Hybrid transform energy based denoising approach for robust noise reduction and structural feature preservation

Synthetic Aperture Radar (SAR) images are affected by multiplicative speckle noise, which degrades image quality and limits subsequent processing. This study proposes Hybrid Transform Energy (HTE), a novel framework integrating Curvelet-based global directional thresholding with Wavelet-based adaptive texture enhancement for effective despeckling and feature preservation. An energy-based fusion strategy combines multiscale features to improve noise suppression while retaining important image details. Experiments on standard SAR image datasets achieve a Peak Signal-to-Noise Ratio (PSNR) of 45.522 dB, an Equivalent Number of Looks (ENL) of 43.313, and a Universal Image Quality Index (UIQI) of 0.99962. Compared with state-of-the-art methods, including SAR-DDPM and transformer-based approaches, HTE improves PSNR by approximately 2–4% and ENL by over 20%. Statistical significance testing further supports the consistency of the results. Its non-iterative design and low computational complexity make HTE a practical and scalable solution for high-quality SAR image denoising.