Fast FFT-based motion compensation for laryngeal high-speed videoendoscopy

(HSV) are compared. Two of them are based on tracking the maximum of the cross-correlation function of two images; two are based on minimization of the L 2 –norm and L 2-like distance between two images ; and the other two make use of the peak present in the cross-power FFT-based spectrum of two images. All six methods are applied to compensate the motion, at the sub-pixel level, of the endoscopic lens relative to the vocal folds in HSV recordings. The new motion compensation methods based on FFT cross-power spectrum demonstrated remarkable computational speed and acceptable accuracy. While accuracy was best for the L 2-minimization techniques , they were slower and had a limited motion-tracking range.