基于改进穿线法和KNN的数码管字符识别研究  

Character recognition research of digital tube based on improved threading method and KNN

作　　者：刘祎爽 黄理瑞 魏敏捷 LIU Yishuang;HUANG Lirui;WEI Minjie(College of Electronics and Information Engineering,Shanghai Univercity of Electric Power,Shanghai 200135,China;GL Tech Co.,Ltd.,Zhengzhou 450001,China;School of Electrical Engineering,Shanghai Univercity of Electric Power,Shanghai 200135,China)

机构地区：[1]上海电力大学电子与信息工程学院,上海200135 [2]光力科技股份有限公司,河南郑州450001 [3]上海电力大学电气工程学院,上海200135

出　　处：《电子设计工程》2024年第4期12-16,共5页Electronic Design Engineering

基　　金：国家自然科学基金面上项目(61872230)。

摘　　要：针对传统穿线法过度依赖数码管字符分割效果、无法对小数点字符进行识别以及机器学习算法识别数码管用时过长的问题,提出了基于改进穿线法与KNN算法相融合的数码管字符识别方法,达到了对不同数码管字符及小数点识别的目的,减少了对字符预处理效果的依赖。在嵌入OpenCV图像处理程序的LabVIEW人机交互平台采集到实时图像后,输出识别结果。经多次实验,该方法的识别时间相比单独使用KNN的识别时间明显缩短,识别率可以达到95%以上,具有识别速度快、精度高的优势。For threading method relying too much on traditional digital tube character segmentation effect,can't the decimal point character recognition and machine learning algorithms of digital works long problems,put forward and the integration of KNN algorithm based on improved threading tube digital character recognition method,the character of different digital tubes and the identification of a decimal point,reduce the dependence on the characters of the pretreatment effect.After the real-time images are collected by LabVIEW human-computer interaction platform embedded in OpenCV image processing program,the recognition results are output.After several experiments,the running time of this method is reduced compared with the recognition time of KNN alone,and the recognition rate can reach more than 95%.It has the advantages of fast recognition speed and high accuracy.

关 键 词：OPENCV 图像处理 KNN 穿线法 LABVIEW 

分 类 号：TN919.8[电子电信—通信与信息系统]