基于机器视觉的电子秤数码管字符识别系统

Research on Digital Tube Character Recognition of Electronic Scale Based on Machine Vision

对送检产品进行质量检测时,需用标准电子秤称量并通过数显仪器显示结果。然而实际检测过程中通常用多种电子秤重复称量并人工记录结果,多种数显仪器通过一般方法难以识别。鉴于此,对基于机器视觉的多种电子秤数码管识别方法进行研究,设计人工检测辅助识别系统。首先,客户浏览器端使用摄像头采集电子秤数码管图像并以Base64编码上传服务器,对服务器端Base64格式解码并进行图像预处理,针对电子秤数码管及小数点识别等问题,提出相应解决方法;其次,对数码管字符进行分割,采用KNN加以识别;最后,将识别结果异步返回客户端,并将数据提交后台MySQL数据库予以存储和计算,实验结果识别率达99%左右。研究表明,该系统识别准确率、效率高,能快速辅助人工提高检测效率。

At present,when the inspection bureau submits for inspection of the factory electronic instrument or product weight,it needs to be weighed with a standard electronic scale and the result is displayed by a digital display instrument.This inspection process requires multiple electronic scales to repeat the weighing and manual recording of the results,It is difficult for digital display instru⁃ments to recognize the results by general methods.Based on this,this paper studies a variety of electronic scale digital tube methods based on machine vision,and designs a recognition system that assists manual detection.First,the client browser uses the camera to collect the digital tube image of the electronic scale and upload it to the server with Base64 encoding,decode the server-side Base64 format and perform image preprocessing,and propose corresponding solutions for the digital tube and decimal point recognition of the electronic scale.The digital tube characters are segmented and recognized by K-Nearest Neighbors(KNN).Finally,the recognition re⁃sults are asynchronously returned to the client,and the data is submitted to the background MySQL database for storage and calcula⁃tion.The recognition rate of the experimental results is about 99%.Studies have shown that the system has high recognition accuracy and efficiency,and can quickly assist manual efforts to improve detection efficiency.