Remote sensing data change detection based on the CI test of Bayesian networks

Authors:
Dai Qin;Ma Jianwen;Ou Yang Yun
Affiliations:
The National Key Laboratory of Remote Sensing Information Science, Institute of Remote Sensing Applications, Box 9718, Datun Road 100101 CAS, Beijing, China;The National Key Laboratory of Remote Sensing Information Science, Institute of Remote Sensing Applications, Box 9718, Datun Road 100101 CAS, Beijing, China;The National Key Laboratory of Remote Sensing Information Science, Institute of Remote Sensing Applications, Box 9718, Datun Road 100101 CAS, Beijing, China
Venue:
Computers & Geosciences
Year:
2006

Citing 5
Cited 2

Probabilistic reasoning in intelligent systems: networks of plausible inference

Probabilistic reasoning in intelligent systems: networks of plausible inference
Bayesian networks without tears: making Bayesian networks more accessible to the probabilistically unsophisticated

AI Magazine
Real-world applications of Bayesian networks

Communications of the ACM
Learning Bayesian networks from data: an information-theory based approach

Artificial Intelligence
Learning the Dimensionality of Hidden Variables

UAI '01 Proceedings of the 17th Conference in Uncertainty in Artificial Intelligence

Review: Bayesian networks in environmental modelling

Environmental Modelling & Software
Integrating spatial relations into case-based reasoning to solve geographic problems

Knowledge-Based Systems

Quantified Score

Hi-index	0.00

Visualization

Abstract

In recent years, the Bayesian network approach has been used as a data-mining tool in many information fields, but it has rarely been used to process remote sensing data. In this paper, we introduce a Bayesian network classifier for remote sensing data change detection. Using the conditional independence (CI) test we can find out relationships among the attributes and construct a Bayesian network that incorporates these relationship constraints. After geometric correction and radiometric normalization, a Bayesian network change detection system based on CI test algorithm is developed and applied to two temporal Landsat TM data acquired in 1994 and 2003 of Beijing area, and the overall change detection classification accuracy can get 92%. The experimental results show that Bayesian network is a newly effective approach for remote sensing data change detection.