摘要
大视角图像匹配算法的鲁棒性与实时性直接影响飞行器对远距离目标定位的性能。针对目前仿射不变图像匹配算法实时性较差的问题,提出一种惯性信息辅助的快速大视角图像匹配方法。该方法对现有的快速图像匹配算法进行改进,避免了构建高斯金字塔,提高了算法效率。然后利用机载惯性导航信息求解实时图与参考图之间的单应性矩阵,并对实时图进行模拟视角变换以此减小图像间视角差异,克服了现有的大视角图像匹配算法盲目多次的匹配计算,实现了大视角图像的快速匹配。实验结果表明,惯性信息辅助的大视角图像匹配算法与现有的快速仿射不变性匹配算法相比,匹配效率提高了至少2倍。
The robustness and efficiency of large viewpoint image＇s matching method immediately influence the performance of target location. Aiming at low efficiency of the affine invariant matching method, a fast matching method for large-viewpoint images aided by inertial information is proposed. The existing fast matching method is improved by the proposed method, which avoids building a Gaussian pyramid and improves the algorithm efficiency. Then the onboard inertial information is used to calculate the homography matrix between real-time image and reference image. The homography matrix is used to transform the real-time image so that the viewpoint change can be decreased. The blindness and quantity of the existing affine invariant matching methods are overcome. Experiment results show that the efficiency of the proposed method is improved by at least 2 times compared with the existing fast affine invariant matching methods.
作者
王云舒
刘建业
曾庆化
刘昇
WANG Yun-shu LIU Jian-ye ZENG Qing-hua LIU Sheng(College of Automation Engineering, Nanjing University of Aeronautics and Astronautics, Nanjing 211106, China Satellite Communication and Navigation Collaborative Innovation Center, Nanjing 211106, China AVIC Luoyang Electro-optical Equipment Research Institute, Luoyang 471009, China)
出处
《中国惯性技术学报》
EI
CSCD
北大核心
2016年第4期504-510,共7页
Journal of Chinese Inertial Technology
基金
国家自然科学基金项目(61533008,61328301)
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资助(NS2014092)
关键词
大视角图像
仿射不变性
惯性导航
图像匹配
目标定位
large viewpoint image
affine invariant
inertial navigation
image matching
target location
