Title :
Robust modelling of local image structures and its application to medical imagery
Author :
Wang, Li ; Bhalerao, Abhir ; Wilson, Roland
Author_Institution :
Dept. of Comput. Sci., Warwick Univ., Coventry, UK
Abstract :
A robust modelling method for detecting and measuring isotropic, linear features and bifurcations is described and applied to analysing 2D electrophoresis and retinal images. Features are modelled as a superposition of Gaussian functions with the Hermite expansion and estimated by a combination of a multiresolution, windowed Fourier approach followed by an EM type of spatial regression. A penalised likelihood test, the Akakie information criteria (AIC) is used to select the best model and scale for feature segments. Results are shown by using samples on both gel and retinal images.
Keywords :
Fourier transforms; Gaussian processes; approximation theory; electrophoresis; eye; feature extraction; image resolution; image sampling; image segmentation; maximum likelihood estimation; medical image processing; optimisation; regression analysis; 2D electrophoresis analysis; Akakie information criteria; EM algorithm; Gaussian function; Hermite approximation; Hermite expansion; bifurcation; expectation maximization algorithm; feature measurement; feature segmentation; image structure modeling method; isotropic linear feature detection; medical imagery; penalised likelihood test; retinal image analysis; robust modelling method; spatial regression analysis; windowed Fourier method; Application software; Bifurcation; Biomedical imaging; Computer vision; Image analysis; Image segmentation; Retina; Robustness; Signal resolution; Testing;
Conference_Titel :
Pattern Recognition, 2004. ICPR 2004. Proceedings of the 17th International Conference on
Print_ISBN :
0-7695-2128-2
DOI :
10.1109/ICPR.2004.1334584
