Adaptive method of speckle reduction based on curvelet transform and thresholding neural network in synthetic aperture radar images

作者：Zakeri Fatemeh^*; Zoej Mohammad Javad Valadan

来源：Journal of Applied Remote Sensing, 2015, 9(1): 095043.

DOI：10.1117/1.JRS.9.095043

• curvelet transform
• synthetic aperture radar images
• speckle noise
• thresholding neural network

Because of the effect of speckles in synthetic aperture radar (SAR) images, its reduction has been considered by many researchers to obtain reliable information. This paper proposes a method based on the curvelet transform to reduce speckles in SAR images. This study is based on the thresholding neural network (TNN) technique, which has been previously used in wavelet transformation. In addition, an adaptive learning TNN with remarkable time saving was introduced. Comparing the obtained results from the method with conventional speckle filters such as Lee, Kuan, Frost, and Gamma filters, curvelet-based, nonadaptive despeckling, wavelet-based TNN despeckling, and curvelet-based particle swarm optimization show better achievement of the proposed algorithm. For instance, noise mean value, noise standard deviation, mean square difference, equivalent number of looks, and beta (an edge-preserving criterion) improved 2%, 9%, 21%, 35%, and 9%, respectively.

• 出版日期2015-11-20