Noisy heart sequences filtering by optimal median based operators

Abstract

Bit errors and impulses introduced into cardiographic image sequences prohibit to process and evaluate the heart dynamics correctly. Thus, biomedical imaging such as vascular imaging and quantification of heart dynamics is closely related to digital filtering. We provide three-dimensional impulse detection based adaptive median filters designed to perform the optimal filtering situation, where the robust median filter affects only noisy samples, whereas the desired image features remains unchanged. The proposed method holds excellent preserving characteristics and takes the advantage of the superposition property despite the nonlinearity of the median filtering.