NSCT-Based Adaptive Denoising for SAR Images

This paper presents an NSCT-based adaptive denoising program for SAR images, which operates within the NSCT toolkit framework. The Non-Subsampled Contourlet Transform (NSCT) represents an advanced wavelet transformation method that outperforms traditional wavelet approaches by better handling signal multiresolution and local characteristics while extracting signal features more precisely. The implementation involves NSCT decomposition through multiscale directional filter banks, followed by adaptive thresholding applied to transform coefficients based on noise statistics estimation. The reconstruction phase then synthesizes the denoised image using inverse NSCT transformation. This method effectively suppresses speckle noise in SAR imagery, significantly improving image quality and clarity through directional sensitivity and shift-invariance properties. The program features configurable decomposition levels and adaptive threshold selection algorithms, making it suitable for diverse SAR image processing applications across aviation, hydrology, seismology, and related fields.