The estimation of spatially varying albedo and optical thickness in a radiating slab using artificial neural networks

The inverse problem of estimating the single scattering albedo and optical thickness of a radiating slab from the knowledge of exit radiation intensities was accomplished by using an artificial neural network. Three training sets were developed which incorporated different amounts of training data. Simulated noisy measurement data was used to test the network. The network estimated the parameterized form of the spatially varying single scattering albedo and the optical thickness of the slab even when the amount of training data was limited to a few examples. The estimation of these parameters was more difficult, however, when the optical thickness was large.