USING THE MODEL OF A MIXTURE OF A UNIFORM DISTRIBUTION AND A VON MISES DISTRIBUTION FOR SEGMENTATION OF ANISOTROPIC IMAGES
I. S. Gruzman
Novosibirsk State Technical University, pr. Karla Marksa 20, Novosibirsk, 630073 Russia
Keywords: segmentation, local orientation estimate, estimation of the parameters of a mixture of distributions, angular random variables
Abstract
An algorithm is proposed for segmenting anisotropic image based on estimates of the local orientation of textures obtained using a gradient structural tensor. A mixture of a uniform distribution and a von Mises distribution is used as the estimate model. An estimation algorithm for the unknown parameters of the mixture based on trigonometric moments is obtained. Examples are given demonstrating the effectiveness of the segmentation algorithm in identifying textures with one dominant direction.
|
