噪声中的线谱检测及自动提取研究

线谱是水中目标检测与识别的重要特征,其自动提取对于水中无人探测系统至关重要。采用统计信号分析方法得出了线谱检测的接收机工作特性曲线,理论分析了线谱自动提取时一定虚警概率下的峰高门限,并提出一种自动门限以适应不同背景方差的变化。实测数据分析表明,本文方法在不同噪声背景下自动提取线谱的效果较好。

Detection and automatic extraction of tonals buried in noise

Tonals are of importance to underwater target detection and recognition,and their automatic extraction is indispensable for unmanned passive sonar system.We obtain the receiver operating characteristic(ROC) of tonal detection with statistical signal analysis,and theoretically analyze the amplitude threshold for recognizing peaks automatically in the power spectrum with a certain probability of false alarm.We then propose a kind of adaptive threshold to meet the fluctuation of ambient noise background in engineering.The results of sea test data processing show that the adaptive threshold works better than the fixed threshold, thus bringing fairly good performance in toanl recognition and automatic extraction.