GPS-assisted Aerial Image Stitching Based on optimization Algorithm

Splicing aerial images makes it faster and more comprehensive to detect the environment of the target area. In this paper, a GPS-assisted image stitching method based on optimization algorithm is proposed for the aerial images of UAV. Predicting the feature points to be extracted in the next key frame by using GPS, the searching time can be greatly reduced and the registration rate can be improved. After obtaining the matched feature point pairs, the pixel difference value of the feature set center is calculated as the initial value of the translation amount, and the re-projection error is minimized by the L-M iterative algorithm to establish the best affine transformation, which is more accurate and less time-consuming than the traditional RANSAC method. The algorithm adjusts the threshold of selecting key frames to adjust the number of feature points between frames, so that it is adapted to a variety of outdoor scenes in real-time.