Leakage Detection Using Ground Penetrating Radar C-Scan Based on 3D Fuzzy C-Means Clustering

This paper presents an unsupervised change detection approach for underground leakage detection using ground penetrating radar (GPR) C-scan images. Superpixel and 3D fuzzy c-means are introduced in difference image(DI) generation and classification processes. Converting DI into superpixels reduces the negative influence of local random noise. The fuzzy c means algorithm was reformulated by introducing a spatial factor to involve 3D neighborhood information among adjacent slices. Experiments on field data demonstrate that the proposed algorithm has good capability to detect leakage areas in time series measurements.