An improved unsupervised threshold determination method for SAR image change detection

Abstract

One of the key techniques for multi-temporal synthetic aperture radar (SAR) image change detection is how to select and determine the threshold. The Kittler and Illingworth (KI) algorithm is an unsupervised threshold determined method based on the least error ratio of the Bayesian theory. Although it is a simple and feasible parameter estimation approach, there are some deficiencies. In order to improve these deficiencies, this paper proposes an improved unsupervised threshold determination method on the basis of the KI algorithm, and it is called mean circulation iteration KI (MCI-KI) algorithm. The main idea of the MCI-KI algorithm is to shorten the scope of selecting the threshold via controlling errors. So it reduces the burden of computation and saves the running time. Finally, the airborne and spaceborne SAR images are used to test the proposed method, and experimental results verify that the method is effective and feasible.