A STUDY OF HYBRID STRIP ADJUSTMENT BETWEEN UAV LIDAR POINT CLOUD AND STRUCTURE-FROM-MOTION POINT CLOUD

Abstract

Laser scanning as know as LiDAR can produce point cloud for creating high resolution and high-accurate digital elevation model. Nevertheless, after a lidar mission, the point clouds from  two adjacent flightlines in the overlapping area were found that there are discrepancy in height in the range of 5 – 20 centimeters. Therefore, these point cloud data needs to be treated e.g. via strip adjustment by hybrid processing in such a way that aerial photogrammetric block co-flight with laser scanning will first processed and be used to produce control features for later strip adjustment. The aerial images are organized for bundle block adjustment (BBA) using photogrammetric computer vision (PCV), together with the ground control points for improving accuracy. While doing the BBA, the key points will be produced from the image matching and lead to point cloud densification by multiple-view geometry of aerial images in Structure-from-Motion (SfM) technique. The SfM point cloud will be used to create tie-line for LiDAR strip adjustment to reduce those discrepancies. Nevertheless, the inadequate density of LiDAR point cloud (5 - 10 point/sqm), it is difficult to adjust the horizontal discrepancies and the drift of attitude. So, only the height discrepancies will be adjusted. In this research there are has two types of strip adjustments, the first one is the strip adjustment of  LiDAR point cloud not using SfM point cloud and the second is the strip adjustment LiDAR point cloud using SfM point cloud. The research’s result was found that the unadjusted LiDAR point cloud has RMSE_Z at 8.3 centimeters, the strip adjustment of LiDAR point cloud not using SfM point cloud has RMSE_Z at 9.2 centimeters and the strip adjustment of LiDAR point cloud using SfM point cloud  improve  the height accuracy upto RMSE_Z of 7.4 centimeters. So it can be concluded that the hybrid processing between UAV LiDAR mapping and aerial photogrammetric block could help improving vertical positioning accuracy of this LiDAR point cloud block.