一种有效的SAR图像分割与目标识别方法

在基于模板匹配的合成孔径雷达(syntheticapertureradar,SAR)目标识别中,一个关键问题就是如何从带有杂波的SAR图像中将目标正确分割出来,以便形成高质量的模板。针对这一问题提出了一种基于对数变换的自适应SAR图像分割方法并将其用于由美国国防高级研究计划署(DefenseAdvancedResearchProjectAgency,DARPA)和空军研究室(AirForceResearchLaboratory,AFRL)提供的实测SAR目标图像识别中。实验结果证明,经有效的目标分割后,不但提高了目标的正确识别率,还有效地提高了对假目标的拒识率,具有良好的鲁棒特性。

Efficient SAR image segmentation and automatic target recognition approach

In the template-based SAR(synthetic aperture radar) target recognition, a key problem is how to (segment) a target image from a noisy SAR image to form a high quality target template. A simple and efficient target (segmentation) method is proposed and applied to the SAR target recognition. Experimental results with MSTAR ((moving and stationary) target acquisition and recognition) SAR data sets provided by the US DARPA/AFRL ((Defense Advanced) (Research) Projects Agency/Air Force Research Laboratory) are presented to illustrate the performance of the proposed approach.