Design and Implementation of a Video-Frame Localization System for a Drifting Camera-Based Sewer Inspection System

Abstract

To reduce the cost of inspecting old sewer pipes, we have been developing a low-cost sewer inspection system that uses drifting wireless cameras to record videos of the interior of a sewer pipe while drifting. The video’s data are transmitted to access points placed in utility holes and further transmitted to a video server where each video frame is linked to its capturing position so that users can identify the damaged areas. However, in small-diameter sewer pipes, locating drifting nodes over the full extent of the pipeline using Wi-Fi-based localization is difficult due to the limited reach of radio waves. In addition, there is the unavailability of a GNSS signal. We propose a function to link each video frame to a position based on linear interpolation using landmarks detected by the camera and image processing. Experiments for testing the accuracy of the localization in an underground sewer pipe showed that all utility holes were successfully detected as landmarks, and the maximum location estimation accuracy was less than 11.5% of the maximum interval of landmarks.

Advancing sanitary surveillance: Innovating a live-feed sewer monitoring framework for effective water level and chamber cover detections

“…The activation of the power switch marked the ensemble's entry into action, signifying the beginning of their coordinated performance. Unlike existing solutions [ [42] , [52] , [53] , [54] , [55] ], our approach consciously moves away from the use of video streams and cameras, allowing these elements to recede into the background. Instead, the study focus on spotlighting the meticulous mapping journey of the sensors, emphasizing their crucial role in the system's functioning.…”

“…In contrast, the foundation of the Data Gathering Station (DGS) rested upon the ESP8266MOD microcontroller, often integrated into the NodeMCU firmware [ 41 ], chosen for its seamless Wi-Fi connectivity. The SPI interface provided simultaneous linkage with additional wireless or alternate modules, leading to the inclusion of a RobotDyn SD card module [ 42 ] coupled with a microSD card (cumulatively priced at 4.5 USD). This arrangement established a provisional repository for sensor data during Wi-Fi disruptions.…”