Combined methods of radiometric correction of typical artifacts on satellite images

Аuthors

Barabin G. V.^1*, Gusev V. Y.^**, Zaitsev V. E.^2***, Yurov N. N.²

1. Research proizvodstvennay firm «Infosystem», 16, 3rd Mytischinskaya Str., Moscow, 129626, Russia
2. Moscow Aviation Institute (National Research University), 4, Volokolamskoe shosse, Moscow, А-80, GSP-3, 125993, Russia

*e-mail: gbarabin@gmail.com
**e-mail: gusev3@mail.ru
***e-mail: umc8@mai.ru

Abstract

Purpose
The aim is to create new methods for correcting brightness distortions in the satellite images. The distortions caused by non-uniformity of the sensitivity of photo detectors, or faulty settings. This is shown in the image in the form of vertical stripes. This method needs to be robust to work with images with very high natural changes in brightness caused by different types of terrain and clouds.
Design/methodology/approach
Proposed two-stage correction methods that take into consideration local characteristics of the image. First stage provides good local correction by taking into consideration local characteristics of image. Second stage provides global control of a general change in brightness across the whole image.
Findings
The methods of removing the vertical stripes on the images with higher reliability due to additional compensation after the first stage of processing. Methods take into consideration the local statistical characteristics of the columns, which allows for smooth adjustment and reduce the impact on the determination of the quantities of natural changes in brightness adjustments due to the terrain. Additional control using moving average of the average brightness of the columns to match the original and the corrected image eliminates the possibility of incorrect correction on a global scale. Another advantage of the proposed methods for processing satellite images of large size is their low time complexity that is linearly dependent on the number of columns.
Research limitations/implications
The methods can be applied at the stage of initial processing of satellite images for automatic correction of vertical stripes. Methods belong to the level of processing 1B according to international classification, which includes radiometric correction and geometric correction of systematic errors of CCD sensors scanning system.
Originality/value
Many existing methods for brightness adjustment on the columns are not ideal, and not always able to take into account the strong natural variations in the brightness of the image because of the terrain or clouds. Suggested methods, thanks to the local image analysis, image processing make it possible to significantly dissimilar in texture and brightness.

Keywords:

radiometric correction, brightness correction, brightness correction, image segmentation, segmentation, local image analysis

References

1. Zlobin V.K., Eremeev V.V. Obrabotka ajerokosmicheskih izobrazhenij (Processing ofspace images), Moscow, FIZMATLIT, 2006, 288p.
2. Gonzalez R., Woods R. Cifrovaja obrabotka izobrazhenij (Digital image processing ), Moscow, Tehnosfera, 2005, 1072p.
3. Forsajt, Djevid A., Pons, Zhan. Kompjuternoe zrenie. Sovremennyj podhod (Computer vizion. Modern approach), Moscow, Williams, 2004, 928p.
4. Gusev V.U. Geodezija ikartografija, 2012, no.12, pp. 111-114.