基于粒子滤波和在线学习的目标跟踪  

Object Tracking Based on Particle Filtering and Online Learning

作　　者：刘海龙[1] 胡福乔[1] 赵宇明[1] 

机构地区：[1]上海交通大学自动化系系统控制与信息处理教育部重点实验室,上海200240

出　　处：《计算机工程》2013年第10期232-235,共4页Computer Engineering

基　　金：国家自然科学基金资助项目"多视角下的多类型目标识别与行为分析"(61175009)

摘　　要：针对粒子滤波跟踪丢失目标后较难恢复的问题,提出一种基于粒子滤波和在线学习的目标跟踪方法。使用粒子滤波有效的跟踪结果作为正训练样本不断更新样本库,将随机蕨作为分类器检测目标位置,当分类器和粒子滤波的检测结果存在较大差异时,重新初始化粒子滤波器。在线学习采用二维二值特征,具有计算简单、尺度不变和光照不变的特点。实验结果证明,该方法的跟踪结果优于传统的粒子滤波,能够准确地跟踪到被遮挡和消失再出现的目标。For the problem that the tracker is hard to be resumed when particle filtering fails to track the target, this paper introduces a method that combines particle filtering with online learning. It uses the validated result of particle filtering as positive sample to update the training set. It uses random ferns as classifier to detect object. When there is a big difference between two results, the particle filter will be reinitialized. Two bit binary pattern is used as the online learning feature. It is easy to be computed, and has invariance to illumination and scale. Experimental result proves that this method has better tracking result than particle filtering and it can track the sheltered and disappeared target.

关 键 词：粒子滤波 在线学习 随机蕨 目标跟踪 二维二值模式 巴氏距离 

分 类 号：TP391.4[自动化与计算机技术—计算机应用技术]