Ground penetrating radar clutter removal based on MC-net network

Abstract Ground-penetrating radar (GPR) is widely used in underground target detection; however, the echo signal is often interfered with by clutter. This affects the quality of data and accuracy of target recognition. To address this limitation, in this study, an MC-Net model based on deep learning was proposed to solve the problem of clutter interference in ground penetrating radar (GPR) data. The proposed model can effectively extract features at different scales through a multi-scale residual dense block (MRDB), which improves its ability to capture the details of GPR target signals. Combined with the channel attention mechanism (CAM), the weight of the characteristic channel was adaptively adjusted to effectively enhance the target signal and suppress clutter. The MC-Net model improved the clutter removal efficacy compared with traditional methods; specifically, when processing measured data, the traditional method of clutter suppression was ineffective, whereas MC-Net effectively removed clutter and maintained a high target signal integrity. The performance and effectiveness of the proposed method were verified using comprehensive tests and measured data. Our proposed model represents an important step forward for improving the accuracy of GPR measurements.