Image target identification of UAV based on SIFT

Abstract

The paper finds a method to meet the large amounts of information and real-time processing of the UAV (Unmanned Aerial Vehicle) image, it could increase efficiency of target identification. The SIFT algorithm has good robustness, it could overcome some affect of image deformation and block, but it could not meet the real-time processing of UAV image. The paper use simplified Forstner operator to improve SIFT algorithm, reduce the computation of feature point recognition. Through the simulation, we prove that the improved SIFT algorithm could meet the accuracy and speed requirements of UAV image in the complex background. © 2011 Published by Elsevier Ltd.