基于时频和形态学滤波的长时间能量积累检测

Long-Duration Energy Accumulated Detection Based on Time-Frequency Morphological Filtering

为了检测强噪声背景下长积累时间、低信噪比、信号波形未知的机动目标,提出了一种时频分析和形态学滤波检测的方法.利用时频分布聚集信号能量在其瞬时频率曲线附近,而散布白噪声能量在整个时频平面上的特点,应用时频分布、阈值处理和数学形态学滤波估计高能量时频支撑区域,并累积这些区域的时频能量以构造检验统计量进行统计判决.该方法不需要先验波形信息,检测积累时间不受处理方法的限制,可在更长观测时间内检测噪声环境中的微弱机动目标.仿真实验表明,在信噪比为-9 dB、虚警概率为10-5时,检测概率可达99%.

A new detection method based on time-frequency analysis and morphological filtering is proposed in order to detect the maneuvering target with unknown waveform, long accumulated time and low signal-to-noise-ratios （SNR） under a severe noise background. Utilizing the characteristics that the time-frequency distribution congregates the signal energy nearby the instant frequency curve and the white noise energy is dispersed over the time-frequency plane, the time-frequency distribution, threshold processing and morphological filtering are used to estimate the support region of high energy, and then these regional energy is accumulated to construct the detecting statistics to perform the statistic decision. The proposed method does not need the prior waveform information, the accumulated time of detection is not restricted by handling methods and the weak maneuvering target can be detected in longer observing time under noisy background. The simulation results show that when the false alarm rate is 10 ^-5 and SNR is -9 dB the detection probability can achieve 99%.