7th International Workshop on Models and Analysis of Vocal Emissions for Biomedical Applications (MAVEBA 2011)

Florence, Italy
August 25-27, 2011

Fully-Automatic Glottis Segmentation with Active Shape Models

J. J. Cerrolaza (1), Víctor Osma-Ruiz (2), Nicolás Sáenz-Lechón (2), A. Villanueva (1), Juana M. Gutiérrez-Arriola (2), Juan Ignacio Godino-Llorente (2), R. Cabeza (1)

(1) Department of Electrical and Electronics Engineering, Public University of Navarra, Pamplona, Spain
(2) Department of Circuits and Systems, Polytechnic University of Madrid, Madrid, Spain

in this paper we present a new fully automatic glottis segmentation scheme that combines traditional bottom-up image processing techniques with high-level shape constraints provided by the active shape Models. unlike previous statistical segmentation approaches, which try to accurately detect the location of the glottis as initialization for the algorithm, we incorporate a new reliability score selector at the final stage of the scheme. the result is a robust and flexible algorithm able to deal with most acquisition techniques, even with stroboscopic videos. the good behavior of the algorithm has been successfully tested in a set of 170 frames extracted from 30 stroboscopic recordings.

Index Terms. Glottis segmentation; active shape Models; region Growing; stroboscopic videos.

Full Paper (reprinted with permission from Firenze University Press)

Bibliographic reference.  Cerrolaza, J. J. / Osma-Ruiz, Víctor / Sáenz-Lechón, Nicolás / Villanueva, A. / Gutiérrez-Arriola, Juana M. / Godino-Llorente, Juan Ignacio / Cabeza, R. (2011): "Fully-automatic glottis segmentation with active shape models", In MAVEBA-2011, 35-38.