Research On Underwater Multi-target Tracking Algorithm Of Forward Looking Sonar

Marine resources are rich and valuable,so it is of great significance to explore the ocean in both military and economic fields.However,the complex underwater environment has been a constraint to the observation and utilization of the ocean.As one of the most effective and important devices in underwater acoustics,sonar is widely used in underwater operations such as seabed survey,obstacle detection,underwater target identification and underwater target tracking.A contemporary focus of research in the realm of ocean development and exploration is the study of underwater multi-target tracking technology based on forward-looking sonar.This is due to the systematic development of sensor technology,communication technology,computer technology,and ocean exploration technology in the modern era.Two major issues with the thorough development and use of underwater multi-target tracking based on forward-looking sonar are as follows: 1.How to implement multi-target tracking based on sonar picture series.2.How to improve the accuracy and robustness of underwater multi-target tracking.To address the above difficulties,this thesis proposes to combine the information theoretical basis,apply the information entropy class criterion to the joint probabilistic data association algorithm,and give some algorithms that can achieve robust multi-target tracking in the underwater non-Gaussian noise and nonlinear environment.The following research is a focus of this thesis.(1)Most current research on underwater target tracking topics only considers single-target tracking,and the proposed tracking algorithms and frameworks are often not applicable to multi-target tracking,this thesis proposes the use of joint probabilistic data association algorithms for underwater multi-target tracking.The traditional joint probabilistic data association algorithm uses the mean square error as the optimization criterion,which has the disadvantage that it only works well in Gaussian environments,and the performance of the algorithm tends to deteriorate in non-Gaussian noisy environments.(2)For the characteristics of underwater environment which is mostly nonlinear and non-Gaussian,we provide the joint probabilistic data association algorithms based on the maximum correlation entropy criterion and the minimum error entropy criterion,respectively,furthermore,their extended algorithms are further given for the study of nonlinear problems.The former replaces the traditional minimum mean square error criterion with the maximum correlation entropy criterion as the optimization criterion,and following derivation,a strong multi-target tracking algorithm that can deal with nonGaussian noise in the environment is suggested.The algorithm works better in the nonGaussian noise environment when the number of targets,the clutter density,and the target proximity varies,according to simulation data.In addition,it derives and proposes a joint probabilistic data association algorithm based on the minimum error entropy criterion and its extension algorithm,analyzes its computational complexity,and substitutes the minimum error entropy criterion for the conventional minimum mean square error criterion.According to the simulation results,the approach works better when dealing with complex non-Gaussian noise.(3)Finally,the proposed algorithm is applied to the forward-looking sonar underwater multi-target tracking,and experiments are conducted for underwater normal target,underwater noise-affected target and underwater small target tracking,the experimental results demonstrate that the suggested method performs better than the conventional technique in terms of tracking accuracy,noise immunity,and robustness when used to monitor several targets underwater.