Research On Key Technologies For Crack Detection Of ROV Underwater Facilities Based On Machine Vision

Underwater facilities are subject to the impact of water flow and microbial corrosion for a long time,and the surface is easy to be damaged and cracked,so cable underwater vehicles are often used to assist in structural damage detection.The underwater facilities are complex and diverse,and the situation of the water area is also more complex.It is very important to accurately and quickly determine the damage in different waters to improve the efficiency of maintenance.For this reason,this paper studies the related technology of structural damage detection of underwater facilities,in order to solve the problem of low underwater image quality,and to realize the fast and accurate detection of crack damage.The main contents of this thesis are as follows:(1)According to the application scenario of this paper,a ROV system is designed to assist the maintenance and repair of underwater equipment.Including the body shape structure design,the drive design,the power board design,the single-chip microcomputer control circuit design and the auxiliary sensing equipment selection.After testing,ROV can achieve underwater forward,steering,heading and depth movement,complete the task of image acquisition,and can adapt to a variety of water environment.(2)In order to solve the problem that the quality of underwater fracture image is low and not conducive to detection,an image preprocessing algorithm is proposed in this paper.The mechanism of underwater image degradation is fully analyzed.according to the characteristics of underwater fracture image,a relatively clear image is obtained by using the restoration method based on physical model and estimating the relevant parameters,and then denoised by the algorithm based on guided filtering.finally,the image that can be used for detection is obtained.The pre-processing of underwater images can obtain high-quality images,and the experimental results show that it can improve the accuracy of crack damage identification.(3)In view of the low detection accuracy of underwater facility crack damage image,an improved real-time image crack segmentation algorithm based on U-Net cut network is proposed in this paper.Firstly,according to the characteristics of semantic information of cracks,the detection method based on semantic segmentation is determined,then the semantic segmentation network is analyzed,and the recognition accuracy is improved based on U-Net network,and the lightweight network model is obtained by model compression,and finally the high-precision real-time segmentation of crack damage is realized.The ROV system designed in this paper can meet the real-time detection of underwater facilities,can assist to improve the efficiency of maintenance,and has a certain reference value for the field of damage detection of complex underwater structures.