基于记忆的多智能体鲁棒运动目标跟踪建模

Memory-based Multi-agent Modeling for Robust Moving Object Tracking

为了寻找目标外观的最佳匹配,将人类三阶段记忆模型引入到多智能体协同进化过程,提出一种基于记忆的多智能体系统模型用于运动目标跟踪。每一个智能体根据其自身的经验都能记住、提取或遗忘记忆系统中的目标外观。多个这样的智能体随机分布在目标区域附近,然后映射到一个二维网格环境并通过实施协同进化行为,如竞争、重组、迁移,估计目标的新位置。实验表明,该方法能够处理目标姿态变化及目标遮挡问题,并优于传统的粒子滤波方法。

The three-stage human memory mechanism was introduced into a multi agent co-evolutionary process to find a best match for the object appearance, and a memory-based multi agent system for tracking the moving objects was presented. Each agent can remember, retrieve or forget the object appearance through its own memory system by its own experience. A number of such memory based agents are randomly distributed nearby the located object region and then mapped onto a 2I） latticelike environment for predicting the new location of the object by their co-evolutio- nary behaviors, such as competition, recombination, and migration. Experimental results show that the proposed method can deal with large appearance changes and heavy occlusions when tracking a moving object, locate the correct object after the appearance changes or the occlusion recovers, and outperform the traditional particle filter based tracking method.