Simulation model for the identification of environmental pollution from hyperspectral imagery in the visible range
V.G. Burlov1, V.N. Ostrikov2, A.V. Kirienko2, O.V. Plakhotnikov2
1Russian State Hydrometeorological University, St. Petersburg, Russia
2Design Bureau Lutch, St. Petersburg Branch, St. Petersburg, Russia
Keywords: hyperspectral imaging, remote sensing, simulation, environmental pollution
Abstract
A mathematical model is proposed that simulates the acquisition and processing of hyperspectral remote sensing data on subtle fragments of environmental pollution (garbage) comparable in size with the spatial resolution of the observation equipment. Spectral mixing of “objects” with “background” is provided by a special coefficient, which takes into account that the area of each element of a scene template related to the “object” is only partially filled with its spectral characteristic, and the rest of the area, with the background characteristic. The options for calculating the probability of detecting objects depending on the observation conditions specified with the MODTRAN atmospheric model are considered. The difference between the model data and real experimental results is no more than 10%.
|
