Change detection in signals using linear regression models

The problem of change detection in signals using linear regression models is addressed. Most of the algorithms presented make use of two AR models: a reference model and a current model updated via a sliding block. Changes are detected when a suitable distance between these two models is high. Three distance measures are considered in the paper: cepstral distance, log-likelihood ratio (justified by GLR) and a distance involving the cross-entropy of the two conditional probabilities laws (divergence test). Finally, a change detection algorithm using three models and the evolution of Akaike´s information criterion is presented. Some results on the application of the discussed algorithms in seismic signal segmentation are included