检测海面小目标的RLS线性预测方法  

RLS Linear Prediction Approach for Detection of Small Target in Sea Clutter

作　　者：刘允峰[1,2] 索继东[2] 柳晓鸣[2] 苏晓宏[2] 

机构地区：[1]渤海大学信息科学与技术学院,辽宁锦州121000 [2]大连海事大学信息科学技术学院,辽宁大连116026

出　　处：《现代雷达》2017年第1期41-43,76,共4页Modern Radar

基　　金：国家高技术研究发展计划(863计划)基金资助项目(2012BAH36B02);国家自然科学基金资助项目(61403041)

摘　　要：为了提高海杂波中的小目标检测能力,提出了基于递归最小二乘线性预测的海面小目标检测方法。首先,建立线性预测模型;其次,利用递归最小二乘法动态调整模型的参数;最后,计算绝对预测误差的均值,通过阈值比较得到检测目标结果。采用加拿大IPIX雷达数据的实验结果表明,该方法的检测性能优于线性预测的检测目标方法和神经网络集成的检测目标方法的检测性能;同极化方式下,HH极性的检测效果优于VV极性的检测效果。该方法实时更新了预测模型参数,同步跟踪海杂波的变化,克服预测模型固定不变的局限,提高了目标检测的能力。In order to improve the detectability of weak targets floated in sea clutter, the method based on recursive least square （ RLS ） linear prediction is proposed. First, a linear prediction model is established. Secondly, the RLS method is used to dynamically adjust the parameters of the model. Finally, the average of absolute prediction errors is calculated and the target results are obtained by threshold comparison. Experimental results of live recorded sea returns collected by the Canada IPIX radar show that, detection performance of the proposed method is superior to that of linear prediction detection method and neural network ensembles detection method. The detection effect of Horizontal-Horizontal （HH） is better than that of Vertical-Vertical （VV） in the same polarization mode. Synchronously with the sea clutter, the parameters of the prediction model are updated real-time. The proposed method overcomes the limitation of the fixed model and improves detection capability.

关 键 词：海杂波 目标检测 递归最小二乘 线性预测模型 

分 类 号：TN957.51[电子电信—信号与信息处理]