Dou-edge evaluation algorithm for automatic thin crack detection in pipelines

Abstract

This paper describes and evaluates a novel computer vision algorithm for automatic thin crack detection in pipelines using dou-edge evaluation (DEE). Inspection for pipes is crucial and it is performed periodically to ensure that the structured integrity of the pipe systems is maintained. Thin cracks and fractures are among the defects which can cause critical damage to pipe systems. Numerous techniques have been used to detect cracks in pipes including machine vision, mostly based on edge-detection algorithms (i.e. Sobel, Laplace). However, these algorithms encounter difficulties in extracting cracks from complicated and noisy environments (i.e. sewer pipes). The DEE algorithm overcomes this problem by evaluating the size and shape of each object in the inspection environment. The results show that thin cracks were automatically extracted by the proposed algorithm.