计算机视觉在仪表数字识别中的应用

Application of computer vision in instrument number recognition

为克服人工读数存在的局限性问题,提高数字仪表读数效率,文章利用计算机视觉技术,自行构建样本数据库并训练卷积神经网络模型识别仪表数字。首先,采集仪表图像并对其进行预处理;其次,对预处理后的字符进行定位和切割;最后,引入训练过的网络模型对仪表数字进行分类识别。实验结果表明,该方法稳定性好、鲁棒性高、识别速度快,可以满足各类场景下的数字仪表读数需求。

In order to overcome the liminations of manual reading and improve the efficiency of unmanned digital meter reading,this paper uses computer vision technology to build a sample database and trains convolutional neural network model to identify meter numbers.Firstly,the instrument image is collected and preprocessed.Then,the preprocessed characters are located and segmented.Finally,the trained network model is introduced to classify and identify the meter numbers.Experimental results show that the method has good stability,high robustnes and fast recognition speed,which can meet the needs of digital instrument reading in various scenarios.