Computer vision algorithms based on image processing in various wavelength ranges (spectrozonal photography) are widely used in modern control systems, navigation, security, diagnostics, robotics, artificial intelligence for aviation, land and sea transport, as well as in various medical diagnostic applications. The aim of research is to develop an information approach to solving the problem of converting RGB data of primary images into spectrum estimation (hyperspectral data) in the presence of a priori information about the original spectrum, as well as additional weakly correlated information from closely located similar RGB-cameras to estimate the parameters of the original spectrum of the image. To reduce errors caused by the limited dynamic range of sensitivity of camera color sensors, it is desirable to use the RAW format with subsequent normalization of the color component values.