Correction of distortions and improvement of information content in satellite-acquired multi spectral images

V. I. Voloshyn and V. M. Korchinsky

Dnipropetrovsk, Ukraine



For a brief description of the methods employed in the present study, we perform image reconstruction from employing one reference image selected within the spectral range under consideration. The reconstruction is performed using an information-geometrical model describing the physical process of image formation in the short wave range of electromagnetic spectra (visual and neighboring infrared). The reconstruction of image pixel intensities is performed by arranging in ideal projection planes (PP) the locations needed for representation in different spectral channels to analyze and validate the available data. The model for reconstruction of the multispectral images was developed from physical principles. Scanner method was applied with coordinate-susceptibility photoreceivers (CSPR). A physics-based model for the formation of multispectral raster images with CSPR scanners in the electro-optic and infrared ranges can be developed by taking into account the tuning of optic-mechanic scanner scheme.  By using basic optic principles, the conformity between points on the earth surface and scanner sensitivity element (SE) can be established. One may note that each SE is fixing the electromagnetic radiation from some site as any cell projection on the earth surface where the center of projection is situated in the optic object-lens center.


Integration is performed on each cell square. This gives the mathematical model for the formation of digital images from CSPR equipped scanners in the multi spectral regime. It also enables explaining the reasons for distortions in these images. It may be noted that different projection distortions caused by optic-mechanic tuning scanner lead to displacements at the subpixel level. Due to construction peculiarities characterizing the satellite and the sensor, the intensity distribution captured on the image plane will suffer distortions due to mixing of images collected along the different spectral channels. A method for the correction of distortions is to select the intensity distribution on one channel as the basic image and construct the intensity distribution data on other channels by image processing operations. Thus the method offers correction in the case of point displacement between different spectral channels due to diffraction effects and spatial coordinate-susceptibility of photo receivers causing positioning instability during sensing.