Virtual sensors for remote sensing: Algorithms and performance

Abstract

Consider data collected by two types of sensors with disparate capabilities. A technique is presented for constructing a high quality image from an image measured by a low quality sensor, given training sets of high quality and low quality images of related scenes. The method involves calculation of a non-linear transformation using the training data. The transformation is then applied to the low quality image to produce a ldquovirtualrdquo high quality image.