基于机器视觉技术的表单图像缺陷矫正及自动识别方法  

Defect Correction and Automatic Recognition Method of Form Image Based on Machine Vision Technology

作　　者：汤力 梁国迪 王菁 刘跃龙 TANG Li;LIANG Guo-di;WANG Jing;LIU Yue-long(Information Center of Yunnan Power Grid Co.,Ltd.,Kuming 650214 China)

机构地区：[1]云南电网有限责任公司信息中心,云南昆明650214

出　　处：《自动化技术与应用》2024年第10期52-55,138,共5页Techniques of Automation and Applications

基　　金：云南电网有限责任公司信息中心项目(059300HK42210001)。

摘　　要：在实际生活中,大部分表单图像的传输仍旧采用人工输入的方式完成,存在工作量大、核对麻烦的问题,因此提出基于机器视觉技术的表单图像缺陷矫正及自动识别方法。首先通过最小二乘算法实现表单缺陷图像信息的矫正修复,其次采用基于小波变换的维纳滤波算法去除表单图像的噪声,最后采用Gabor特征结合字典学习的稀疏方法完成表单图像的自动识别。实验结果表明,所提方法的图像校正效果好,图像识别效果更好。In real life,most of the form image transmission is done by manual input,which has the problem of heavy workload and trouble-some verification.Therefore,a form image defect correction and automatic recognition method based on machine vision technolo-gy is proposed.Firstly,the form defect image information is corrected and repaired by the least square algorithm.Secondly,the form image noise is removed by the Wiener filtering algorithm based on wavelet transform.Finally,the Gabor feature combined with the dictionary learning sparse method is used to complete the process.Automatic recognition of form images.The experi-mental results show that the proposed method has good image correction effect and better image recognition effect.

关 键 词：机器视觉 表单图像校正 图像识别 最小二乘算法 维纳滤波 字典学习 

分 类 号：TP391.41[自动化与计算机技术—计算机应用技术]