Research Of Probability Hypothesis Density Algorithm Based On Sequential Monte Carlo Method

Many classic multiple target tracking algorithm have been sprung up since the multiple target tracking is been proposed,such as kalman filter and particle filter target tracking algorithm has good filter performance,has been widely used in military,civil,and many other fields.Since the classical filter hypothesis model has certain limitations,lead to it is not available to many complicated situations,This urgently requestes a more universal algorithm in order to satisfy the modern accurate real-time tracking of the target of science and rapid positioning technology on the basis of the classical algorithm,Hence the study of multiple target tracking is still has very important practical significance and theoretical value.The Probability Hypothesis Density Filter Based on Sequential Monte Carlo Method combines the monte carlo thoughts and the probability hypothesis density filter algorithm effectively,abandones the data correlation method,has good filtering effect and not limited by the model,can be applied to all kinds of nonlinear situation with a lot of samples to approximate posterior probability hypothesis density of the target.On this basis,this paper carried on the thorough research to its,the main contents are as follows.Firstly,this paper study in depth the bayesian filtering,the core ideology of monte carlo theory and the main step of particle filter algorithm;Secondly,we analyzed multi target model based on random finite set in detail;Furthermore research main steps of the probability hypothesis density filter and one way of its implementation--SMC-PHDF.When system measurement noise is bigger,a lot of small samples of normalized weights become invalid samples,leading to result of the SMC-PHDF filtering precision is low.Aiming at this problem,this paper proposes an adaptive SMC-PHDF algorithm,The algorithm keep trying the thorough analysis of adaptive likelihood distribution weights update method,illuminate the applicability of the algorithm in the SMC-PHDF,and improves the update steps of SMC-PHDF algorithm based on this,effectively introduce the method of dynamic adjustment particle weights to adaptive adjusting the particle weight in order to improve the filtering performance.Then,compare the Matlab simulation the improved the SMC-PHDF algorithm with the original algorithm,the simulation results show that:when the measurement noise is bigger,the proposed adaptive SMC-PHDF algorithm filtering performance is superior to the conventional SMC-PHDF algorithm.In order to enhance the execution efficiency of SMC-PHDF algorithm,and make full use of parallel computing features of multi-core CPU,This paper proposes a SMC-PHDF algorithm based on the mode of parfor parallel,In addition,we analyzed the structure of SMC-PHDF,one part of the program is found out which is satisfied the parallel mode,then with introducing parfor mode to modify SMC-PHDF algorithm,thus reduce the time complexity of the algorithmis.Finally parfor is used to deal with this part of parallel computing,each particle is updated using Matlab parallel mode.Simulation experiments show that the parfor parallel mode can be used smoothly,the computational time is reduced in the update procedure,and the closed feature is satisfied.