基于视频图像的工业仪表高速自动识别

针对工业现场高速变化的灯板仪表全自动检测的问题,提出了一种基于图像识别的检测方法。本方法采用高速工业相机获取仪表视频图像,并采用数字图像处理技术,对单帧图像用OSTU算法进行二值化分割,运用邻域平均法进行平滑处理,通过闭运算以完成去噪,运用Hough变换以实现对灯的逐一定位,并用灰度检测法对灯状态进行判读,最后保存视频。研究结果表明,该算法可靠有效,系统能够高速准确地应对仪表位置的变化,为进一步的一般化视频检测奠定了基础。

An high-speed and automatic recognition method of the industry meter based on video images

Aiming at the problems of the automatic detecting the status of a highly variable lights meter which is located in industrial site,a method based on digital image recognition was introduced.A high-speed industrial camera was used to catch video images of a lights meter.Having digital image processing technology being used,every single image frame was segmented by means of OSTU algorithm,smoothed by adjacent region average method and denoised by closing operation.Then every light was located through Hough transformation and the lights＇ status were then judged by way of gray-scale detection.At last the whole video was saved.The results indicate that the algorithm is reliable and effective while the system is sensitive and accurate to the variation of the meter＇s location which lays foundation for the futher more generalization video detecting.