Open AccessSub‐region non‐local mean denoising algorithm of synthetic aperture radar images based on statistical characteristics
Author(s) -
Ma Wei,
Xin Zhihui,
Liao Guisheng,
Sun Yu,
Wang Zhixu,
Xuan Jiayu
Publication year - 2022
Publication title -
iet image processing
Language(s) - English
Resource type - Journals
SCImago Journal Rank - 0.401
H-Index - 45
eISSN - 1751-9667
pISSN - 1751-9659
DOI - 10.1049/ipr2.12516
Subject(s) - synthetic aperture radar , non local means , artificial intelligence , noise reduction , noise (video) , algorithm , computer science , image (mathematics) , transformation (genetics) , pattern recognition (psychology) , probability distribution , pixel , mathematics , computer vision , image denoising , statistics , biochemistry , chemistry , gene
When synthetic aperture radar (SAR) images are denoised by non‐local mean (NLM) algorithm, logarithmic transformation will lead to the loss of some image information. To keep the details and smooth the noise of the SAR images better, a new sub‐region NLM denoising algorithm with the statistical characteristics of SAR image is proposed in this paper. Firstly, the probability distribution image is generated by calculating the probability value of every pixel. Then the images can be divided into the heterogeneous region and the homogeneous region by the threshold obtained with the variation coefficient of the probability image. A new filtering weight using both the original and probability images is generated based on NLM in the heterogeneous region. The filtering weight is obtained using the probability image in the homogeneous region. This method fully considers the characteristics of noise in different regions. Multi‐SAR image experiments demonstrate the advantages of noise smooth and detail protection.