Scholarly record

COMPARISON OF DIFFERENT NOISE REDUCTION METHODS FOR INTERFEROMETRIC SAR IMAGES

Justyna Bała

Abstract

The reducing noise is one of the most challenging problem in digital image processing. This is a fundamental issue both in satellite images, medical images and other application connected with computer vision and pattern recognition. Many numerical methods have been adapted for noise reduction problem. In this article we describe filtering techniques, which were developed for the problem of removing noise from the interferometric SAR images. The problem of SAR interferogram noise reduction is especially important for detection of subsidence troughs. Different types of noise can be observed in the SAR interferograms. Speckle noise as well as low-amplitude Gaussian noise are the most common ones. Speckle type noise fills relatively large areas. Due to its characteristics, this type of noise is difficult to remove without significant interference in no-noise areas. The noise structure observed on SAR interferograms may be similar to the image of subsidence troughs and therefore be a source of false alarms. On the other hand structure of noise observed on SAR interferograms may hide the location of real subsidence areas. Optimal filtration should maximize noise reduction and minimize changes in subsidence areas. The appropriate method of filtration should work in two ways: allow to reduce noise with simultaneous reinforcement of subsidence troughs. In this article difference filtration method were tested. SAR images were denoised in spatial domain (median filtration) and spatial-frequency domain (wavelets filtration). Evaluation of chosen filtering methods have been carried out using synthetic SAR interferograms.

Publication Impact Profile

• Captures
• Mendeley - Readers: 4

PLUMX see details

DOI resolver: doi.org/10.5593/sgem2019/2.1/s08.104

Publication details

Back to publication list