Automatic Segmentation of Glottal Space from Video Images Based on Mathematical Morphology and the hough Transform

Abstract

Vocal disorders directly arise from the physical shape of the vocal cords. Videostroboscopic imaging provides doctors with valuable information about the physical shape of the vocal cords and about the way these cords move. Segmentation of the glottal space is necessary in order to characterize morphological disorders of vocal folds. One of the main problems with the methods presented is their low level of accuracy. To solve this problem, an automatic method based on Mathematical Morphology edge detection and the Hough transformation is presented in this article to extract the glottal space from the videostroboscopic images presented. This method and two other popular algorithms, histogram and active contour, are performed on 10 sets of videostroboscopy data from excised larynx experiments to compare their performances in analyzing videostroboscopy images. The accuracy in computing glottal area of these methods are investigated. The results show that our proposed method provides the most accurate and efficient detection, and is applicable when processing low-resolution images. In this paper we used edge detection based on geometric morphology to detecting the edges of vocal cords. Then in the next step the edges that were extracted, using Hough transform change to some lines. After that through using proposed algorithm, we omit the extra lines and extract the glottis.

DOI:http://dx.doi.org/10.11591/ijece.v2i4.324