IAM   02674
INSTITUTO ARGENTINO DE MATEMATICA ALBERTO CALDERON
Unidad Ejecutora - UE
artículos
Título:
A method for mixed states texture segmentation with simultaneous parameter estimation
Autor/es:
AGUSTÍN MAILING; BRUNO CERNUSCHI FRÍAS
Revista:
PATTERN RECOGNITION LETTERS
Editorial:
ELSEVIER SCIENCE BV
Referencias:
Lugar: Amsterdam; Año: 2011 vol. 32 p. 1982 - 1982
ISSN:
0167-8655
Resumen:
In this work a method for mixed-state model motion texture segmentation and parameter estimation ispresented. We use the Expectation Maximization algorithm for mixture parameter estimation, introducingthe Gibbs distribution for moving points, excluding zero discrete component associated with nomotion regions. We use then the a posteriori probabilities to generate an alternative field to segmentthe textures according to its statistical parameters