FLDA and PCA Classification Supported by an Adapted Block Matching Algorithm to Diagnose Vocal Folds Paralysis

Amaia Méndez Zorrilla, Eneko Lopetegui Alba, and Begoña García Zapirain


Block Matching, Glottal Space Segmentation, FLDA, PCA


This paper presents the study of vocal videostroboscopic recordings to detect vocal folds paralysis using a combination of segmentation, block matching and classification techniques. This approach involves four process steps: 1) a pre-processing stage 2) the segmentation of the glottal area is made analyzing the image textures applying Gabor filtering, 3) an adapted version of exhausted block matching algorithm is used, and 4) two step classification stage using FLDA and PCA techniques. The block matching adaptation consists on for each frame a personalized block size and search window is applied due to segmented region features. Finally, the results show that our proposal works correctly to detect automatically vocal folds with paralysis and to distinguish them from healthy or pathological vocal folds.

Important Links:

Go Back