摘要
景象匹配通过将飞行器实拍图像与机载的基准图像比较而进行导航。当飞行高度较低时,由地面大起伏引起的实时图几何变形会严重影响匹配性能,造成匹配精度下降甚至出现误匹配,这一问题严重制约了现有景象匹配系统的应用范围,比如无法应用于建筑林立的城镇地区和梯田、丘陵地带。本文首先推导了小视场内无几何变形的高空竖直透视投影模型;在此基础上,依据三视约束关系,把相邻两幅下视实时图像的特征赋给该透视模型所成的像,得到消除几何变形的特征模板,再和基准图进行匹配。因此,所提方法不仅能处理由飞行器姿态引起的实时图几何变形问题,而且适用于城镇等地面存在较大起伏的地形。理论分析和仿真实验表明,该算法取得了较好的结果,具有良好的环境适应能力,实用性较强。
The geometric distortion of real-time images caused by attitude aberration of aircraft or the undulation of terrain is one of the challenges in terms for low-altitude scene matching. Therefore, a novel method to eliminate geometric distortion for scene matching is proposed in this article. In the approach described, contribution mainly consists of two respects： first, a vertical perspective model restricted in a small visual field with no geometric distortion is introduced; second, a matching template is obtained for further matching by transferring homologous features extracted from a real-time stereo pair to the new perspective model. Simulation results have verified the effectiveness of the method, especially in the application to urban terrain scene matching.
出处
《航空学报》
EI
CAS
CSCD
北大核心
2008年第5期1288-1296,共9页
Acta Aeronautica et Astronautica Sinica
关键词
景象匹配
地形起伏
几何变形
三视约束
特征模板
scene matching
terrain undulation
geometric distortion
three-view constraint
feature template
