Non-coherent change detection in real SAR satellite imagery with usage of the CFAR algorithms

Abstract

Change detection in synthetic aperture radar (SAR) imagery is a demanding task due to the intrinsic speckle-noise accompanying the images. In general, it can be divided onto coherent (CCD) and non-coherent (NCCD) change detection methods. In the case of coherent methods, we have an increased sensitivity to changes (due to taking into account phase information) at the expense of incorporating phase noise, which in turn increases the level of false detections. In the case of incoherent methods, we do not take into account signal phase information but only its amplitude (or power), which results in a lower sensitivity to changes and also much lower level of false alarms. Therefore, incoherent detectors often demonstrates better operating performance (ROC <sup xmlns:mml="http://www.w3.org/1998/Math/MathML" xmlns:xlink="http://www.w3.org/1999/xlink">1</sup> curves) than coherent ones - while being simpler, more reliable and easier to handle. In this article, author will focus on SAR imagery non-coherent change detection based on constant false alarm rate (CFAR) algorithms.