Retrieval of land surface emissivity using spectral and texture features based on neural network

Land surface emissivity is one of the most important parameters in thermal infrared remote sensing and plays a significant role in the quantitative study of thermal infrared remote sensing, the surface energy balance, and feature mapping. Retrieving the land surface emissivity from thermal infrared remote sensing data is a challenge because it presents an ill-posed problem. In this work, a method, which takes advantage of spectral and texture features of many visible channels available in the moderate resolution imaging spectroradiometer (MODIS) data and is based on back-propagation neural network to obtain land surface emissivity pixel-by-pixel, is presented. The method obtains the land surface emissivity map without the categorization of the land cover and the analysis indicates that the average error, compared to MODIS emissivity product, is within 0.002. It builds a direct relationship between reflectance data and emissivity data, and provides the possibility of obtaining precise emissivity data through the single channel thermal infrared satellite.