摘要
从实际出发,以昆明钢铁集团公司中板厂钢板运动为研究对象,在系列帧图像中对运动目标以直方图为模型的模板方法进行匹配,由于模板匹配计算量非常大,要想在整幅图像中对目标进行搜索匹配又要达到实时是不可能的,对目标状态进行可靠的估计,就可以在相对较小的区域完成对模板的搜索,Kal-man滤波器就是一个对动态系统的状态序列进行线形最小方差估计的算法,通过以动态的状态方程和观测方程来描述系统,它可以任意一点作为起点开始观测,采用递归滤波的方法计算,它具有计算量小,可实时计算的特点。
In view of practical use ,taking the moving steel sheet of the medium-plate plant of Kunming Iron & Steel Company as object,a series of image frames of the moving steel sheet to the template method as a model for histogram matching, due to a very large amount of template matching, it is not possible to target the whole image, but also meet the real-time search and matching. If aim to conduct a reliable estimate, it can be completed in a relatively small region of the search template. Kalman filter is a linear sequence of the dynamic systems, the state minimum vafiance estimation algorithm. Through dynamic equations to describe the equation of state and observation systems, it can arbitrarily observe point as a starting point, and recursive filtering method. It is a small amount with the characteristics of real-time computation.
出处
《冶金设备》
2009年第5期6-8,5,共4页
Metallurgical Equipment
基金
云南省教育厅科学研究基金项目[基于DSP的信息整合技术的研究与开发]项目编号为042140D
关键词
卡尔曼滤波器
目标跟踪
运动钢板的预测
目标匹配
估计
Kalman filter Target tracking Moving steel sheet forecast Target matching Estimates