【摘要】 引入分布式Agent系统进行SAR图像海洋分布目标检测,针对分布目标的特性提出了知识继承行为和2类繁殖方法。并利用邻域直方图的χ2距离作为差异度,提出利用差异度分布分析方法得到Agent判断的先验知识。实验表明,对于SAR海洋图像,算法不拘泥于目标的种类和形状,能快速排除无目标区域,在目标区迅速扩大检测结果,同时算法大大减少了直方图分析次数,提高了运算速度。更多还原

【Abstract】 To detect distributed targets in the SAR sea images,an algorithm based on distributed Agents system is proposed.According to the features of the distributed targets,the knowledge inheriting behavior and two kinds of breeding behaviors are proposed.Then through analyzing the χ2 distance of the local histograms,the threshold in the judgment behavior can be obtained.Experimental results show that the algorithm can quickly get rid of the region without targets,but find the target region.The time of histograms computation are reduced.And it is effective for different distributed targets.更多还原