Landsat-8 Sensor and Sentinel-2 Sensor Data Fusion With Multiscale Detailed Information

Abstract

With the increasing demand for high temporal and spatial resolution multispectral data sequences, many studies have been carried out on fusion on Landsat-8 and Sentinel-2 sensor data. However, current fusion methods suffer from the loss of detailed spatial and spectral information. To address this problem, a Landsat-8 and Sentinel-2 data fusion with multiscale detailed information (MSDI) method is proposed. MSDI combines well the initial spatial prediction obtained from the Landsat-8 data at the target date and the detailed part extracted from the Sentinel-2 data at the reference date. Thin plate spline interpolation is implemented on the Landsat-8 data for upsampling. Smoothing-sharpening filter (SSIF) is employed to separate the high- and low-frequency components of data from the two sensors. The multiscale SSIF is then utilized to migrate the details from the Sentinel-2 data to the upsampled Landsat-8 data. Experiments at two sites confirm that the proposed MSDI method could efficiently generate Sentinel-2-like data with high spatial and spectral resolution.