With the advancement of urbanization and infrastructure construction, the development of underground space is becoming increasingly important. Ground Penetrating Radar (GPR) is widely used for underground detection because of its high efficiency and non-destructiveness. However, the echo signals of ground penetrating radar are often interfered with by noises including white noise, resulting in a decrease in detection accuracy. Traditional digital filters have limited effect in separating signals from noise and may cause phase distortion. To this end, this study proposes a wavelet denoising algorithm based on circular convolution to preprocess the ground penetrating radar echo signal. The simulation results show that the optimal performance can be achieved by using two-layer decomposition and sym8 wavelet basis. The proposed algorithm can effectively suppress Noise while retaining the key features of the Signal, increase the signal-to-noise Ratio (SNR) from 10 dB to 17.53dB, and significantly improve the accuracy of target positioning. The depth measurement error decreased from 5.12cm to 1.27cm. This algorithm provides a high-precision solution for underground detection and shows significant advantages compared with the existing methods.