Abstract
ABSTRACT To detect linear structure, model-based approaches using Hough and Radon transforms are often used but, are not recommended for thick line detection, whereas methods based on image derivatives need further tedious step-by-step processing. In this paper, a novel detection paradigm is presented, where the 3D image gray level representation is considered as finite mixture model of statistical distributions, called linear anchored Gaussian and parametrized by radius, angle and scale parameters dealing with structure location and thickness. These parameters could estimated by Expectation-Maximization algorithm. To rid the data of irrelevant information brought by nonuniform and noisy background, a modified EM algorithm is detailed. The proposed method gives very accurate results on real-world and synthetic images, where, for the latter with strong Gaussian blur and Additive White Gaussian Noise ( σ n = 150 ), the mean estimation errors on the orientation, the distance from the origin and the thickness reach 0.35 ∘ , 0.4 and 0.48 pixel, respectively.
Published Version
Talk to us
Join us for a 30 min session where you can share your feedback and ask us any queries you have
Schedule a call