2019 CSCE Annual Conference - Laval (Greater Montreal) Conference
Ms. Yuan Chen, University of Alberta
Mrs. Qin Zhang
Mr. Ahmed Bouferguene, University of Alberta, Campus Saint-Jean
Mr. Hamid Zaman, University of Alberta
Dr. Mohamed Al-Hussein, University of Alberta
Manual defect identification for sewer pipes based on closed-circuit television (CCTV) monitoring is time-consuming and error-prone. In order to address this issue, an innovative approach is to extract image frames from the video, examine whether the frames include defects, and classify these defects into different types (e.g., cracks or fractures). Specifically, a classifier based on a neural network is proposed in this paper, which consists of four parts: (1) extracting the color frames including defects from the video; (2) tailoring the frames in order to highlight the defect part; (2) transferring information in the tailored frames including defect shape and depth information into a matrix; (3) Using the matrix as inputs to generate a classifier by means of neural network; (4) Testing the built classifier. The proposed framework is then applied to a case study of Edmonton in order to automatically detect the number and location of defects on sewer pipes and conduct a condition assessment for each pipe.