A MULTI-FREQUENCY GPR DATA FUSION TECHNOLOGY BASED ON END-TO-END DEEP NEURAL NETWORKS

As one of the most important non-destructive testing technologies, ground penetrating radar (GPR) technology detects the internal electromagnetic characteristics and distribution by transmitting electromagnetic waves. High-frequency electromagnetic waves have high resolution but shallow detectable depth, while low-frequency electromagnetic waves have wide detectable range but terrible resolution. In order to obtain GPR signals with both high resolution and great detectable depth, we propose a GPR multi-frequency data fusion technology based on end-to-end deep neural network. In the first part, the features of the A-scan echo signals at different scales are obtained through the convolutional layer. In the second part, these features are fused using maximum values. Finally, the high-dimensional information is decoded to return the A-scan signal. 濜n our paper, compared with the original single-frequency signal, the fusion signal achieves higher resolution at greater depth.