Features of the formation of pseudo-hyperspectral images from RGB components

Abstract

The article discusses the formation of spectral images in television multispectral and hyperspectral systems. A hyperspectral data cube contains more information than a multispectral system cube, however, hyperspectral systems have limitations in the spatial and temporal domain. The existing correlation between hyperspectral and multispectral information and the availability of data on the reflectivity of the studied scene make it possible to build a pseudo-hyperspectral system based on a multispectral one with a limited number of visualized spectral channels. At the same time, as a result of post-session processing, a limited number of multispectral images are converted into pseudo-hyperspectral images corresponding to several hundred spectral channels. The minimum multispectral system can be considered a television system consisting of three channels R, G and B. To implement the procedure for obtaining pseudo-hyperspectral images at the first stage of processing R, G, B components, it is necessary to implement methods to improve their color resolution and ensure white balance. The article considers the possibility of using the video information processing method based on the theory of nonlinear two-component vision of the Soviet engineer S. D.Remenko for the above problem.