An Unsupervised Image Segmentation Using B-Splines Functions

Abstract

We propose in this paper an unsupervised Bayesian image segmentation based on non-parametric expectation-maximization (EM) algorithm. The non-parametric aspect comes from the use of the B-spline probability density function (pdf) estimation, which is reduced to the estimation of the parameters of B-splines of the pdf.

Topics

    6 Figures and Tables

    Download Full PDF Version (Non-Commercial Use)