摘要
水银血压计检定过程中人工读数误差大且效率低,为提高测量精度和工作效率,提出使用机器视觉方法实现水银血压计检定的自动读数问题。首先通过HSV颜色识别快速提取水银柱区域,利用二值化、概率霍夫直线变换、形态学处理和轮廓检测等技术对水银血压计的水银柱液面、刻度线和刻度值分别进行识别定位,采用优化训练的卷积神经网络模型(CNN)识别刻度值,通过水银柱液面与刻度线及刻度值的空间关系确定水银血压计读数。水银血压计自动读数为水银血压计自动检定过程的实现解决其核心技术问题。
To solve the problem of manual reading error and low efficiency during the verification process of mercury sphygmomanometer,the automatic reading system based on machine vision is proposed. Firstly,the mercury column area is quickly extracted by HSV color recognition. Secondly,the liquid level,tick marks and scale values of mercury sphygmomanometer are identified and located respectively by using the methods of binarization,Progressive Probabilistic Hough Transform,morphological processing and contour detection. An optimized convolutional neural network model is used to identify the scale values. And the reading of mercury sphygmomanometer is determined by the spatial relationship between the liquid level,tick marks and scale values. The automatic reading solves the core technical problem for the realization of automatic verification process of mercury sphygmomanometer.
作者
徐志凯
颜黄苹
王金珍
XU Zhikai;YAN Huangping;WANG Jinzhen
出处
《计量与测试技术》
2020年第2期1-5,共5页
Metrology & Measurement Technique
基金
莆田市科技计划项目(2018G2015)
关键词
机器视觉
卷积神经网络
HSV
自动读数
machine vision
convolutional neural network
HSV
automatic reading system