图片文字干扰的快速自动消除  

Fast and automatic elimination of text interference in images

作　　者：董剑龙 王浩全[2] DONG Jianlong;WANG Haoquan(Academic Affairs Office,Taiyuan Institute of Technology,Taiyuan 030008,China;Key Laboratory of Instrumentation Science & Dynamic Measurement,Ministry of Education,North University of China,Taiyuan 030051,China)

机构地区：[1]太原工业学院教务处,山西太原030008 [2]中北大学仪器科学与动态测试教育部重点实验室&电子测试技术重点实验室,山西太原030051

出　　处：《电视技术》2019年第4期11-13,24,共4页Video Engineering

基　　金：山西省回国留学人员科研资助项目(No.2017-089);山西省自然科学基金项目(No.201801D121160)

摘　　要：通过将边缘信息加入到传统TV模型中,改善扩散系数的收敛性,提出一种快速自动消除文字的改进TV模型。采用LoG算子和数学形态学对待处理图片中干扰文字区域进行检测与准确定位,将已确定的干扰文字区域作为改进TV模型的约束条件并进行求解即可有效消除文字的干扰。仿真结果表明,该方法对非纹理图片中干扰文字的检测和消除快速实用。By adding edge information to the traditional TV model, the convergence of the diffusion coefficient is improved, and an improved TV model for quickly and automatically eliminating text interference is proposed. LoG operator and mathematical morphology are used to detect and accurately locate the interference text region in the image to be processed. By taking the determined interferential text region as the constraint condition of improving TV model and solving it, the interferential text can be effectively eliminated. The simulation results show that this method is fast and practical for detecting and eliminating interference texts in non-texture images.

关 键 词：文字干扰 文字消除 TV模型 数学形态学 

分 类 号：TP391.4[自动化与计算机技术—计算机应用技术]