Despeckling of Ultrasound Images Using Non-subsampled Shearlet Transform and Enhanced Gradient Domain Guided Filter

Ultrasound imaging is widely used in medical diagnostics. However, the presence of speckle noise degrades
image quality by reducing contrast and obscuring structural details. To overcome this problem, this study proposes a
despeckling algorithm that integrates the non-subsampled shearlet transform, an enhanced gradient-domain guided
filter (eGDGF) augmented with the maximum local variation (MLV) operator, and an improved thresholding shrinkage
rule derived from the residual image weighted by a bilateral filter. The integration of the edge-aware eGDGF with the
MLV operator enhances local contrast and structural sharpness in low-frequency subbands, while the improved
thresholding shrinkage rule in high-frequency subbands enables adaptive speckle suppression and perceptual
enhancement. Experimental results demonstrate that the proposed algorithm achieves effective noise reduction while
preserving structural details and improving the perceptual quality of ultrasound images with varying noise levels.