摘要
针对视频中的多目标跟踪问题,提出一种改进的基于数据关联矩阵的多目标跟踪算法,实现视频场景复杂环境下的多个目标跟踪。使用区间分布模型获取图像的背景和前景,对前景目标建立相应的运动模型。根据运动模型和Kalman滤波器的位置预测,建立相关的匹配代价函数、关联矩阵和匹配链表。实验结果表明,该算法对目标在场景中的频繁出现和消失、交叉运动和短暂遮挡等均有较好的处理效果。
In order to solve multi-object tracking problems,based on data association matrices,an improved multi-object tracking algorithm is established,and multi-object video tracking is achieved under complex scenes.Section-distribution model is established to acquire background and foreground,and motion model is built for foreground target.According to the prediction of motion model and Kalman filter,corresponding matching cost function,correlation matrix and matching list are set up to solve kinds of tracking problems under complex scenes.Experimental results demonstrate better processing effect when object is exposed in scenes such as frequent appearance and disappearance,mutual cross-motion and short-time occlusion.
出处
《计算机工程》
CAS
CSCD
北大核心
2010年第23期158-161,共4页
Computer Engineering
关键词
智能交通系统
多目标
跟踪
Intelligence Traffic System（ITS）
multi-object
tracking
