The Use of Geometrical Methods in Multispectral Image Processing

摘要

The information geometric model of multispectral images whose individual components are obtained in different spectral ranges of carrier radiation of video information at non-dotty correspondence of the object and the image caused by diffraction phenomena has been proposed. The proposed techniques allows one to restore position parameters of forming and wavelength of a spectral range for the fixation of analyzed image. The new combined invariants allow one to identify objects in a fuzzy scene without restoring or filtration.