基于HALCON的发票号字符识别研究

Character recognition research of invoice numbers based on HALCON

为了解决人工方式识别发票号效率低下、费时费力的问题,以HALC(ON软件为平台,对发票号进行识别,首先,对读取的图像进行图像预处理,包括把原图像转化成单通道图像、图像增强、高斯滤波、阐值分割和连通区域分割,然后对预处理后的图像进行数学形态学处理,接着把粘连的数字拆分开,最后用基于多层感知分类器(MLP)对发票号数字进行识别。通过与人工方式的方法对比,结果表明,用基于HALCON的方法.发票号数字能被快速、精确的识别出来,有效的减少了人员的工作量,提高了工作效率。

In order to solve the problem of low efficiency by manual way of identification,this paper uses HALCON software as a platform to identify the invoice number.Firstly,the image preprocessing is carried out,which includes transforming the original image into single channel image,image enhancement,Gaussian filtering,threshold segmentation and connected region segmentation;Secondly,the mathematical morphology of the pretreated image is processed.And then,the figures of the split;Finally,the number of the invoice number is identified by MLP（multilayer perceptual classifier）.By comparing with the manual method,the results show that the invoice number can be quickly and accurately identified by the method based on HALCON,which can effectively reduce the workload of the staff and improve the efficiency of the work.