论文部分内容阅读
针对被动声呐多目标信号检测中的噪声背景归一化问题,提出了一种基于数学形态学滤波的噪声背景归一化新方法。该方法利用数学形态学处理中的膨胀和腐蚀算子,以及基于多项式拟合的数据均值估计方法,构造出了一种能够较为准确的估计噪声门限的方法,并以之进行噪声背景归一化,在较好保留信号信息的前提下较大程度的抑制了噪声,有效降低了多目标信号检测的虚警概率。通过计算机仿真对比了该算法与S3PM算法、OTA算法的性能,结果表明该噪声背景归一化算法能够在检测概率损失较小的情况下较大幅度地降低检测的虚警率。实际被动声呐数据处理的对比结果同样验证了该算法的有效性。
Aiming at the normalization of noise background in passive sonar signal detection, a new normalization method of noise background based on mathematical morphology filtering is proposed. This method uses the expansion and erosion operators in mathematical morphology processing and the data mean estimation based on polynomial fitting to construct a method which can estimate the noise threshold more accurately and normalize the noise background , Which can restrain the noise to a great extent while keeping the signal information well and effectively reduce the false alarm probability of multi-target signal detection. The performance of the algorithm and the S3PM algorithm and the OTA algorithm are compared by computer simulation. The results show that the noise background normalization algorithm can greatly reduce the detection false alarm rate under the condition of small detection probability loss. The comparison of actual passive sonar data processing also verifies the effectiveness of this algorithm.