研究目的
Investigating the method for the reconstruction of the sensor trajectory from a multi-echo ALS point cloud acquired over various forests.
研究成果
The method for trajectory reconstruction from an airborne lidar point cloud presented in this article uses the intersection of multi-echo pulses which direction is estimated with first and last echoes. The method was tested on three ALS datasets acquired with different sensors and various forest environments (temperate and tropical deciduous, mountainous coniferous). The trajectory inversion of a flightline of St Gobain show a certain sensitivity to dmin and Nest, although we observe a stable performance with dmin in the range [5, 20] m and Nest in the range [100, 400] pulses. With these parameters, the average error of location of the sensor stays inferior to 25 cm and a scan angle error inferior to 2·10-2°, which seems an acceptable accuracy for the applications aforementioned.
研究不足
The method shows a certain sensitivity to dmin and Nest, and the errors of estimate are not equally distributed among axes. The bias could be linked to a lever arm between the IMU/DGPS and the sensor itself that would not have been corrected on the reference trajectories.
1：Experimental Design and Method Selection:
The method is based on the intersection of multi-echo pulses and was tested on three data sets acquired over a deciduous, a tropical and a mountainous forest, respectively. It involves an analytical resolution of the minimal distance between a sensor defined as a unique point, and a set of lines corresponding to the pulses of a scanline, based on the least square criterion.
2：Sample Selection and Data Sources:
The data used in this study are extracted from ALS acquisitions over three different sites: the deciduous forest St Gobain in France, the mountainuous coniferous forest of Prenovel in France, and the tropical forest at Paracou Research Station in French Guyana.
3：List of Experimental Equipment and Materials:
Lidar data were delivered in LAS 1.2 format. Sensor trajectories, produced from the inertial measurement unit (IMU) and the DGPS, were specifically requested and were delivered in ASCII file including X, Y and Z coordinates of the sensor at each IMU time-stamp.
4：2 format. Sensor trajectories, produced from the inertial measurement unit (IMU) and the DGPS, were specifically requested and were delivered in ASCII file including X, Y and Z coordinates of the sensor at each IMU time-stamp.
Experimental Procedures and Operational Workflow:
4. Experimental Procedures and Operational Workflow:
From the LAS data, only X, Y and Z coordinates of the echoes, as well as the corresponding time-stamp of laser pulse emission (known as gpstime) were used. Scanlines and flightlines are identified a posteriori based on differences in time-stamps between successive pulses greater than Δtsl = 1 ms and Δtfl = 30 s respectively.
5：Data Analysis Methods:
The residual angle between the estimated direction nj, and the reference direction is analyzed, as well as the sensitivity of dmin and Nest on the sensor location estimate.
独家科研数据包，助您复现前沿成果，加速创新突破
获取完整内容
