Anisotropic-Scale-Space-Based Salient-Region Detection for SAR Images

作者：Zhang, Qiang^*; Wu, Yan; Wang, Fan; Fan, Jianwei; Zhang, Lei; Jiao, Licheng

来源：IEEE Geoscience and Remote Sensing Letters, 2016, 13(3): 457-461.

DOI：10.1109/LGRS.2016.2519241

• Anisotropic diffusion
• constant false alarm rate (CFAR)
• multiple scales
• salient region
• synthetic aperture radar (SAR)

Currently, salient-region detection, which can be used to provide important man-made target information for synthetic aperture radar (SAR) image analysis and interpretation, is a valuable tool in SAR image processing. However, because SAR images are blurred by a large amount of multiplicative speckle noise, it is difficult for existing methods to produce satisfying results with simple intensity variation. Based on the anisotropic scale space, which can describe edge variations, this letter presents a new salient-region detection method for intensity SAR images. In this method, by using the constant false alarm rate operator of the statistical distribution of the intensity, the edge intensity of each scale is measured with a corresponding-scale window to create the anisotropic scale space. Then, the dual Kullback-Leibler divergence between adjacent scales is computed to obtain the saliency metric of the pixel at a salient scale. To select salient regions, a simple iterative algorithm is presented. Experiments demonstrate the descriptive capacity of our anisotropic scale space and the performance of our method for intensity SAR images.

• 出版日期2016-3
• 单位西安电子科技大学