一种医疗票据图像数据结构化方法及系统

Data structuring method and system for medical bill image

提出一种医疗票据图像数据结构化方法,针对医疗票据图像,设计了红章去除、文字定位、文字识别和费用明细项目数据结构化等方法。在此基础上,设计了一套医保零星报销系统,实现医疗票据从图像扫描、OCR识别、医保目录匹配到数据入库自动化操作。经多个省市使用,累计处理10万份票据,使用该系统后报销票据每份处理时间由原先的2小时缩短到15分钟。

In this paper,we propose a medical bill image data structuring method,which is designed for medical bill images,including red stamp removal,character positioning,character recognition and expense item data structuring.Based on this,a medical insurance sporadic reimbursement system is designed to realize automated operation of medical bills such as image scanning,OCR recognition,medical insurance catalog matching and data warehousing.It has been used in many provinces and cities,and 100,000 bills have been processed in total.The processing time of each reimbursement bill has been shortened from 2 hours to 15 minutes after using this system.