K-means based noisy SAR image segmentation using median filtering and otsu method
Enugula Niharika, Hafsah Adeeba, A. Shiva Rama Krishna, P. Yugander · 2017
In this paper we propose a segmentation algorithm for noisy Synthetic Aperture Radar (SAR) images. This method is based on k-means (KM) clustering and thresholding techniques. SAR images have huge employment in topography, remote sensing, and subsurface imaging. The segmentation of SAR images is always demanding because of the noise present in it. Speckle noise is the common noise present in SAR images. The median filter is utilized for noise removal. Clustering methods are customary segmentation techniques, but they do not give meticulous results. Hence we have performed image filtering and traditional thresholding methods along with clustering technique for effective results. In our proposed method, we have used k-means clustering method and Otsu thresholding for effective segmentation. Finally, morphological closing is performed for accurate results. The experimental results unveil that the proposed method has less error percentage.