Change Detection based on Bayes' Theorem for Intensity Wavelength-Resolution SAR Difference Images

Gabriel L Espindola Pedro, Dimas Irion Alves, Diego da Silva de Medeiros, Paulo Ricardo Branco da Silva, João V R Negri, Arthur CM Barcella

DOI: 10.14209/sbrt.2025.1571156659

Evento: XLIII Simpósio Brasileiro de Telecomunicações e Processamento de Sinais (SBrT2025)

Keywords: CARABAS II change detection bivariate Gamma distribution background statistics

Abstract

Detecting concealed targets under foliage remains a significant challenge for wavelength-resolution synthetic aperture radar (SAR) systems. This paper proposes the usage of the bivariate Gamma distribution as a clutter model, shown to be a well fitted distribution for intensity SAR difference images, on an noniterative Change Detection (CD) algorithm based on Bayes' Theorem. The results were compared by using ROC curves and a probability of detection of 98.94\% at false alarm ratio of 1 per kilometer squared was achieved, outperforming previously used distributions in the literature.

Download