-SLAM3, SVO DSO, our experiments.