摘要
在智能视频监控领域,作为人体运动分析重要内容的行人跟踪正受到广泛关注。在实际视频监控中,利用运动摄像机拍摄会造成背景运动与行人尺寸变化,增加行人跟踪的困难。针对这一问题,为改善实现动态背景下的行人跟踪效果,文章提出了一种综合应用梯度方向直方图(HOG)和粒子滤波的行人跟踪算法。此方法以粒子滤波为跟踪框架,利用小型化的改进HOG方法实现了小尺度行人检测,并根据其检测结果确定目标,不断修正粒子采样,实现了动态背景下的行人跟踪。仿真实验结果表明,与传统的粒子滤波算法相比,本算法能够更加准确有效地跟踪动态背景中尺寸变化的行人目标。
In the field of intelligent video monitoring,pedestrian tracking has drawn great attention as an important content of human motion analysis.In actual video monitoring using moving camera to track people causes size change of people in the dynamic background,leading to more difficult to track people.To solve this problem for getting a better result of pedestrian tracking,this paper proposes an improved pedestrian tracking algorithm which synthesizes particle filter and histograms of oriented gradients（HOG） detection.The algorithm takes the particle filter as the tracking framework,and uses the improved small-scale HOG method to realize pedestrian tracking for the small scale.Also,it identifies the target area according to the result of HOG detection,and constantly modifies particle sampling,thereby realizing the pedestrian track in dynamic background.Compared with the traditional particle filter algorithm,the simulation results show that the proposed algorithm can track pedestrian target whose size is changing in the dynamic background more accurately and efficiently.
出处
《电子技术（上海）》
2011年第8期23-25,共3页
Electronic Technology
基金
安徽省科技攻关项目(09010306042)
关键词
智能视频监控
行人跟踪
动态背景
粒子滤波
梯度方向直方图
intelligent video monitoring
pedestrian track
dynamic background
particle filter
histograms of oriented gradients
