Underwater robot image enhancement method for detecting cross-sea bridge structure damage

Abstract

The invention provides an underwater robot image enhancement method for detecting damage to the cross-sea bridge structure, the cameras carried by the underwater robot carry out underwater shooting in a three-camera parallel mode, and the problems that the focal length of underwater shooting becomes large, and the field angle becomes small are solved. Through the auxiliary light source and the jitter elimination compensation device, scattering of water is reduced, jitter is eliminated, and image definition is improved. The method comprises the following steps: preprocessing an underwater image, graying the image, reducing the interference of image dithering by adopting a dithering elimination algorithm, reducing interference noise generated by underwater floating animals by adopting a multi-image averaging method, and processing the problem of image degradation caused by uneven underwater illumination by adopting a homomorphic clarification method; then performing image segmentation operation, an image segmentation algorithm is combined with an edge segmentation algorithm based on a Canny operator and closed operation, and a complete crack can be obtained while noise is effectively suppressed; and finally, performing crack identification by adopting an identification method based on a convolutional neural network.

Description

technical field [0001] The invention belongs to image recognition technology and deep learning, and in particular relates to an underwater robot image enhancement method for detecting structural damage of a cross-sea bridge. Background technique [0002] At present, the commonly used detection methods for bridge cracks include: optical fiber sensing detection technology, ultrasonic detection technology, ground penetrating radar detection technology and detection technology based on image processing. These detection methods have certain advantages in the field of non-destructive testing, but the detection technology based on optical fiber sensing needs to lay fiber optic sensors in advance, which is costly; ground penetrating radar detection technology is limited by the environment, and its use under water is limited; ultrasonic detection technology It is impossible to obtain visual images of defects, making qualitative determination difficult and quantitative accuracy low. ...

AI Technical Summary

Problems solved by technology

[0006] The underwater camera structure includes a three-column combination camera, two upper and lower anti-shake compensation devices, a waterproof protective case, and a support structure. The three-column combination camera is specifically composed of three groups of cameras placed side by side. To solve the problem that the focal length of underwater photography becomes larger and the field of view becomes smaller, the anti-vibration compensation device is composed of a buffer tube, a buffer rod and a soft plate to reduce the jitter during shooting and improve image clarity. The shape of the waterproof protective shell It is streamlined, which can reduce the resistance when the underwater robot moves, and the built-in card slot is used to fix the camera

Images