SubFuz: Subsurface Utility Detection Interface Employed Through Composite Fuzzy Clustering and Watershed Technique

• Main Authors: Enriquez, Mike Louie, Dominic Ducut, Jullian, Concepcion, Ronnie, Aristotle De Leon, Joseph, Vicerra, Ryan Rhay, Francisco, Kate, Relano, R-Jay, Bandala, Argel, Espanola, Jason, Co, Homer, Renosa, Claire Receli, Dadios, Elmer
• Format: Conference Proceeding
• Language: English
• Subjects: blob detection; Classification algorithms; Clustering algorithms; Conductivity; electrical resistivity profile; fuzzy clustering; image recognition; Image segmentation; Metals; Plastics; subsurface imaging; watershed algorithm; Watersheds
• ISSN: 2770-0682
• DOI: 10.1109/HNICEM57413.2022.10109519

Geophysical exploration is acknowledged as a useful tool for identifying and planning work quite effectively, limiting the scope of excavations, and reducing risks in urban settings. Aquifers, cavities, underground utilities, leaky pipes, and other objects can all be found using the low-cost, non-destructive subsurface imaging method known as electrical resistivity tomography. The difficult part of collecting geospatial data was using absolute or relative accuracy, along with supporting evidence. A series of steps are taken by ERT profiles to produce the 1D and 2D maps of the surveyed area. Depending on the true resistivity without inversion, this study will quickly identify underground utilities from the examined area, about a potential target object, and the identified underground utilities, such as metal and plastic pipe. Fuzzy clustering, computer vision, and edge detection techniques, such as Blob detection and the watershed algorithm, were created for analysis which denotes as SubFuz interface. Furthermore, this will also divide up various regions in the ERT profile to create an interface for underground utility detection. The following were the benefits of the new approach: (1) identify subsurface utility emphasizing previously undiscovered texture features with analyzation instead of requiring convolution from true resistivity data-intensive computation; (2) it lessens spurious blobs; and (3) it eliminates noisy spots. This method is effective for both single-feature and multiple-feature spatial feature data, making it a potent approach for segmenting noisy images.