采用分布式并行子阵波束形成的水下三维成像  

Underwater 3D imaging by distributed and parallel subarray beamforming algorithm

作　　者：韩业强[1,2] 田翔[1,2] 陈耀武[1,2] 

机构地区：[1]浙江大学数字技术及仪器研究所,浙江杭州310027 [2]浙江省网络多媒体技术研究重点实验室,浙江杭州310027

出　　处：《浙江大学学报（工学版）》2014年第2期368-376,共9页Journal of Zhejiang University：Engineering Science

基　　金：国家自然科学基金资助项目(41276090);国家"863"高技术研究发展计划资助项目(2010AA09Z104);中央高校基本科研业务费专项资金资助

摘　　要：针对水下三维声纳成像技术因计算负载过大而无法满足实时性需求的问题,提出一种频域分布式并行子阵波束形成算法.基于大规模二维方形平面换能器阵列,将全面阵分解成两级分布式子阵.所有一级子阵采用并行计算架构,同时进行并行波束形成;一级子阵和二级子阵之间采用流水线分布式计算架构,在二级子阵中计算得出波束强度值.基于Matlab软件对该算法进行仿真测试,并与传统波束形成算法相对比.综合考虑主瓣宽度、旁瓣峰值、内存需求量和计算需求量4个参数,给出最合理的子阵分解方法.结果表明:该算法可以实现水下三维声纳成像,并且符合工程实践的实时性需求.A distributed and parallel subarray beamforming algorithm （DPS） was proposed in order to reduce the computational load associated with the signal processing in the underwater real-time 3D sonar imaging system. A full-populated transducer array was subdivided into two stage distributed subarrays. All of the first-stage subarrays used parallel computing framework and performed parallel beamforming simultaneously. The distributed computing framework was adopted between the first and second stage suharray. The second-stage subarray acquired the final beam pattern after beamforming. The Matlab software simulated the results of the algorithm in different conditions, and compared them with the conventional beamforming algorithm （CBF）. The most reasonable method of subarray decompositions was given based on four parameters： main lobe width, side lobe peak, memory requirement, and computational requirement. The results demonstrate that the DPS algorithm implements the underwater 3D sonar imaging and meets the demand of engineering utilizations.

关 键 词：水下实时三维声纳成像 分布式并行子阵 波束形成 

分 类 号：TB56[交通运输工程—水声工程]