稀疏MIMO声纳沉底目标特征提取及识别研究

项目名称： 稀疏MIMO声纳沉底目标特征提取及识别研究

项目编号： No.11204051

项目类型： 青年科学基金项目

立项/批准年度： 2013

项目学科： 物理学I

项目作者： 朱广平

作者单位： 哈尔滨工程大学

项目金额： 27万元

中文摘要： 稀疏布放的MIMO声纳体制，采用空间分集的思想，通过接收不同方位目标散射波，可有效获得目标散射场信息，可大大提高对沉底目标的识别能力，对沉底目标声探测具有重要的实际意义。稀疏MIMO声纳沉底目标识别的关键问题是如何有效地将目标不同方向的散射波信号进行信息综合并提取特征。针对此问题，重点研究稀疏MIMO声纳沉底目标散射信息重建及特征提取方法。研究思路是，首先分析宽带信号激励下混响干扰与目标信号的时频特性，采用时频滤波方法抑制混响，然后研究基于压缩感知理论的MIMO声纳目标散射场的空间信息重建方法，将不同方位的沉底目标回波信息有效进行综合，并分析目标、混响及非目标的空间散射特性差异，提取二维散射特征，最后设计基于数据特征驱动的支持向量机作为分类器进行目标识别研究，并通过实验验证特征提取及识别效果。本项目研究可为稀疏MIMO声纳体制下沉底目标特征提取及识别问题提供有效的解决途径及技术基础。

中文关键词： 特征提取；支持向量机；沉底目标；混响；MIMO声纳

英文摘要： The widely separated MIMO sonar system, which is by means of idea of spatial diversity, via receiving scattering waves from different azimuth of object to achieve the information of scattering field, could improve ability of recognition of bottomed objectives and is practical significance for detecting bottomed objects. The key problem using widely separated MIMO sonar system to recognize bottomed object is how to effectively synthesize the information of scattering waves from different azimuth of object and extract the features. For this problem, the emphasis of study is the methods of reconstructing the scattering information and extracting feature under the widely separated MIMO sonar system. Firstly, analyze the time-frequency characters of reverberation and object signals under wide-band transmitting signals, and restrain the reverberation by means of time-frequency filter. And then study the methods of reconstructing the spatial information of scattering field of the objects under MIMO sonar system basing on theory of Compressed Sensing. Effectively synthesize the information of scattering waves from different azimuth of bottomed object and analyze the differences from objects, reverberation and non-objects. And extract their two-dimension scattering features. At last, design support vectors machine to rec

英文关键词： extracting feature；SVM；bottomed target；reverberation；MIMO sonar