混响环境下基于频率-波数谱分析的水下慢速目标回波检测方法

Detection Method for Underwater Slow Moving Targets Based on Frequency-wavenumber Spectrum Analysis in Reverberation Environment

针对水下慢速目标回波信号往往被随机噪声、混响信号等湮没的问题,提出一种基于频率-波数(FK)谱分析的回波检测方法。FK谱分析相当于速度滤波器,回波中的混响信号与运动目标信号在FK域呈现不同形态。利用此特性,通过Radon变换提取接收信号经过二维傅里叶变换后的FK谱线信息。构建低维数特征空间,采用支持向量机识别运动目标和混响信号。进一步根据运动目标信号的FK谱线斜率构建滤波器,进行信号滤波。该方法将目标分类前置,不需要预先对信号做波束形成,较传统方式更高效。采用海试数据验证算法,成功从强混响信号提取目标信号,并获得大约3 dB增益。结果表明,回波检测方法能够提取受混响严重干扰的慢速运动目标信号。

For the problem that detected signals of underwater slow moving targets are usually immersed by random noise and reverberation,this paper presents a detection method based on frequency-wavenumber(FK)spectrum analysis.FK analysis is like a velocity filter,in which the signals of targets and reverberation have different distributions.According to this feature,the information in FK spectrum obtained through two-dimensional Fourier transform is extracted by Radon transform.A low dimensional feature space is constructed based on the FK spectrum,and the supported vector machine is used to recognize the signals of moving targets and reverberation.Additionally,the signals of moving targets are filtered according to the slope of FK spectrum.The proposed method is more effective than traditional method because it is“classify before detect”without beamforming.The feasibility of the method is verified using sea trial data.Target signal is successfully extracted from the strong reverberation signal,and the obtained gain is 3 dB.The result shows that the proposed method has ability to extract slow moving target signal critically disturbed by reverberation.