GPR Missing Data Recovery via Deep Unfolding Network with Mask Guidance

Ground penetrating radar can use electromagnetic waves to conduct non-destructive testing on the ground. However, due to the lack of data in the actual acquisition, it will have a negative impact on the recognition, interpretation, matching, and localization of ground penetrating radar images. This paper proposes a method based on deep unfolding networks for recovery missing data. Based on the maximum a posteriori framework, the problem is iteratively optimized using a half-quadratic splitting algorithm. We designed the corresponding network structure based on the algorithm framework and introduced mask information to guide the optimization process. The experimental results show that our proposed method is superior to existing ground penetrating radar data recovery methods and has significant advantages.