Improving Water Supply System Components Visualization into GPR Images

Abstract

This work focuses on the use of easy-to-apply procedures that allow rapid visualization of components of water supply systems (WSSs) by non-highly qualified personnel. We use a methodology that does not alter the conditions and characteristics of the environment (nondestructive methods), specifically the study of images obtained with ground penetrating radar (GPR). The study is based on the analysis and interpretation of the wave amplitude, then applying a series of image corrections, so that the display and handling of data is improved. The results are promising as a subjective and repeatable methodology to visualize buried pipes efficiently. The goal is to generate know-how to be able to train intelligent systems for the characterization of components of WSSs.

Advancing towards semi-automatic labeling of GPR images to improve visualizations of pipes and leaks in water distribution networks using multi-agent systems and machine learning techniques

“…Thus, non-destructive testing methods have been incorporated into the surveys to address the need for non-invasive procedures that provide relevant information on buried assets, including WDNs, gas pipelines, and more [7]. Nondestructive testing methods such as ground penetrating radar (GPR) are easy to deploy, but they present several difficulties (e.g., absence of consistent GPR data interpretation protocols, excessive noise, and heterogeneity of environmental conditions and equipment settings [8]) in the data interpretation process.…”

“…GPR signals are capable of detecting materials, both metallic and non-metallic [14] (e.g., PVC, asbestos cement, polyethylene), which are commonly used in assets of WDNs. GPR has the sensitivity to detect leaks in the media surrounding pipes [8], and this non-destructive method also offers accurate depth estimation and has an advantage of high resolution of up to a centimeter [15]. In the field of leak detection, other non-destructive methods such as thermography and tracer gas are limited by temperature, leak depth, and disruption to WDNs [16].…”

“…The 3D models generated from the information obtained after processing the GPR images can be used to feed augmented reality visualizations that can facilitate the assets assessment process [25] in a dynamic manner. The literature has examples of 3D model generation ranging from simple raw data plotting [16] to contours created from a series of pre-processed cross sections [8,26]. These contours created from pre-processed cross sections are a significant step towards interpretable GPR images; however, they rely on comparison of initial and final states of leaking pipes to extract the relevant features and construct a 3D image.…”