A NEW METHOD FOR PROCESSING THE HIGH-SPEED GLOTTIS IMAGE SEQUENCE

摘要

High-speed digital imaging (HSDI) is an emergingmodality for endoscopic laryngeal imaging. HSDI has ahigh acquisition rate sufficient to resolve the fastphonatory vocal fold vibrations. The characteristics of thevocal fold vibration play an important role in the clinicalassessment of voice disorders as well as understanding themechanism of phonation. It is therefore of great interestfor voice clinicians to measure and quantify informationon the vocal fold motion during phonation. In this paper, asemi-automatic algorithm is introduced for efficient andaccurate glottis segmentation. This method consists oftwo major steps: first, an automatic tracking of the glottiscontour and second, a global threshold segmentation. Thereliability and accuracy of the algorithm are evaluatedusing representative clinical HSDI recordings.