A two-stage shearlet-based approach for the removal of random-valued impulse noise in images.

•Two-stage denoising algorithm to (i) detect and (ii) restore corrupted pixels.•Improved HEIND routine is highly accurate to detect corrupted pixels.•Innovative combination of noisy pixel detection and inpainting routines.•Shearlet inpainting is especially efficient at restoring edges and curvilinear features.•Outstanding performance for high-level noise with respect to competing methods.