ºìÐÓÊÓƵ

Topic

Automatic Characterization of Surface-Breaking Crack Defects in Pipe Walls Using Ultrasound Images

Department of Electrical and Computer Engineering

Date & location

• Friday, April 25, 2025

• 1:00 P.M.

• Virtual Defence

Reviewers

Supervisory Committee

• Dr. Daler Rakhmatov, Department of Electrical and Computer Engineering, University of Victoria (Supervisor)

• Dr. Panajotis Agathoklis, Department of Electrical and Computer Engineering, UVic (Member) 

External Examiner

• Dr. Barbara Sawicka, Department of Mechanical Engineering, UVic 

Chair of Oral Examination

• Dr. Caterina Valeo, Department of Mechanical Engineering, UVic

   

Abstract

Ultrasound imaging is a widely used technique in non-destructive testing for detecting and sizing defects in industrial pipelines. Accurate defect localization and sizing are critical for diagnosing the structural integrity of pipelines and preventing failures that pose significant environmental, economic, and safety risks. Motivated by the need to mitigate such risks, this thesis presents an effective technique for localization and sizing of surface-breaking cracks on the outer walls of liquid-filled pipes.

The proposed approach combines traditional and novel data processing techniques applied to a sequence of multi-view ultrasound image frames of inspected pipe sections. Tri-sectional sliding windows are utilized for frame-by-frame view-specific defect localization, followed by establishing correspondence among potentially differing crack location estimates across all considered views and frames, as well as for defect sizing. Additionally, a similarity-based sizing method is developed to increase the accuracy by comparing synthetic images to the original ones. Our evaluation results using real-world experimental data demonstrate that the sliding window method is computationally inexpensive and yields accurate localization and sizing results in most cases, while the similarity-based method provides superior sizing accuracy in more complex scenarios.