Signal Decomposition

Abstract

The goal of signal decomposition is extraction and separation of signal components from composite signals, which should preferably be related to semantic units. Examples for this are distinct objects in images or video, video shots, melody sequences in music, spoken words or sentences in speech signals. Signal decomposition methods are closely related to classification of underlying features, which characterize the component to be separated. An important category of signal decomposition methods is segmentation of image, video, audio and speech signals, which is often a prerequisite for useful feature extraction and classification, but can also be used to improve the performance of compression algorithms. Segments are localized either over a time interval or over a spatial region; they will typically be defined by criteria that indicate homogeneity of certain features or semantic coherence. In segmentation, usually no overlap of the signal components to be isolated occurs. A more challenging approach of decomposition is separation of single components from mixed signals, where the composite signal consists of a sample-wise superposition from multiple components.