Application of morphological segmentation to leaking defect detection in sewer pipelines

Area of Science:

• Civil Engineering
• Computer Vision
• Image Processing

Background:

• Sewerage systems are critical urban infrastructure, with leaks being a common and significant problem.
• Closed-circuit television (CCTV) inspection is the standard method for identifying sewer pipe defects, guiding rehabilitation efforts.

Purpose of the Study:

• To propose and evaluate a novel computer vision method, morphological segmentation based on edge detection (MSED), for detecting defects in sewer pipeline CCTV images.
• To compare the effectiveness of MSED with other mathematical morphology-based segmentation methods (OTHO, CBHO) for identifying specific sewer pipe defects.

Main Methods:

• The study employed computer vision techniques, specifically mathematical morphology-based image segmentation.
• Morphological segmentation based on edge detection (MSED), opening top-hat operation (OTHO), and closing bottom-hat operation (CBHO) were applied.
• Experimental data comprised CCTV inspection images from vitrified clay sewer pipelines in Taichung City, Taiwan.

Main Results:

• MSED proved effective in detecting cracks, a common leakage defect in sewer pipelines.
• OTHO demonstrated utility in identifying open joints, another prevalent leakage defect.
• CBHO was also applied, contributing to the comparative analysis of segmentation methods.

Conclusions:

• MSED is a valuable tool for assisting inspectors in identifying cracks in sewer pipelines from CCTV data.
• OTHO effectively detects open joints, complementing MSED's capabilities for comprehensive defect assessment.
• These computer vision methods enhance the accuracy and efficiency of sewer defect detection, supporting timely infrastructure maintenance.