Change detection in SAR images based on the logarithmic transformation and total variation denoising method

ABSTRACT

Noise interference and the need to process massive image data present challenges to change detection in synthetic aperture radar (SAR) images. In order to improve the change detection accuracy and decrease the processing time, this paper proposes a novel unsupervised change detection algorithm for SAR images. The logarithmic transformation is applied to transform images into the logarithmic domain, while the multiplicative noise in images is transformed into additive noise. The total variation (TV) denoising algorithm is then used to reduce image noise, and the difference operator in the logarithmic domain is employed to provide the difference image. The k-means clustering algorithm, which does not require consideration of the statistical properties of an image, is employed to cluster the difference image into two disjointed classes: changed and unchanged. The experimental results demonstrate that change detection results achieved by the proposed algorithm offer great improvement over existing algorithms in terms of objective quantitative indices and the visual effect.

Disclosure statement

No potential conflict of interest was reported by the authors.

Additional information

Funding

This work was supported in part by the International Cooperative Research and Personnel Training Projects of the Ministry of Education of the People’s Republic of China [Grant number DICE 2014-2029].