Research on Crack Size Identification Method Based on Digital Image Processing Technology

Cracks is one of the common hazards in offshore structures, and continuous crack propagation can cause serious consequences such as steel structure fracture. The steel structure adopts a regular inspection method to detect structural cracks. Once cracks are found, emergency repair and maintenance work is carried out to ensure the safety of the structure. One of the keys of crack visual inspection is to accurately extract crack size from image. Based on digital image processing technology, the camera calibration, image processing and crack feature extraction were achieved by OpenCV and Skimage open source algorithm package, and then the crack pixel size was extracted. The crack pixel size is converted to the actual size by pixel calibration. The crack length inspection error is 4.8%, and the maximum width inspection error is 13.7%. The results show that the crack identification method has high crack inspection accuracy,which is worth spreading and applying.