LAUSR

ABSTRACT Land remote-sensing satellite (System, Landsat) 8 operational land imager (OLI) can provide land cover target mapping (LCTM) information. Due to the limitation of hardware and complexity of environment, Landsat 8 OLI image sometimes contains a large number of mixed pixels, which brings a challenge for LCTM. Super-resolution mapping (SM) handles the mixed pixels to obtain LCTM at subpixel scale. However, the scale information is single and the infrared information is not rich in the existing SM methods. In order to solve these issues, this paper proposes SM based on multiscale-infrared information (MII). There are two terms (i.e. multiscale and infrared terms) in MII. The multiscale term with multiscale information is produced through deep Laplacian pyramid network (DLPN), multiresolution segmentation, and extended random walker in turn. The infrared term with infrared information is derived by calculating the normalized difference target index (NDTI). The two terms are combined to generate a minimization term with multiscale-infrared information. Particle swarm optimization algorithm is applied to the minimization term to obtain the LCTM result. The experimental results on four Landsat 8 OLI images show that the MII provides better LCTM results than the traditional SM methods.