Research On Detection Method Of Damage And Biofouling Of Underwater Netting Based On Machine Vision

In recent years,in response to the requirement of the strategy of strong marine nation,China has vigorously developed marine aquaculture,initially using shallow beaches as the main aquaculture area,but with the development of marine aquaculture,many problems such as eutrophication of water body and over-dense aquaculture areas have emerged.This requires the development of a broader breeding space,and then aquaculture industry gradually shifts to the deep-sea area,and deep-sea breeding equipment has gradually become a research hotspot.However,due to the severe deep-sea aquaculture environment,typhoons,storm surges and other marine disasters often occur,and deep-sea areas are often accompanied by different degree of wind,wave and current coupling effects,which has high requirements for deep-sea aquaculture equipment being strong against wind and waves,some supporting facilities must also be developed in concert.The most important part in the cage structure system is the status of its netting system.If the netting system damage,it will cause quantities of fish escape,which will not only bring serious economic loss,but also may lead to the invasion of biological species,which requires regular inspections of the cage system.The current inspection work is mainly through the staff to sneak into the deep-sea environment to check the netting,mark the damaged location,and then carry out remedial work.But this method not only has low efficiency and high cost,but also cannot guarantee the safety of the staff.Therefore,the research on the non-contact detection of the netting system of the cage structure is urgently needed to provide new ideas for the smart ocean.In this paper,based on machine vision,the state of the underwater netting is obtained through the underwater robot ROV,and new non-contact netting damage detection method is proposed using digital image processing related methods to detect damage and biofouling degree of the underwater netting video.The detection is realized by the MATLAB algorithm program,and the underwater netting obtained from the real sea is tested to verify the feasibility and universality of the proposed method.The non-contact method mentioned in this article mainly includes two modules: image pre-processing module and netting status detection module.The main function of the image pre-processing module is to filter out redundant information,retain key information and enhance the quality of key information.Firstly,the key frame technology is used to filter the underwater netting image to filter out images with excessive repetition or unclear images.Then select the center area of the image through the ROI area selection technology to filter out the unclear edge of the image.After that,the underwater noise is removed from the underwater netting image through bilateral filtering operations,while retaining effective information such as the edge of the netting line.Finally,the OSTU binarization method is used to perform the binarization operation on the netting image,and the netting image is simplified into a 0-1 pixel image.Two methods are proposed in the netting state detection module.The first method is the Harris corner detection method,by detecting the corners of the net line in the netting image,and judging whether the netting is damaged according to the arrangement of the corner points of the net line,this method is simple to calculate and easy operating,but only suitable for clean netting without biofouling.The second method is the mesh feature gradient detection method,which divides the mesh into multiple independent mesh connected domains through the connected domain segmentation technology,statistically calculates the area and feature gradient of each mesh connected domain,and passes the mesh feature gradient curve to determine the abnormal point of the netting,and then locate the damaged position and display the degree of damage of the netting.This method can also detect the biofouling of the netting in parallel,and display the biofouling degree of the netting in the image.