Method for Character Domain Extraction from Image Using Wavelet Transform

被引：4

作者：

Taniguchi, Taiki ^{[1
]}

Yoshitomi, Yasunari ^{[2
]}

机构：

[1] Zensho Holdings Co Ltd, Minato Ku, 2-18-1 Konan, Tokyo 1080075, Japan

[2] Kyoto Prefectural Univ, Grad Sch Life & Environm Sci, Sakyo Ku, Shimogamo, Kyoto 6068522, Japan

来源：

JOURNAL OF ROBOTICS NETWORKING AND ARTIFICIAL LIFE | 2015年 / 2卷 / 02期

关键词：

Character domain extraction; Image processing; Internet; Private information; URL of illegal Web site; Wavelet transform;

D O I：

10.2991/jrnal.2015.2.2.8

中图分类号：

TP24 [机器人技术];

学科分类号：

080202 ; 1405 ;

摘要：

There has been an increase in the number of images on the Internet that contain private information and/or the URL of an illegal Web site. This information could result in a violation of human rights and/or a crime. We present a method for extracting from an image the region(s) that contain characters. Our method uses the discrete wavelet transform and the empirical knowledge that characters have strong vertical and/or horizontal elements. The experimental results show the usefulness of our method.

引用

页码：103 / 106

页数：4