Hybrid Image Denoising Framework Based on Transform-Domain Signal Enhancement Using Sliced Ridgelet Transform for Medical and Industrial Applications

Effective image denoising is crucial for ensuring high-fidelity results in medical diagnostics and industrial defect analysis. This paper presents the first phase of a two-phase AI-augmented denoising framework using a Sliced Ridgelet Transform (SRT) - based preprocessing module. Phase I aims to convert noisy images into sparse, edge-preserving representations through advanced transform techniques including Radon projections, ridgelet decomposition, and adaptive thresholding. Results on benchmark datasets demonstrate significant noise suppression, enhanced directional sensitivity, and improved coefficient sparsity, forming a robust foundation for AI-based refinement in Phase II.