摘要
如今图像成为重要的信息载体,图像中含有大量有价值内容。文字作为图像的重要内容蕴含了大量的信息,并且文字能够对于自然场景的定位识别提供重要线索。本文简述了现在使用广泛的OCR系统,并且依照不同的文字特征,介绍了三类自然场景文字定位的方法:基于纹理特征方法、基于连通域分析方法和基于边缘特征方法。目前,国内外大量的研究机构和人员力求开发出高鲁棒性、高召回率的文字识别定位系统。假如可以实现对这些文字信息的自动定位与识别,为人们生活提供极大的便利。
Nowadays,image has become an important information carrier,and there are a lot of valuable content in the image. Text as an important content of the image contains a lot of information,and the text can provide important clues for the location and identification of natural scenes. This paper briefly describes the widely used OCR system,and introduces three kinds of methods of text localization based on different characters,such as texture feature,connected domain analysis and edge feature. At present,a large number of research institutions and personnel at home and abroad strive to develop a high robust and high recall character recognition and positioning system. If we can realize the automatic positioning and identification of these text information,it can provide great convenience for people's life.
作者
季昊龙
Ji Haolong(Criminal Investigation Police University of China Technical,Department of Sound and Image Data Inspection,Shenyang 110000,China)
出处
《山东化工》
CAS
2018年第11期59-61,64,共4页
Shandong Chemical Industry
关键词
图像
文字
自然场景
特征
自动定位
鲁棒性
召回率
image
text
natural scene
feature
automatic positioning
robustness
recall rate