Research on UAV Image Matching Algorithm Based on Improved AKAZE Feature and VFC Algorithm

Abstract

Aiming at the problems of UAV image matching difficulty, time-consuming and low matching efficiency of traditional AKAZE algorithm, an improved algorithm combining AKAZE and LATCH algorithm is proposed. The algorithm first uses the AKAZE algorithm to extract feature points from the image to ensure the accuracy of feature detection, then uses the LATCH algorithm to calculate feature descriptors, and finally uses Vector Field Consensus (VFC) algorithm to accurately match the original image extracted feature points, and eliminate Mismatch points in the rough matching process. Experimental research shows that: comparing and analyzing the feature point extraction time and matching accuracy of image matching, compared with the traditional SIFT algorithm, ORB algorithm, and AKAZE algorithm, the feature point extraction time is increased by about 30%, and the feature matching time is increased by about 22 %, the feature accurate matching effect is the best. The method in this paper has significant matching effect and matching efficiency, and is more suitable for UAV image matching.