加入地面控制点进行几何精校正后，采用非线性最小二乘算法不断优化生成稀疏

After addition of ground control points of geometric correction, using the nonlinear least squares algorithm constantly optimize sparse point cloud generation, and finally through intensive matching algorithm by matching pixel dense point cloud generation, complete recovery has relative coordinates of the 3D point cloud.

After adding the ground control point for geometric fine correction, the sparse point cloud is continuously optimized by nonlinear least square algorithm, and finally the dense point cloud with relative coordinate series is generated by dense matching algorithm, and the recovery of the three-dimensional point cloud with relative coordinates is completed.

After adding ground control points for geometric precision correction, non-linear least square algorithm is used to optimize and generate sparse point cloud. Finally, dense point cloud is generated by pixel matching through dense matching algorithm to complete the recovery of 3D point cloud with relative coordinates.