Improvement of CSF based on a wide range of urban complex scenes

Most airborne LiDAR point cloud filter algorithms are low-precision, ineffective and low-robust in mountain region. In order to improve the precision, efficiency and robustness in this region, a normalization CSF- modified algorithm presented in this paper based on CSF (Cloth Simulation Filtering). This algorithm has high precision and robustness in a wide range of complex scenes. In the first place, the pretreatment of point cloud reject gross error. Then, establish a grid index by grid and use the lowest point of each grid mesh surface equation. Thirdly, calculate the distance between raw point cloud and fitting surface, getting normalized point cloud. Finally use CSF algorithm to simulate filtering process, getting the final shape of cloth and filtering result obtained by the shape of cloth and limit of threshold. Use a big campus area to verify the algorithm, the result shows that the algorithm can effectively correct the top information of mountains removed by CSF algorithm and improve the accuracy and robustness of point cloud filtering.