Research On Underwater Artificial Target Detection Of Fractal-based Isolation Forest Algorithm

There are vast sea areas on the earth,and rich marine resources have become an important factor for people to survive.How to use and protect this resource has gradually become the focus of people's attention.The detection of underwater artificial targets is an important part of protecting water resources,and the sonar image analysis associated with it has attracted more and more attention from academic circles at home and abroad.Related detection algorithms have been proposed so far.However,some of these algorithms have complex and massive calculations,and some use only limited data information,and there is still room for improvement.Therefore,underwater artificial target detection technology still needs to be further explored.In this paper,the method of detecting the underwater artificial target of isolated forest based on fractal theory is studied,which realizes the fast and accurate detection of the artificial target in sonar image.The main research work is as follows:(1)Based on the acoustic characteristics of artificial targets in the sonar image,a new detection algorithm is proposed which combines with fractal theory and the isolated forest algorithm.First,according to fractal theory,the fractal dimensions of the data points in the sonar image is calculated.Find the dimensions with obvious acoustic characteristics of the artificial target data points,and combine the fractal dimensions to extract the fractal characteristics of the data points.Second,under the premise of ensuring the original data distribution characteristics,the sonar image data is sampled to reduce the impact of a large amount of data on the detection process,and quickly build separation trees.Then,use the separation trees to separate the sonar data and determine the position each data point is on the separation tree.Finally,according to the position of the data points in the separation tree,the abnormal value of each data point is calculated,and the detection of the artificial target data points is finished.(2)An artificial target detection optimization algorithm based on genetic algorithm is proposed.This method first uses genetic algorithm to encode the parameters involved in the setting of the parameters.The automatic selection of the optimal values of the parameters is achieved while eliminating the impact caused by manual values.Subsequently,in view of the problem of dimensional redundancy in multi-dimensional data,this paper proposes a screening criterion to improve the detection rate of the detection algorithm while removing the redundant data dimensions.Finally,the method utilizes a new screening method to determine the cut point,and optimizes the construction of the separation tree,thereby achieving a more accurate positioning of the data points on the separation tree.(3)The algorithm in this paper is verified by simulation experiments in different types of sonar data,and compared with the more classical detection algorithm.In the simulation experiment,noise group data is added to verify the performance of the algorithm in complex environment.The experimental results show that the algorithm proposed in this paper is effective.