针对传统机载合成孔径雷达(SAR)对海面舰船的检测算法难以适应复杂的滨海背景环境,提出一种基于SAR 图像背景与目标统计特性的粗检测和舰船运动特性精检测的两步分级处理算法。依据图形学处理方法实现滨海区机载SAR图像的海陆分离与陆地粗剔除,并采用自适应背景窗提取待检测目标;针对传统依据单一背景分布模型的恒虚警检测方法存在漏警和高虚警问题,提出基于K-lognormal 混合分布的恒虚警检测方法对待检测目标进行粗筛选。对筛选后的待检测舰船目标进一步根据舰船自身运动特性与聚焦效果剔除伪目标,降低虚警率,实现对滨海区舰船目标的精检测。实测数据结果验证了所提分级处理算法的有效性。

For the conventional ship detection algorithms of airborne synthetic aperture radar (SAR) aren′t suitable in the application of coastal region, a two-step hierarchical scheme including ship prescreening with the statistical model of SAR image background and targets, and ship discrimination by motion properties is proposed. Firstly, a preprocessing based on graphics processing methods is performed which is introduced to implement the separation of the land and sea for eliminating the effect of coastal regions, and adaptive windows are adopted to extract the ship candidates before prescreening. Then, for the problem of the conventional single distribution models combined with constant false alarm rate (CFAR) exist the high false dismissal and false alarm probability, the modified mixture distribution model K-lognormal for ship candidate detection is analyzed in the procedure of prescreening. After the prescreening, the discrimination via the difference between the true ship targets and the false ones in the aspects of motion properties is performed to reduce the false alarm rate. Finally, SAR real data processing is presented to validate the proposed scheme.

TN957.52