摘要
提出了一种基于知识的SAR图像目标检测算法。针对军用车辆,利用各种先验知识,以地形类型信息、距边界的距离信息、目标聚集程度为影响目标出现概率的因素,通过分类获得SAR图像的地形及边缘信息,得到影响因子,并综合地形信息使用MAP准则,从而获得目标检测的结果。使用真实SAR图像进行了测试,结果表明,与CFAR检测算法相比,该算法有效地提高了目标的检测率,虚警目标数目明显减少。
A target detection method in SAR images is proposed. The prior knowledge is used to detect the military vehicles. The target probability is influenced by the terrain type, the hedge proximity and the proximity of other targets. Classification technique is used to get the terrain types and hedge of the image, and the influ- ence factor is calculated. Then with the terrain types, the MAP technique is used to detect the targets. Experi- ments using SAR images indicate that compared with CFAR techniques, the new method proposed in this article increases the detection rate and decreases the number of false targets.
出处
《系统工程与电子技术》
EI
CSCD
北大核心
2009年第6期1314-1318,共5页
Systems Engineering and Electronics
基金
国家自然科学基金项目资助课题(60702011)
关键词
合成孔径雷达
目标检测
最大后验概率
基于知识
恒虚警检测
synthetic aperture radar
target detection
maximum a posterior
knowledge based
constant false alarm rate