基于权重滤波的时空融合算法对比分析

Comparative Analysis of Spatio-temporal Fusion Algorithm Based on Weight Filter

多源遥感数据时空融合技术可生成高空间高时间分辨率的图像数据,有效解决了遥感应用中的遥感图像"时空矛盾"问题。ESTARFM和STNLFFM是目前常用的两种基于权重滤波的时空融合模型,本文对上述方法进行了对比分析,为模型选择提供依据。通过两组不同地表覆盖类型变化的实验,开展了模型精度对比分析,结果表明:当地表覆盖物时相变化属于物候变化时,两种算法融合结果均能还原出较为丰富的空间细节信息,且算法融合精度差异较小,效果相当;而当地表覆盖类型发生较大突变时,两种算法融合结果空间细节信息还原较少,同时,相比于ESTARFM,STNLFFM融合结果能够获取较高精度。

The multi-source remote sensing data spatio-temporal fusion technology can generate image data with high spatial and high temporal resolution, which effectively solves the problem of "time-space contradiction" of remote sensing images in applications. ESTARFM and STNLFFM are two commonly used time-space fusion models based on weight filtering. In this paper, comparing the above methods to provide a basis for model selection. Through the experiments of two different types of surface cover types, the comparison of model accuracy is carried out. The results show that when the phase change of the local cover is phenological change, the fusion results of the two algorithms can restore more spatial details. The difference of algorithm fusion precision is small and the effect is quite good. When the local table cover type has a large mutation, the spatial detail information of the two algorithm fusion results is less restored. At the same time, compared with ESTARFM, the STNLFFM fusion result can obtain higher precision.