多源DEM空间域加权最优融合方法研究

Study on Weight Optimal Fusion of Multi -Source DEM in Spatial Domain

全球许多区域都拥有不同类型、来源和精度的数字高程模型（DEM）数据，为利用多源DEM的互补信息开发出质量更高的DEM，提出多源DEM的空间域加权最优融合方法。以秦岭典型高山峡谷地貌类型区为试验样区，选取相同位置的航天飞机雷达地形测绘任务（SRTM）与先进星载热发射和反射辅射仪全球数字高程模型（ASTERDEM）数据，并以1：5万DEM作为参照数据，通过重采样、数据配准、系统误差消除等步骤形成融合数据源，以均方根误差（RMSE）性能指标最优为准则遍历权重系数，生成融合DEM。将融合前后的数据分别与参照数据作精度比较，总体统计与抽样剖面检查表明：融合DEM精度较源数据均得到了提高，该融合技术为应用多源DEM生成精度和可靠性更高的DEM产品提供了可行方案。

There exist multiple DEMs of different kinds and of various accuracies in most regions of the world. In order to develop DEM with better quality with complementary information provided by multi - source DMEs, a weight optimal fusion of multi - source DEM in spatial domain is put forward. The typical physiognomy of the Qinling mountains is chosen as region of in- terest, SRTM and ASTER DEM data of the same areas are selected and 1：50,000 -scale DEM is used as reference data. The fused data source is formed after preprocessing SRTM and ASTER DEM data by re - sampling, co - registering and systemic error eliminating. Then, the performance index of RMSE is optimized through ergodic process of weight, and the fused DEM is pro- duced. The accuracy of data before and after fusion are compared with 1：50,000 DEM respectively, and the collectivity statistic and spot check results show that the accuracy of fusion DEM has improved compared with that of original DEM. It proves that the weight optimal fusion provides an effective and reliable method for producing DEM from multi - source DEM.