基于OpenCV形态学的发票定位研究  

Invoice Location Research Based on OpenCV Morphology

作　　者：谢阳 程艳云[1] XIE Yang;CHENG Yanyun(Automatic&AI College,Nanjing Uuniversity of Posts and Telecommunications,Nanjing 210000)

机构地区：[1]南京邮电大学人工智能学院、自动化学院,南京210000

出　　处：《计算机与数字工程》2021年第4期809-812,817,共5页Computer & Digital Engineering

摘　　要：为了提高发票识别的准确率,对票据定位算法进行了研究。现提出一种基于OpenCV形态学的发票定位算法。该算法利用电子发票的表格线规整性,利用形态学操作检测垂直和水平等高线并画出发票表格框架线,再利用右下角基准点及各区域框的长宽定位出发票需要识别的区域。论文算法用Python及OpenCV等库实现,实验结果表明:该算法可以准确定位并裁剪出待识别区域。In order to improve the accuracy of invoice recognition,bill location algorithm is studied.This paper presents an in⁃voice location algorithm based on contour detection.The algorithm utilizes the regularity of the form line of electronic invoice,draws the frame line of the invoice form by using the vertical and horizontal line detection,and locates the area to be identified by using the benchmark point in the lower right corner and the length and width of each area frame.The experimental results show that the al⁃gorithm can locate and tailor the region to be recognized accurately.

关 键 词：发票 定位 等高线 表格框架 形态学 

分 类 号：TP3-05[自动化与计算机技术—计算机科学与技术]