摘要
传统水准测量数据处理作业中,采用人工方式读取标尺检定证书信息,一定程度制约了水准测量数据处理实现全流程自动化。本文基于Paddle OCR模型识别标尺检定证书图片中的文字信息,对块状识别结果进行分行整理,经过数据清洗、联系上下文、信息互补校验等操作后获取关键信息,实现了水准标尺检定信息的自动化提取。经验证,该识别方法精度达到98.7%,可以承担工程应用中对算者的角色,为实现人工智能技术与测绘地理信息专业的交叉融合提供技术参考,也为未来开发智能化无人水准测量数据处理软件提供技术储备。
In the traditional leveling data processing operation,the information of leveling staff verification certificate is read manually,which restricts the realization of full process automation of leveling data processing.This paper recognizes the text information in the picture of leveling staff verification certificate bases on Paddle OCR model,arranges the block recognition results in rows,extracts the key information after data cleaning,context contract and information complementary verification,and realizes a picture text automatic recognition method for a large number of leveling staff verification certificates.The accuracy of the recognition method is 98.7%.It can replace the role of calculator in engineering application,and provide technical reference for realizing the cross integration of artificial intelligence technology and surveying and mapping geographic information specialty,and provide technical reserves for the future development of intelligent unmanned leveling data processing software as well.
作者
齐文
刘晓云
王建文
马润霞
刘和平
QI Wen;LIU Xiaoyun;WANG Jianwen;MA Runxia;LIU Heping(Geodetic Data Processing Center of the Ministry of Natural Resources,Xi'an 710054,China)
出处
《测绘与空间地理信息》
2024年第9期188-191,196,共5页
Geomatics & Spatial Information Technology
关键词
水准标尺检定证书
Paddle
OCR
图片文字识别
人工智能
verification certificate of leveling staff
Paddle OCR
picture and text recognition
artificial intelligence