The Application of Improved Small Sub-Domain Filtering in Ground Penetrating Radar Data Enhancement Processing: A Case Study of Embankment Detection in Hunchun City

Abstract

The quality of ground penetrating radar data is vital for subsequent geological interpretation. Conventional filtering methods often damage the boundary features of geological bodies and affect the accuracy of the interpretation. This paper proposes the use of small sub-domain filtering to enhance ground penetrating radar data. We improve the "Mi" type sub-domain division method and mean square deviation discrimination criterion of traditional small sub-domain filtering. The improved small sub-domain filtering uses the parallelogram division and mean square deviation of the total gradient magnitude as discrimination criterion. Using gprMax to construct geological models of different depths of soft soil and different soil contact interface fluctuations for test, we discuss the effectiveness of traditional and improved small sub-domain filtering in identifying geological body boundaries at different window sizes under the influence of Gaussian noise. Model experiments show that using improved small sub-domain filtering to enhance ground penetrating radar data leads to a higher degree of compression of the phase axis of the geological boundary reflection waves, making them more continuous, clearer, and less deformed. Increasing the filtering window can reflect the boundaries of deep geological structures. In application, the enhanced ground penetrating radar data of the Hunchun City embankment detection show that the improved small sub-domain filtering can preserve the boundary features of geological body boundary, enhance their recognizability, and obtain clearer and more accurate boundary feature information.