论文部分内容阅读
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。
From a practical point of view, taking the plate movement of the medium plate factory of Kunming Iron and Steel Group as the research object, matching the template of the moving target with the histogram as the model in the series of frame images. Because the calculation of the template matching is very large, It is impossible to search and match the target in real time in the image. The reliable estimation of the target state can search the template in a relatively small area. The Kalman filter is a dynamic state system This algorithm can describe the system by using dynamic equation of state and observation equation. It can start from any point as a starting point and calculate by recursive filtering. It has the characteristics of small calculation and real-time calculation.