B Panchami, SP Kumar, AJ Rajkumar, V Rishika, Shri Harini Ramesh
Proceedings of the 2023 International Conference on Intelligent Systems for Communication, IoT and Security (ICISCoIS)
Abstract: Laryngeal high-speed video endoscopy is used to visualize the vocal folds pattern of vibration for appropriate diagnosis and evaluation of voice disorders. However, it is tedious to examine LHSV frames merely through visual perception by explicating underlying vibration patterns, time dynamics and spatial dynamics. As a result, its crucial to employ a visualization approach that examines the vocal folds vibratory characteristics. To investigate and assess the vibratory patterns of vocal folds, image processing techniques can be employed to interpret motion, and to provide facilitative playbacks, and quantitative characteristics. Additionally, optical flow approaches can be utilized to investigate the local and global dynamics of the vocal fold vibrations. In this paper, optical flow kymograms and glottal optical flow waveforms are developed applying two distinct motion estimation method, namely Lucas-Kanade and Horn-Schunck. The vocal fold characteristics are analyzed both visually and quantitatively from OFKG images and GOFW waveforms. OFKG enables visualization of light artifacts. GOFW peak value was estimated using facilitative playback and the mathematical model was used to compute physical parameters like opening speed and closing speed. The correlation of these parameters has been derived from GOFW peak values of varying upper and lower margin amplitudes. The linear proportionality between the GOFW peak value and opening speed is evident in these quantifiable characteristics.