Research On Underwater Target Trackingtechnology Of Forward Looking Sonar

Forward looking sonar as a kind of underwater detection and environment sensing equipment,has important significance for underwater exploration,research and marine development in the human beings,it mainly used in underwater target location,underwater target tracking and obstacle avoidance.The task of underwater target tracking is to predict the position of the tracking target in each frame quickly and accurately above the continuous image sequence based on the data collected by forward looking sonar.Research on fast,accurate and effective forward looking sonar underwater target tracking technology has important theoretical and practical significance no matter in military or civilian.Since the original sonar data is a gray value stored in a polar coordinate form,and the image display device is in rectangular coordinate form,and the human eye is sensitive to the change of the color than that of the gray level.Therefore,it is necessary to preprocess the forward looking sonar data,the preprocessing mainly includes the forward looking sonar image generation and the pseudo-color processing.The forward looking sonar image is generated by using parabolic eight-point interpolation to interpolate the beams and the sampling points.Then,the backward mapping coordinate transformation is adopted to generate forward looking sonar image.On this basis,an adaptive hot metal coding algorithm is proposed to complete pseudocolor processing,the node values of the hot metal coding are selected adaptively by the maximum entropy threshold method.The experimental results demonstrate that the image generation method can eliminate mosaic phenomenon and get detail-prominent sonar images.Combined with the adaptive hot metal coding algorithm can get rich colors and outstanding targets in generated sonar images,which is beneficial for subsequent tracking processing of the targets in forward looking sonar image.Because of the great difference between forward looking sonar image and optical image,many of the target tracking algorithm suitable for optical image cannot be applied to the forward looking sonar image.Combined with the characteristics of the forward looking sonar image,an adaptive particle swarm optimization(APSO)algorithm is proposed to track the underwater target in forward looking sonar image sequences.Specifically,an adaptive inertia weight is first applied to solve the problem of diversity loss and premature convergence in the PSO algorithm.Then,the individual optimal fitness value of a random particle selected from particle swarm is compared with that of the current particle,and the larger one is used to update the velocity of the particle to further prevent particles from falling into the local optimum.On this basis,when the underwater target is occluded,a new update strategy based on an adaptive discrete swarm optimization algorithm(ADSO)is used to update the particle's position according to the level of occlusion so as to achieve a better tracking result.Finally,the experimental results show that the proposed APSO can accurately complete underwater target tracking.Compared with other algorithms,the proposed APSO has higher tracking accuracy and faster tracking speed,and it has a certain effectiveness and adaptability.