This paper addresses the seeking of an identification algorithm for image target fitting UAV features to increase the efficiency of image identification according to the characteristics of UAV that the image information captured is huge and the real-time processing requirement is high. SIFT algorithm has good accuracy and robustness and can overcome some effect of image deformation and occlusion, but it is still hard to achieve real-time processing of UAV images. In the paper we use simplified Foretner operator to improve SIFT algorithm, reduce the computation of feature point recognition process in it. Through the simulation experiment, it proves that the improved SIFT algorithm can raise the accuracy and matching speed of the identification, meet the requirements of UAV in its target identification precision and speed in complicated background.%根据无人机获取图像信息量大、处理实时性要求高的特点,寻找一种符合无人机特点的图像目标识别算法,提高图像识别的效率.SIFT(Scale Invariant Features Transform)算法具有良好的准确性和鲁棒性,能够克服一定的图像形变及遮挡影响,但其还难以满足无人机图像的处理实时性,利用简化的Forstner 算子对SIFT 算法进行改进,降低SIFT算法特征点获取过程的计算量.通过仿真实验,证明改进的SIFT 算法可以提高识别和匹配的速度和准确率,可以满足复杂背景下无人机目标识别精度与速度的需求.
展开▼
