Doubly Adaptive Nonlocal Means Image Denoising Algorithm Based on Mathematical Morphology

An doubly adaptive nonlocal-means (DANLM) algorithm for image denoising is proposed in this paper. The wavelet-Based denoising method is first used to get the pilot image, and then the pilot image is divided into texture and smooth regions base on mathematical morphology. Finally, the noisy image is denoised by nonlocal means method with patch window and search window adaptively adjust to the local property for different regions. Experiment results demonstrate the superior denoising performance of the proposed denoising technique.