The Acoustic Emission (AE) technique plays a progressively significant role in the field of non-destructive testing (NDT) especially in structural health monitoring (SHM). Acoustic emissions are commonly defined as transient elastic waves in a material caused by the of localized stress release. In using AE for structural diagnostics, noise has always been a potential barrier. AE can be produced from sources not related to material damage including traffic or friction. The major challenge is the differentiation of signals relevant to the purpose of the monitoring - such as crack growth in a member - from noise of various origins. This paper deals with noise discrimination and introduces a novel approach for noise interpretation in AE data. AE activities recorded in field and lab environments for concrete and steel specimens are investigated in this study. Approaches for clustering and separation of AE signals based on multiple features extracted from experimental data are presented.