基于GST-Hough变换的LFM信号识别方法

针对线性调频(linear frequency modulated, LFM)信号在低信噪比条件下的信号检测问题,提出将广义S变换(generalized S transform, GST)与Hough变换相结合(generalized S transform based on Hough transform,GSTH)信号检测方法。从理论层面推导出LFM信号在进行GST后对应的参数特性,论证Hough变换的可行性,推导出GSTH变换后LFM信号与噪声的概率密度分布函数,给出了基于奈曼-皮尔逊准则进行峰值检测时,检测门限的计算方法与确定流程。利用GST时频聚焦性提供良好的直线线性,有易于Hough变换的直线检测,提升变换后主峰峰值并降低副峰高度。通过与WHT (Wigner Hough transform)、分数阶傅里叶变换与周期WHT算法的仿真对比,定量评估算法的适用性,并与经典算法对比,定性的描述出算法良好的时频聚焦性,凸显GSTH算法在强噪声背景下具有更好的检测精度与适用范围。

Recognition method of LFM signals based GSTH transform

Aiming at the linear frequency modulated (LFM) signal detection under the low signal-to-noise ratio, generalized S transform based on Hough transform (GSTH) which combines generalized S transform and Hough transform is put forward. Parameter characteristics of LFM signals after GST were derived in theory and the possibility of Hough transform was demonstrated, the probability density distribution functions of LFM signal and noise after GSTH are derived, the calculation method and flowchart of the detection threshold are given when conducting peak detection based on the Neyman-Pearson rule. The time frequency focusing property of GST is used to provide nice linear characteristics which benefit line detection of Hough transform and promoted main peak while reducing the height of the second peak. The feasibility of the algorithm is tested according to the GSTH’s simulation comparisons with Wigner-Hough transform (WHT), fractional Fourier transform and periodic WHT. Besides, compared with the classical algorithm, the nice time frequency focusing property is qualitatively described, the better detection precision and application scope of GSTH under strong noise are highlighted.