基于结构特征的机载LiDAR数据与航空影像自动配准

针对现有机载LiDAR数据与航空影像配准方法对匹配特征具有较强的依赖性，易受数据等影响的问题，提出了一种基于结构特征的自动配准方法。该方法首先提取LiDAR距离图像与对应影像的结构特征，利用初始姿态参数将LiDAR结构特征投影至影像坐标系下，根据结构特征的几何约束条件获取初始匹配点集，完成粗匹配；接着利用粗匹配结果计算直接变换模型（DLT）参数，并以此为初值引入双点几何约束，采用循环迭代的匹配策略，不断剔除错误匹配，获得一组新的匹配点集，完成精匹配；最后根据精匹配结果，采用基于单位四元数的空间后方交会方法解算航空影像的姿态参数，实现机载LiDAR数据与航空影像的自动配准。实验证明，该方法受噪声影响小，能实现机载LiDAR数据与航空影像的自动配准。

Auto-registration of aerial imagery and airborne LiDAR data based on structure feature

Current algorithms of registration of aerial imagery with airborne LiDAR data has the major issue of strong dependency upon the matching feature, so these methods are impressionable to the texture feature of image and the density of LiDAR point cloud. A new method of auto-registration of aerial imagery with airborne LiDAR data based on structure feature was proposed. The first step was the automated extraction of structure feature from LiDAR range image and aerial imagery. After that the LiDAR structure features were projected onto aerial imagery and corresponding features were determined using geometry constraints. The second step was the wrong matches eliminating by two points geometric constraint after calculating the DLT parameters as the initial value, and iteration strategy was adopted to obtain optimal results. The last step was the pose parameters calculated by the optimal matching results using quaternion-based solution of space resection. Experimental studies have demonstrated that this algorithm is effective in auto-registration of aerial imagery with airborne LiDAR data and little influenced by noise.