惯性信息辅助的大视角目标快速精确定位

发布时间：2018-03-29 05:24

本文选题：目标定位　切入点：惯性导航　出处：《航空学报》2017年08期

【摘要】：目标定位技术广泛应用于航空领域的侦察机、无人机等各类侦察打击任务中,目标定位精度的高低及效率对作战效果具有重要影响。针对仿射尺度不变的特征变换(ASIFT)算法对远距离大视角目标定位精度较低、速度较慢的问题,提出了一种基于惯性信息辅助的大视角目标快速精确定位方法。该方法首先对目标实测序列图像构造尺度空间,结合FAST特征检测与FREAK特征描述的方式进行匹配,实现对待定位目标的快速提取;然后利用机载惯性信息求解目标实测图与参考图之间的透视变换矩阵,利用该矩阵对实测图进行变换以减小图像间视角差异,克服了ASIFT算法盲目匹配计算的弊端,并通过FAST特征检测与FREAK特征描述相结合的方式提升了大视角图像的匹配速度;最后通过单应性矩阵映射关系实现对目标的精确定位。实验结果表明,大视角目标快速精确定位方法匹配耗时比ASIFT算法的减小了1个数量级,定位精度比目标平均值定位算法精度提高了1个数量级,有效提高了图像匹配定位在航空领域的应用效率。
[Abstract]:Target positioning technology is widely used in reconnaissance aircraft, UAV and other reconnaissance and strike missions in the aviation field. The accuracy and efficiency of target positioning have an important impact on the combat effectiveness. The problem of low precision and slow speed of target location with long range and large viewing angle is solved by the affine scale invariant feature transform (ASIFT) algorithm. In this paper, a fast and accurate target location method based on inertial information is proposed. Firstly, the scale space is constructed for the target measured sequence images, and the FAST feature detection is combined with the FREAK feature description. Then the airborne inertial information is used to solve the perspective transformation matrix between the measured image and the reference map of the target, and the matrix is used to transform the measured image to reduce the visual angle difference between the images. It overcomes the disadvantage of blind matching calculation of ASIFT algorithm, and improves the matching speed of large view image by combining FAST feature detection with FREAK feature description. Finally, the homography matrix mapping is used to realize the accurate location of the target. The experimental results show that the matching time of the fast accurate location method of large angle target is 1 order of magnitude less than that of the ASIFT algorithm. The accuracy of image matching is improved by one order of magnitude compared with that of the average target location algorithm, and the application efficiency of image matching in aviation is improved effectively.
【作者单位】：南京航空航天大学自动化学院;卫星通信与导航协同创新中心;中航工业洛阳电光设备研究所;
【基金】：国家自然科学基金(61533008,61104188,61374115,61603181) 江苏省普通高校研究生科研创新计划项目(KYLX15_0277) 中央高校基本科研业务费专项资金(NS2015037)~~
【分类号】：E926.3;TP391.41

【相似文献】