Aerospace Heterogeneous Image Matching Algorithm Based on Spatiotemporal Continuity

There are problems such as large differences in angles, colors, and resolutions in aerospace heterogeneous images. It is difficult to extract image features, and the image matching accuracy is relatively low, which has a negative impact on the precise target positioning of unmanned aerial vehicles (UAVs). Due to its characteristics of self-supervision, easy training, and high accuracy, the SuperPoint-SuperGlue algorithm has been widely applied in the field of image matching in recent years. However, the SuperPoint-SuperGlue algorithm cannot correct the mismatching phenomena in the continuous matching sequence according to the existing matching information, and there is still room for improvement in matching accuracy. From the perspective of engineering applications, this paper makes full use of the temporal continuity and spatial continuity of UAV reconnaissance images and proposes an aerospace heterogeneous image matching algorithm based on spatiotemporal continuity. The algorithm in this paper can use the historical matching information of UAVs to correct the mismatching phenomena during the flight, and it is more suitable for the matching scenarios of aerospace heterogeneous images. Through experimental verification, on the uav visloc dataset, the number of matchable images within the error interval of 100 pixels can reach 81.07%. Compared with the original SuperPoint algorithm, the number of matchable images within the error interval of 100 pixels has increased by 3.82%, effectively improving the matching accuracy of aerospace heterogeneous images.