GPR Image Recovery Effect on Faster R-CNN-Based Buried Target Detection

Measurements acquired through ground-penetrating radar (GPR) may contain missing information that needs to be recovered before the implementation of any post-processing method, such as target detection, since buried detection methods fail and cannot produce desired results if input GPR image contains information. This study proves recovery in a has direct influence on performance subsequent methods. Thus, state-of-the-art matrix completion are applied with both pixel- column-wise cases different rates, 30% 50%. After is successfully recovered, faster region-based convolutional neural network (Faster R-CNN) method applied. The correlation between accuracy method’s confidence score analyzed using quantitative visual results. obtained demonstrate importance prior implementation, detection.