Boundary estimation for star-shaped objects
Mats Rudemo, Henrik Stryhn · Lecture notes-monograph series · 1994
The present paper gives a motivation for the study of star-shaped contour models, surveys briefly some related models in statistical image analysis, and presents the asymptotic distribution of a histogram-like boundary estimator.We consider a two-region image with a star-shaped contour and continuous nonstationary observations.The image model contains two-dimensional Gaussian white noise and a representation of the contour by a regressogram in radii, i.e., a piecewise circular boundary.The maximum likelihood estimator of arc radii is studied when the noise variance and the regressogram sector width tend to zero, and the asymptotic distribution is specified in terms of the location of the maximum of a two-sided Brownian motion with negative drift.