Investigation of multiscale product for change detection in difference images

Author

Beauchemin, M. ; Fung, Ko B.

Author_Institution

Canada Centre for Remote Sensing, Natural Resources Canada, Ottawa, Ont., Canada

Volume

6

fYear

2004

fDate

20-24 Sept. 2004

Firstpage

3853

Abstract

We investigate the potential of multiscale product as a measure for change detection in difference images. The rationale relies on exploiting the cross-scale correlation characteristics of signal and noise. We use Monte Carlo simulations to assess the detection and localization performances of the multiscale product for the detection of linear features of different widths. Comparisons are made with change detection based on single scale as well as other multiscale combination rules. Preliminary simulation results show that, overall, a good performance balance is obtained for change detection over the range of SNRs and line widths investigated.

Keywords

Monte Carlo methods; correlation theory; feature extraction; geophysical signal processing; multidimensional signal processing; remote sensing; Monte Carlo simulations; change detection; cross-scale correlation characteristics; detection performance; difference images; linear feature detection; localization performance; multiscale analysis; multiscale combination rules; multiscale product; noise correlation; signal correlation; Acoustic noise; Algorithm design and analysis; Change detection algorithms; Computer vision; Image analysis; Image edge detection; Noise reduction; Pixel; Remote sensing; Signal to noise ratio;

fLanguage

English

Publisher

ieee

Conference_Titel

Geoscience and Remote Sensing Symposium, 2004. IGARSS '04. Proceedings. 2004 IEEE International

Print_ISBN

0-7803-8742-2

Type

conf

DOI

10.1109/IGARSS.2004.1369964

Filename

1369964