Artificial Reef Detection And Evaluation Based On Multi-Beam Point Cloud

Artificial reefs are closely related to the construction of marine ranching.The quality of their delivery has greatly affected the effects of artificial reefs on the proliferation and maintenance of fishery resources and the restoration of marine biological habitats.However,due to the limitation of technical level and the complex and changeable sea conditions,there is often a lack of corresponding monitoring and evaluation methods after the release of artificial reefs.With the development of acoustic means such as side scan sonar and multi-beam sounding system,the traditional artificial diving exploration has been gradually transformed into artificial reef detection and evaluation based on acoustic means.It is the main research direction in recent years to use the side scan sonar scanning to obtain the seabed artificial sonar image,and then use the image target recognition method to extract the artificial reef.However,the twodimensional sonar image takes the shadow feature of the seabed target as the main recognition feature,and there are few studies on the recognition and extraction of the seabed target point cloud.With the deepening of research,the detection of artificial reefs based on sonar images has also exposed some problems.For example,the twodimensional image of artificial reefs obtained by side scan sonar has large deformation,and the three-dimensional information such as artificial reef height and internal structure is missing.Therefore,the recognition and extraction of targets such as artificial reefs using underwater three-dimensional point cloud data has gradually become a new research hotspot.Based on this actual demand,this paper has carried out a series of research work based on artificial reef multi-beam point cloud data,mainly in the following aspects :(1)In this paper,the underwater sonar image of the artificial reef area is obtained by using the multi-beam sonar carried by the unmanned ship.Through a series of processing work such as tide level,sound velocity and attitude correction,the original three-dimensional point cloud is generated.Through a series of preprocessing work such as point cloud segmentation,point cloud radius filtering denoising and Laplace point cloud smoothing in non-research area and research area,high-quality point cloud in artificial reef area is finally generated.(2)The location analysis of artificial reef area is introduced systematically,and the automatic identification and extraction of artificial reef based on RANSAC algorithm and CSF algorithm are described in detail.The recognition and extraction of RANSAC and CSF algorithms are carried out under the basic conditions of wide and flat underwater terrain in artificial reef area.The recognition and extraction of artificial reefs are realized by separating seabed point cloud and reef point cloud,and the actual extraction effects of the two are compared and analyzed.The experimental results show that the accuracy and integrity of artificial reefs extracted by RANSAC algorithm are 94.79 % and 91 %respectively,while the accuracy and integrity of artificial reefs extracted by CSF algorithm are 91.45 % and 87 % respectively.Compared with the CSF algorithm,the RANSAC algorithm has more advantages in extracting artificial reefs,and has better results.In particular,it has higher recognition of single reefs with lower distribution density,higher accuracy and easier operation.(3)There is a certain synergy between artificial reefs.Whether the empty volume of artificial reefs meets the predetermined target is an important aspect to measure the quality evaluation of artificial reefs.This paper improves the point cloud clustering algorithm DBSCAN and applies it to the clustering analysis of artificial reefs.The improved method adopts the kernel density estimation method to realize the adaptive acquisition of Eps parameter interval,and then introduces the Min Pts parameter method to realize the reef.The improved method adopts the kernel density estimation method to adaptively obtain the Eps parameter interval,and then introduces the Min Pts parameter method to realize the clustering of the reef point cloud,and introduces the convex hull algorithm to calculate the volume for the clustering.In this paper,the Convex hull function is called in python to calculate the convex hull.By calculating the total empty volume of 8 units of reefs,the calculation of the empty volume of artificial reefs is realized,which provides a certain reference for the scientific quantitative evaluation of artificial reefs.