复杂地形下基于三视约束的景象匹配方法

 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。

Novel Scene Matching Method for Complex Terrain Based on Three-view Constraints

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 ob...