合成孔径雷达(SAR)图像分割是SAR 图像处理的基础,国内外研究者提出了很多行之有效的分割方法。典型的算法如基于单阈值形态学分割算法、基于马尔科夫随机场的分割算法等。然而,考虑实际需求,图像分割需要同时兼顾快速性和准确性,这是当前手段相对缺乏的。文中提出了一种柔性自适应SAR 图像目标分割算法,将峰值点的提取过程与恒虚警率检测算法相结合分割SAR 图像中的目标。该算法可以将散射中心信息融入到目标分割中,同时完成目标分割和峰值点提取,是一种快速而又精确的图像分割算法。最后,该文基于数据集对算法进行了验证,证实了该算法的合理性与可行性。

Target segmentation is the base of synthetic aperture radar (SAR) image interpretation. Many target segmentation algorithm has been proposed for SAR image processing, including single-threshold morphology (STM), markov tandom gield (MRF),etc. This paper introduces a flexible adaptive SAR image target segmentation algorithm, in which the extraction of peak point and the constant false alarm rate algorithm (CFAR) are combined to segment targets in SAR image. MSTAR data set is used to validate the algorithm. Compared with STM and MRF, the algorithm introduced in this paper is both fast and accurate.

TN959