探雷声纳目标的概率分类技术

摘要

探雷声纳搜索海底时发现目标数量多,重合点少,现有方法难以快速有效识别水雷.对此提出一种基于概率特征的目标分类技术,通过多次搜索获得小目标发现率,根据错判概率最小准则确定分类判据,对概率差异明显的目标实现有效分类.对以往的实际搜索数据分析结果证明,该方法能有效剔除大部分低概率小目标,减小待处理目标数量,显著提高探雷声纳的综合搜索效能.%The targets which found by mine detection sonar while searching sea bed are large quantity and can`t be found again easily.The current measures are difficult to classify the mines from the sonar targets.Now a new method has been developed to do this efficiently, which based on the difference between the targets` detecting probabilities acquired from multiple searches.The method has been approved effective when it applies on some practical examples.The results express that a large percentage of lower probability targets are rejected, the numbers of reserved targets are reduced obviously and the comprehensive effectiveness of the mine-detection sonar is raised remarkably.