METHOD OF FILTERING ARTIFACTS ON MULTISPECTRAL DIGITAL IMAGES OF PROJECTION NATURE

Abstract

A method is proposed to reduce graphic interference (artifacts) on digital images of projection nature, obtained in an arbitrary number of spectral intervals of radiation – the range of specific information or under different positional conditions of projection. The algorithmic basis of the proposed method is the compression of the brightness distributions of spectral bands, optimized according to the criteria for minimizing the deviation of the ratio of signal energies of the compressed image and noise from a given value.

The set of images of different spectral bands is represented as a single multidimensional geometric object, which is described by a two-dimensional data mass ordered by raster and spectral intervals.

Suppression of artifacts with simultaneous compression of digital image brightness codes is implemented on the basis of decompositions of these codes according to discrete orthonormal functional bases, zeroing of part of the expansion coefficients and subsequent reconstruction of image brightness distributions. Determination of the thresholds for zeroing the expansion coefficients of digital brightness levels in the form of an optimization problem for achieving a given ratio of signal energies of the compressed signal and residual noise.

The implementation of the proposed method includes the following steps: combining the brightness codes of spectral bands in a single two-dimensional graphic object (raster image); Peano-Hilbert scan of the combined image with the receipt of a one-dimensional digital signal; compression of digital sweep values according to the specified criteria; reconstruction of the brightness codes of compressed images of spectral bands by a functional transformation inverse to that used at the stage of decomposition of the brightness distributions of directly recorded images of spectral bands.

Comparative analysis of various discrete functional bases as a basis for compression of brightness distributions showed the greatest efficiency in terms of the criteria of the discrete functional basis of Walsh.

Keywords: multispectral image, signal-to-noise ratio, orthonormal discrete functional basis.