Glottal Area Segmentation without Initialization using Gabor Filters

Abstract

This paper describes a method to automatically obtain the glottal space segmentation without user initialization from healthy and pathological vocal folds video sequences captured by the laryngoscope. The segmentation is mainly based on a Gabor filter bank, studying the texture differences inside vocal folds images, and combining it with others advanced image processing techniques to achieve the expected results. The authors want to emphasize that the proposed algorithm is independent of images' resolution and zoom, but the quality of them depends on specialist experience with the instrumentation. Our proposal has worked correctly in all database test videos and it shows a great advance in design, and in the nearby future, a complete method to diagnose vocal folds pathologies.