基于部分卷积的文字图像不规则干扰修复算法研究  

An irregular interference repair algorithm of text images based on partial convolution

作　　者：段荧 龙华[1,2] 瞿于荃 邵玉斌[1,2] 杜庆治[1,2] DUAN Ying;LONG Hua;QU Yu-quan;SHAO Yu-bin;DU Qing-zhi(Faculty of Information Engineering and Automation,Kunming University of Science and Technology,Kunming 650504;National Key Laboratory of Computer Science of Yunnan Province,Kunming University of Science and Technology,Kunming 650504,China)

机构地区：[1]昆明理工大学信息工程与自动化学院,云南昆明650504 [2]昆明理工大学云南省计算机重点实验室,云南昆明650504

出　　处：《计算机工程与科学》2021年第9期1634-1644,共11页Computer Engineering & Science

基　　金：国家自然科学基金(61761025)。

摘　　要：针对文字图像中不规则干扰与文字粘连导致误识字的问题,提出了基于部分卷积的文字图像不规则干扰修复算法。研究分析了若干常见字体的文本图像特点,建立文字图像数据库,使其与干扰掩码数据库进行图像融合后对模型的修复效果进行评测,并对不同等级的修复情况进行分类测试。实验表明,所提模型在保证原有文字信息不损失的前提下,根据当前文字的现有部件对缺失部分进行预测,峰值信噪比最高达到32.46 dB,结构相似性最高为0.954,最佳损失值达到0.015,修复前后文字识别率提升27.85%,对隶书、篆书、甲骨文、行书4种古代文字缺损图像进行修复后峰值信噪比最高达到30.46 dB,结构相似性最高为0.964。Aiming at the erroneous literacy caused by irregular interference and text adhesion in text images,this paper proposes a text image restoration model based on partial convolution operations.This paper studies and analyzes the text image characteristics of several common fonts,establishes a text image database,and integrates it with the interference mask database,then evaluates the repair effect of the model,and conducts classification tests on different levels of repair.Experiments have proved that the text model predicts the missing parts based on the existing parts of the current text under the premise of ensuring that the original text information is not lost.The peak signal-to-noise ratio is up to 32.46 dB,the structural similarity is up to 0.954,and the best loss value is up to 0.015.The text recognition rate after repair is 27.85%higher than that before repair.After repairing the defective pictures of four ancient scripts,including official script,seal script,oracle bone inscriptions,and running script,the peak signal-to-noise ratio reached 30.46 dB,and the structural similarity reached 0.964.

关 键 词：文字图像修复 部分卷积 光学字符识别 深度学习 

分 类 号：TP391.41[自动化与计算机技术—计算机应用技术]