摘要
针对巡检中对变压器油位仪表读数人工采集存在效率低、准确度差的问题,论文提出一种基于机器视觉的变压器油位仪表识别技术。该技术首先对采集的指针式油位计图像进行二值化处理,随后结合膨胀腐蚀和边缘检测技术对图像进行预处理,最后采用霍夫变换算法取得指针方向和指针中心定位,完成油位计读数的自动识别。通过实验证明,论文所述的指针式油位计在实验测试中的的平均识别准确率达到98.22%,具备工程应用价值。
Aiming at the problem of low efficiency and poor accuracy in the manual collection of oil gauge reading in the inspection,this paper proposes a transformer oil level instrument recognition technology based on machine vision.The technology of collecting first binarizes the acquired pointer oil level meter image,then combines the technique of expansion corrosion and edge detection of image preprocessing,finally uses the direction of hough transform algorithm to obtain a pointer and pointer center positioning,completes the automatic identification of the oil level reading.It is proved by experiment that the average recognition accuracy of the pointer oil level gauge mentioned in this paper is 98.22%,which has the value of engineering application.
作者
郁飞
党晓鹏
张建华
YU Fei;DANG Xiaopeng;ZHANG Jianhua(State Grid Jiangsu Electric Power Co., Ltd. Maintenance Branch, Nanjing 222004)
出处
《计算机与数字工程》
2019年第7期1810-1814,共5页
Computer & Digital Engineering
关键词
指针式仪表
二值化
膨胀腐蚀处理
CANNY算子
霍夫变换
pointer instrument
binarization
expansion corrosion treatment
Canny operator
Hough transform
