基于ASIFT的飞行器辅助导航图像匹配技术

摘要

对一种新型算法仿射尺度不变特征变换(Affine-SIFT,ASIFT)算法进行研究,将完全仿射不变性引入尺度不变特征变换(SIFT)中,对飞行器光学成像辅助导航中参考图像与实时获取图像存在的不同尺度、不同视角、不,同成像条件、甚至不同传感器条件下的灰度特征差异性有很好的鲁棒性,并与现阶段仿射不变性方面最优的SIFT、最大稳定极值区域(MSER)等算法比较,实验结果表明,ASIFT算法大大提升了在大视角变换时的图像匹配的成功率和准确率.%The Affine Scale-Invariant-Feature-Transform (SIFT) algorithm extends the Scale-Invariant-Feature-Transform (SIFT) method to the fully affine invariant.ASIFT matches the real-time images and reference images which usually obtains in different scale,different view point,different imaging codition,even different CCD has a good robustness.The classical algorithms of image matching hardly get correct matches in these disadvantageous conditions.Affine-SIFT (ASIFT) compares with the classical SIFT and MSER which considers as the best algorithm at aspect of the affine invariant.The Simulation results show that ASIFT method provides robust image matching during optical aided navigation and is more robust than traditional methods.