Ship detection based on coherence images derived from cross correlation of multilook SAR images

A new simple technique is presented to extract ships from synthetic aperture radar (SAR) images. The procedure is to compute the cross-correlation values between two images extracted by moving windows of a small size from the multilook SAR intensity (or amplitude) images. A coherence image, consisting of the cross-correlation values of the intensity images, is then produced. Ships are deterministic targets, so that their interlook subimages possess higher degree of coherence than the uncorrelated random images of the surrounding sea surface. The main advantage of this method over the conventional constant false-alarm rate is its ability to detect, under favorable conditions, "invisible" images of ships embedded in the speckled image of the sea surface. The technique is tested using a RADARSAT-1 image in which one known and several unknown ships are present. The use of complex images and the exploitation of short decorrelation times of small-scale ocean waves to obtain an extra look are also discussed.