Texture segmentation with optimal linear prediction error filters

Trygve Randen and John Håkon Husøy
Rogaland University Center
Department of Electrical and Computer Engineering
P.O. Box 2557 Ullandhaug, N-4004 Stavanger, Norway

Piksel'n, 1994, vol. 11, number 3, pages 25-28

Full paper


An approach to feature extraction for texture images using optimal linear (autoregressive) predictors is presented. The features used for classification are calculated from the prediction error using a local energy function. Experimental results are given to show the applicability of the method.