|
Activity Number:
|
288
|
|
Type:
|
Contributed
|
|
Date/Time:
|
Tuesday, August 8, 2006 : 10:30 AM to 12:20 PM
|
|
Sponsor:
|
Biometrics Section
|
| Abstract - #307597 |
|
Title:
|
Image Analysis by Spatial Point Process Modeling in Irregular Area
|
|
Author(s):
|
Weimin Zhang*+ and Suojin Wang
|
|
Companies:
|
Texas A&M University and Texas A&M University
|
|
Address:
|
6602 Harbor Town Drive, Apt. 201, Houston, TX, 77036,
|
|
Keywords:
|
area-interaction process ; maximum pseudolikelihood estimate ; Markov chain Monte Carlo ; Monte Carlo maximum likelihood estimation ; simulation study
|
|
Abstract:
|
Spatial point pattern analysis in an irregular area with heavy edge effects is generally computationally expensive due to heavy dependence on Monte Carlo simulations. In this paper, an aggregated point pattern model, the area-interaction process model, was applied to a lattice image data analysis application in an irregular area. The spatial distributions are examined by fitting area-interaction processes to the data. The Monte Carlo maximum likelihood estimation is considered. The statistical properties of the estimation procedure are investigated by a group of simulation studies. Despite the complexity of the formulation, the simulation research shows that the estimator is consistent. We compared this with the maximum pseudo-likelihood estimation.
|
