Wavelet transform,empirical mode decomposition and singular value decomposition are effective means to highlight the deep anomaly and improve the signal-to-noise ratio of ground penetrating radar (GPR) data.This paper proposes a fast short-time Fourier transform technique for interpreting ground penetrating radar (GPR) noise signal.The frequency spectrum differences between transmit signal and noise signal are used to improve signal-to-noise ratio of radar data.It is verified by using a numerical example.The method and the wavelet transform as well as the advantage of empirical mode decomposition method are analyzed in detail respectively from single signal frequency energy difference,frequency equivalent distribution difference,and other aspects.The results show that the short time Fourier analysis technology can achieve rapid interpretation of GPR data and,for the ground penetrating radar data inversion imaging,it can provide more precise constraints.