Home // SPACOMM 2020, The Twelfth International Conference on Advances in Satellite and Space Communications // View article

Region-based N-cuts Polarimetric SAR Image Segmentation Algorithm

Authors:
Lingkai Zhao
Mingchuan Yang

Keywords: Polarimetric SAR; Polarization Characteristic; K-means; region-based N-cuts; Image Segmentation

Abstract:
Due to the imaging principle, there is a large number of noise points in polarimetric Synthetic Aperture Radar(SAR) images. This coherent noise leads to inaccurate segmentation results. To solve this problem, this paper proposes a region-based N-cuts polarimetric SAR image segmentation algorithm by combining the K-means clustering algorithm and N-cuts algorithm. Firstly, K-means clustering algorithm is used to pre-segment polarimetric SAR images to form segmented regions. Secondly, the similarity measurement matrix is constructed on the basis of pre-segmentation. Finally, the N-cuts algorithm is introduced to cluster regional nodes and to realize image segmentation. This method makes full use of the over-segmentation characteristic of K-means clustering algorithm, and it significantly reduces the computation. Combined with the global optimization of atlas segmentation algorithm, the performance of segmentation results is improved. In this paper, full polarization E-SAR(the Experimental airborne SAR System of DLR) images are used for experiments, and processing results of various polarization characteristics are compared. Experimental results show that region-based N-cuts algorithm is efficient and practical for image segmentation. The results also show that this method has higher level of precision and shorter computation time than both K-means clustering algorithm and N-cuts algorithm.

Pages: 1 to 6

Copyright: Copyright (c) IARIA, 2020

Publication date: February 23, 2020

Published in: conference

ISSN: 2308-4480

ISBN: 978-1-61208-769-6

Location: Lisbon, Portugal

Dates: from February 23, 2020 to February 27, 2020