Automatic local thresholding algorithm for SAR image edge detection

Abstract

Using the combined method of thresholding algorithm and nonmaximum suppression, edge-pixels can be extracted rapidly from the edge strength map gotten by a synthetic aperture radar (SAR) image edge detector. Most thresholding algorithms require one or more thresholds which are hardly determined empirically. An automatic local thresholding algorithm (ALTA) is proposed in this article. ALTA can quickly and conveniently predefine the required thresholds through a simple computation process. ALTA, firstly divides the detected SAR image into a series of sub-images, secondly computes the mean of a sub-image, and thirdly determines a threshold for each sub-image through the computed mean. The receiver-operating-characteristic curve is supposed to evaluate . Experiment results indicate that by using ALTA, an edge detector can attain a greater edge-pixels detection effect.