A fast environmental change detection approach based on unsupervised multiscale texture clustering

In this paper, an analytic model for mapping land cover change that uses a novel multiresolution analysis in concert with image differencing to identify environmental change/no-change areas is presented. The method adaptively chooses thresholds to segment targets from background, by using multiscale...

Full description

Saved in:
Bibliographic Details
Published in:International journal of environmental studies 2005-02, Vol.62 (1), p.79-93
Main Authors: OUMA, Y. O., TATEISHI, R.
Format: Article
Language:English
Subjects:
Animal, plant and microbial ecology
Applied ecology
Automated environmental change detection
Biological and medical sciences
Conservation, protection and management of environment and wildlife
Decomposition
environmental changes
Fundamental and applied biological sciences. Psychology
General aspects
General aspects. Techniques
Geographic information systems
Landsat
Mapping
Multiresolution analysis
Q1
Remote sensing
Risk assessment
Teledetection and vegetation maps
vulnerability
Wavelet transforms
Citations: Items that this one cites
Items that cite this one
Online Access:Get full text
Tags: Add Tag
No Tags, Be the first to tag this record!
Description
Summary:In this paper, an analytic model for mapping land cover change that uses a novel multiresolution analysis in concert with image differencing to identify environmental change/no-change areas is presented. The method adaptively chooses thresholds to segment targets from background, by using multiscale decomposition of the image difference. Changes due to atmospheric influences are automatically suppressed and resolution heterogeneity problems minimized. Examples that demonstrate the efficiency of the technique on medium spatial resolution Landsat multitemporal imagery are presented. Empirical evaluation supports the suitability of this technique for fast identification of changed scenes. As opposed to the traditional change detection methods, we directly detect land cover changes rather than simply pixels. This is evident from the less-pixelated appearance of the change maps. The technique is recommended for regional/local environmental vulnerability and risk assessment as well for automated updating of GIS databases.
ISSN:0020-7233
1029-0400
DOI:10.1080/0020723042000286374