Title :
Modeling a Random Field by Two-Dimensional AR Process and its Application to Texture Analysis
Author :
Oe, Syunichiro ; Tomita, Yutaka
Author_Institution :
Department of Information Science and Systems Engineering, Faculty of Engineering, University of Tokushima, Tokushima 770, Japan
Abstract :
This paper deals with a method for the texture segmentation or classification using two-dimensional AR model. We propose a new distance measure which shows the statistical difference between two textures based on Kullback information. Finally, the effectiveness of the present algorithm is shown by numerical examples.
Keywords :
Covariance matrix; Data mining; Equations; Gaussian processes; Information analysis; Information science; Parameter estimation; Predictive models; Systems engineering and theory;
Conference_Titel :
American Control Conference, 1986
Conference_Location :
Seattle, WA, USA
