摘要:
星载激光雷达(Light Detection And Ranging,LiDAR)与合成孔径雷达干涉(Interferometric Synthetic Aperture Radar Digital,InSAR)数字高程模型(DigitalElevation model,DEM)的融合在林下地形估计、树高反演、DEM精度评价及修正等方面均有广泛的应用，实现二者的有效配准是融合应用的关键前提。现有的配准模型并未充分考虑不同地形因素影响下的平面偏移量与高程改变量变化情况。本文提出基于地形坡度的定权模型，并将现有配准模型线性化，利用最小二乘平差的方法求解偏移量。为验证配准方法的有效性，选用了西班牙地区四个具有不同地形条件和地表覆盖类型的试验区进行了测试，实验结果表明，配准后四个试验区的高程向均方根误差(root mean square error,RMSE)分别达到0.931m、1.381m、1.034 m、4.526 m，与配准前相比，精度分别提升33.02%、8.78%、37.67%、10.00%。实验结果验证了本文算法对复杂地形下星载LiDAR与InSARDEM配准的有效性。
Abstract:
Objectives: The fusion of spaceborne Light Detection And Ranging(LiDAR) and InSAR(Interferometric Synthetic Aperture Radar Digital) Digital Elevation model(DEM) is widely used in understory terrain estimation, tree height inversion, DEM accuracy evaluation and correction, etc. However, due to limitations in onboard platforms, imaging technology, and data processing, There exist deviations in plane position and height measurement in the InSAR DEM. Therefore, the effective registration of spaceborne LiDAR and InSAR DEM is the key premise of the fusion application. Methods: An improved registration model that fully considers the changes in plane offset and elevation variation under different topographic factors has been proposed, which includes a stochastic model and a functional model. The stochastic model is proposed based on the relationship between changes in terrain slope and plane offset, while the functional model linearizes the original registration model. Finally, The plane offset can be solved using the registration model through least squares adjustment method. In order to verify the validity of the registration method, four test areas with different terrain conditions and land cover types in Spain were selected for testing. Results: The experimental results showed that the RMSE of height difference in the four test areas reached 0.931m, 1.381m, 1.034m and 4.526m respectively after registration, and the accuracy was improved by 33.02%, 8.78%, 37.67% and 10.00% compared with that before registration. Among them, the contribution of slope greater than 45° to the overall registration of elevation changes in the 4 study areas reached 72% to 99%, while the registration accuracy decreases from 33.02%, 8.78%, 37.67%, 10.00% to 20.60%, 0.27%, 35.89%, 4.62% respectively for a canopy height of 15m compared to 5m. Conclusions: This method weakens the systematic error variation of height deviation caused by plane offset under the influence of terrain slope and aspect. The experimental results have verified the effectiveness of the model proposed in this paper for the registration of spaceborne LiDAR and InSAR DEM in complex terrains.
下载:
