Open AccessDe‐speckling method based on non‐local means and coefficient variation of SAR image
Author(s) -
Chen Shaobo,
Hou Jianhua,
Zhang Hua,
Da Bangyou
Publication year - 2014
Publication title -
electronics letters
Language(s) - English
Resource type - Journals
SCImago Journal Rank - 0.375
H-Index - 146
eISSN - 1350-911X
pISSN - 0013-5194
DOI - 10.1049/el.2014.0630
Subject(s) - synthetic aperture radar , artificial intelligence , filter (signal processing) , similarity (geometry) , speckle pattern , euclidean distance , computer vision , pixel , speckle noise , mathematics , euclidean geometry , enhanced data rates for gsm evolution , window (computing) , computer science , image (mathematics) , pattern recognition (psychology) , algorithm , geometry , operating system
Proposed is a speckle reduction method for synthetic aperture radar (SAR) images. This method can be considered as a new adaptive non‐local (NL) means filtering technique since different weights based on Euclidean distance and coefficient variation (CV) of SAR images are applied. The Euclidean distance of similarity windows and CV of the search window can be calculated first; next, the decay parameter will be adjusted by the CV; then, the speckled‐image pixel will be restored by weighted averaging based on the new weight. The visual and numerical experimental results show that the proposed filter can suppress speckle effectively and keep details (such as edge and texture) simultaneously. The proposed filter outperforms the existing state‐of‐the‐art de‐speckling filters based on NL means in terms of runtime.