基于相机标定和最大连通区域算法的指针式仪表读数识别

Pointer instrument reading recognition based on camera calibration and maximum connected region algorithm

针对现有指针式仪表判读技术很少利用摄像机标定参数,自动判读准确度较低,无法满足工业生产实际需求的现状,提出将摄像机标定技术应用于圆形指针式仪表的自动判读,极大地减少了摄像机自身在图像采集过程中产生的误差;提出基于最大连通区域的仪表轮廓识别方法,提高了圆形仪表轮廓检测速度。利用SIFT算法进行仪表图像的关键点提取,利用模板匹配的方式实现图像中仪表的倾斜校正。实验结果表明,该方法的检测精确度可以达到95%以上,检测效率较传统方法提高了30%左右。

The camera is rarely used in existing automatic reading recognition technologies of pointer instrument for parameter calibration, and the low accuracy of automatic reading recognition can′ t meet the actual needs of industrial productions. Therefore,the camera calibration technology is proposed for the automatic reading recognition of the circular pointer instruments to greatly reduce the camera′ s error generated in the process of image acquisition. The instrument′ s contour recognition method based on maximum connected region is presented to improve the detection speed of the circular instrument contour. The SIFT algorithm is used to extract the key points of the instrument image. The template matching method is adopted to correct the instrument tilt in the image. The experimental result shows that the detection accuracy of this method can reach up to 95%,and the detection efficiency is improved by about 30% than that of the traditional method.