基于视觉的运动物体特征点参数测量方法研究

Research on the method of measure of mechanism motion characteristic parameters based on the machine vision technique

主要研究利用机器视觉的非接触式测量方法实现对运动物体特征点的轨迹、速度和加速度三个运动参数的测定。通过在运动物体上设定一个特征点,采用基于运动估计的模板匹配技术实时跟踪特征点在各个时间点的位置。通过对动态图像序列的分析和处理从而在线得到运动物体特征点的三个运动参数。为了实现实时在线测量,采用了基于运动估计的图像匹配算法加速了图像匹配算法的计算速度,实现实时在线测量。此算法计算一次位置点需要的时间为20ms,图像标定的精度为1μs,达到了实时在线测量的要求。测试结果表明:研究的基于运动估计的运动物体特征点测量算法运算速度快、可靠以及测试简单等特点。

It mainly researchs adopting machine vision to measure track velocity and acceleration of mechanism institution,which need not touch the measure object.It adopts pattern matching tecnnology to track label＇s location in real-time on each time ,the label is placed on the mechanism. By analysising and processing the serial image,we can get the three parameters of the mechanism institution.In order to achieving online measuring, it adopts image matching based on motion estimate to accelerate matching velocity,achieving real-time online measure.the method needs 20 ms to accomplish one image matching ,and its precision is 1μs ,it gets to the requirement of real-time online measurement. The test result shows ：the method is reliable and applied.