首页 | 本学科首页   官方微博 | 高级检索  
     

一种基于直方图模式的运动目标实时跟踪算法
引用本文:胡明昊,任明武,杨静宇.一种基于直方图模式的运动目标实时跟踪算法[J].计算机工程与应用,2004,40(3):71-74.
作者姓名:胡明昊  任明武  杨静宇
作者单位:南京理工大学计算机系,南京,210094
摘    要:动态图像的分析和理解是当前研究的热点之一,基于视觉的目标跟踪技术有着广泛的实用价值。目标跟踪的难点在于完成帧与帧之间的快速且稳定的目标匹配。该文给出了一种运动目标的跟踪算法,它与云台设备控制相结合,可使被跟踪目标始终位于图像的中心区域。直方图具有较好稳定性,可以不受目标的外形和比例变化的影响;而均值平移(MeanShift)算法可以得到局部最优解,并具有快速和有效的特点。因此,该文以直方图为模式特征,以均值平移算法为跟踪核心算法。对候选目标进行运动检验,过滤了伪目标,保证了跟踪的可靠性。在搜索过程中,通过Kalman滤波器的运动预测,减少模式匹配的搜索范围,提高了处理速度。最后,该文给出了实地测试结果,验证了跟踪算法的实用性和有效性。

关 键 词:目标跟踪  均值偏移  直方图  卡尔曼滤波器
文章编号:1002-8331-(2004)03-0071-04

A Real-time Moving Object Tracking Algorithm Based on Histogram Model
Hu,Minghao Ren Mingwu,Yang Jingyu.A Real-time Moving Object Tracking Algorithm Based on Histogram Model[J].Computer Engineering and Applications,2004,40(3):71-74.
Authors:Hu  Minghao Ren Mingwu  Yang Jingyu
Abstract:Analyzing and understanding dynamic images have been focused in many research fields.Object Tracking is one of its applications,which has great practical value.The difficulty in visual tracking is how to perform and match targets from frame to frame quickly and reliably.This paper presents a real-time tracking system which tracks moving object and makes the target always lie in the center of image frame assisted by controlling a servo device.Histogram character is not affected by object figure's variance and scale and has good stability.Mean shift algorithm is an algorithm that can calculate optimal result fast and effectively.Therefore,the kernel-tracking algorithm in the system is based on the mode of histogram and its searching method is Mean Shift algorithm.The motion verification will remove the false targets to guarantee the reliability of algorithm.Kalman filter is used to estimate motion parameter of targets,which will reduce searching scope during model matching.Finally,the paper gives the experiment results,which show that the algorithm is practical and efficient.
Keywords:Object Tracking  Mean Shift  Histogram  Kalman filter  
本文献已被 CNKI 维普 万方数据 等数据库收录!
设为首页 | 免责声明 | 关于勤云 | 加入收藏

Copyright©北京勤云科技发展有限公司  京ICP备09084417号