Enhancing the UAV Localization Accuracy by Removing False Matches Using Affine Consistency

Unmanned Aerial Vehicles (UAVs) face significant challenges in localization within Global Navigation Satellite System (GNSS) denied environments, where traditional methods often produce false matches during image matching. This paper introduces a novel approach that integrates affine consistency with flight path verification to enhance image localization accuracy. By applying affine transformations, we initially reduce false matches, but residual inaccuracies remain. To address these, our method incorporates flight path data and flight headings, to further refine the matching process. Experimental results demonstrate a substantial reduction in false matches across diverse terrains, using images sourced from Google Maps and Bing Maps. The proposed approach achieved a zero false positive rate in all tested areas, including challenging regions such as a desert area with homogeneous and less distinct terrain features. The results highlight the efficacy of our methodology in improving UAV localization accuracy, making it a robust solution for GNSS-denied environments.