基于嵌入式ARM的指针式仪表读数的识别研究

Research of Recognition for Pointer Angle of Dial Based on Embedded ARM

指针式仪表自动读数识别,是实现各种指针式仪表自动检定最重要的基础.针对嵌入式ARM系统,采用了一种适合实际设备运行的指针角度识别算法——差影法,并利用最小二乘法拟合指针直线方程.系统充分利用图像动态采集设备和基于多线程系统平台,把图像处理的算法和特征提取的算法进行优化结合,使得采集和判读以多线程方式并发执行.该系统实时性较好、误差小,通过实验结果证明这种算法对于嵌入式ARM系统下的指针式仪表读数的识别具有高效性.

The automatic recognition technology of analog measurement instruments is very important for automatic calibration devices. In this article, a new method for the point angle recognition is brought forward, which is specially fit to running equipments, and then least square method is used to approximate the point equation. It make full use of dynamic image acquisition equipment and multi-threaded platform, the image processing algorithms and feature extraction algorithms to optimize the combination of acquisition and interpretation in order to make multi-threaded implementation of the complicated way. The system has the traits of easy calculation, good real time and low recognition error. Through a great deal of different condition experiments, this kind of calculation method is proved to be very valid for the point readings of the embedded ARM system.