Rima Glottidis Segmentation by Thresholding Using Normalized Cuts

Datum konání: 02.11.2012
Přednášející: Adam Novozámský
Odpovědná osoba: Sedlář

This presentation will introduce a new segmentation by thresholding algorithm that computes the threshold value automatically by a normalized graph cut measure. The weight matrices used for computation of graph cuts are based on the gray levels in the image, rather than on the commonly used image pixels. We used this algorithm for segmentation of rima glottidis in videokymographic images of vocal fold vibrations.