An Object-Based Strategy for Improving the Accuracy of Spatiotemporal Satellite Imagery Fusion for Vegetation-Mapping Applications

摘要: Spatiotemporal data fusion is a key technique for generating unified time-series images from various satellite platforms to support the mapping and monitoring of vegetation. However, high similarity in reflectance spectrum different vegetation types brings an enormous challenge similar pixel selection procedure spatiotemporal fusion, which may lead considerable uncertainties fusion. Here, we propose object-based data-fusion framework replace original with object-restricted method address this issue. The proposed can be applied any algorithm based on pixels. In study, modified spatial temporal adaptive model (STARFM), enhanced (ESTARFM) flexible (FSDAF) using framework, evaluated their performances fusing Sentinel 2 Landsat 8 images, Moderate-resolution Imaging Spectroradiometer (MODIS) MODIS study site covered by grasslands, croplands, coniferous forests, broadleaf forests. results show that improve all three algorithms significantly delineating boundaries more clearly, improvements FSDAF greatest among algorithms, has average decrease 2.8% relative root-mean-square error (rRMSE) sensor combinations. Moreover, improvement significant (an 2.5% rRMSE). By fused generated result reducing “pepper-salt” effect. We believe great potential used high-resolution remote-sensing applications.