摘要
目标跟踪是计算机视觉应用的重要任务,兼顾算法性能和实时性是所有方法的共同目标。通过分析背景信息在目标跟踪的重要作用和影响,提出一种基于自适应背景信息概率密度函数的背景模板表示方法,改进传统核函数跟踪中的相似函数表达。实验表明,该算法在稍增加原有算法复杂度的情况下,抗背景干扰能力大大增强,并能准确跟踪快速运动目标和小目标。
Object tracking is one of the most important tasks in the application of computer vision.The consistent aim of all methods is to get better performance in the consideration of real-time.The importance and effect of background information on object tracking is analyzed in this paper,and a background template based on adaptive background information is proposed and combined with the traditional similarity function in kernel based object tracking.The proposed method is proved to be of better performance in the resistance of background interference and tracking of fast moving object and small object,at the cost of a little increase of computation.
出处
《信息与电子工程》
2011年第5期596-599,共4页
information and electronic engineering
关键词
目标跟踪
背景信息
自适应
核函数
相似函数
object tracking
background information
adaptive
kernel
similarity function
