Detection and Assessment of Hull Plate Corrosion Damage Based on Image Recognition Techniques

Corrosion damage can lead to a decrease in the ultimate strength and carrying capacity of ship structures, and even result in buckling or fracture. With the rapid development of deep learning, image recognition processing, and convolutional neural networks (CNN) are playing an important role in the field of corrosion damage detection and assessment. In this paper, corrosion damage images are obtained by taking photos of test pieces of accelerated corrosion tests, and a corrosion damage database is established to provide data for the establishment of the subsequent image corrosion damage level dataset and the training of the CNN model. Digital image processing methods and unsupervised learning clustering algorithms are used to identify and process corrosion images and obtain corrosion parameters. Based on the corrosion parameters, the images are classified into different corrosion damage levels to establish a corrosion damage dataset with corrosion damage level labels. This dataset is used to train a CNN model and establish a corrosion damage level assessment interface. By inputting corrosion damage images, the corrosion damage level assessment results can be obtained.