Morphological signal adaptive median filter for noise removal

Abstract

A novel extension of the classical signal-adaptive median filter (SAM) is proposed in this paper. It is the so-called morphological signal-adaptive median filter (MSAM). Two modifications are introduced in the SAM filter aiming at: (1) enhancing SAM impulse detection mechanism so that it detects not only impulses of a constant amplitude but randomly-valued impulses as well, (2) employing an anisotropic window adaptation based on binary morphological erosions/dilations with predefined structuring sets. Performance results are reported by evaluating both objective criteria (e.g. SNR, MAE) and subjective criteria (e.g, the perceived quality of the filtered images). The proposed MSAM filter outperforms the classical SAM filter in all cases.