Research On Target Tracking Technology Based On Improved Chicken Swarm Optimization

With the rapid development of computer vision,numerous research results have been successfully applied to modern civil and military missions.The tracking algorithm based on matching can track the target by establishing a reference template containing target information and searching the whole image for template matching.However,in practical application,single feature can't better describe the target characteristics in complex tracking scenes.When some interference factors is occluded in the tracking process,it is easy to lead to the phenomenon of deviation and loss of the tracked target.Therefore,the fusion of different features can more effectively express the target characteristics and improve the accuracy of the tracking algorithm.On the other hand,the chicken swarm optimization algorithm is a kind of intelligent optimization algorithm with multi group cooperation,which has the adaptive ability and good expansibility to solve the optimization problem.But the traditional chicken swarm optimization algorithm has some defects,such as the lack of population diversity,premature convergence and easy to fall into the local optimum.This paper analyzes the influence of the limitations of chicken swarm optimization algorithm on the target tracking performance,and proposes a target tracking algorithm based on improved chicken swarm optimization.The main work of this paper is from the following five aspects:(1)Target feature fusion: As HSV color feature is more suitable for the human visual observation angle and not sensitive to target deformation,and Hu-moment invariants have the advantages of translation,expansion,rotation invariance and small calculation in continuous image frames.Thus HSV color feature and Hu-moment invariants feature are combined to establish target reference template.The experimental results show that the chicken swarm optimize target tracking algorithm based on feature fusion can effectively improve the accuracy of target tracking and has good adaptability in a wide range of scenes.(2)Weighted central individuals: Due to the most of the chicken individuals gather to the local extreme points in the late stage of the search,the diversity of the chicken group is lost and it is easy to fall into the local optimal solution leading to the phenomenon of premature convergence.In order to avoid these defects,a weighted central hen is introduced into thehen group.It competes with all the individuals in the whole chicken group for the global optimal individual position,which increases the diversity of the population.(3)Adaptive inertia weight: An adaptive inertia weight is introduced to coordinate with individual's search ability in different periods.The size of inertia weight is determined by the individual fitness value and the current number of iterations.Generally,in order to quickly locate the area around the global best point in the early stage of search,the inertia weight should be set larger to enhance the global search ability.While in order to get the global best,the inertia weight should be reduced gradually to enhance local search ability.(4)Improved chicken group position updating method: During location update,the rooster individual should not only inherit their own position,but also learn from the best individuals in the current population.The hen individual updates the position by learning the position information of the best rooster and other random individuals,but hen individuals lack of self-learning ability.Also in the process of chick updating,it only learns from the mother of the hen,which lacks the direction of optimization.In view of these shortcomings,this paper introduces weighted center individual and adaptive inertia weight to improve the above problems.(5)Anti-occlusion template update strategy: In order to effectively respond to scene changes,this paper proposes an anti-occlusion template updating strategy.The occlusion threshold judgment mechanism is introduced to determine whether the target is occluded,and the template is updated accordingly.By adding the anti-occlusion template updating strategy,when the target is occluded,it can ensure the timely recovery of accurate and effective tracking,and improve the robustness of target tracking.