基于最大稳定极值区域的自然场景文本检测摘要： 近些年来，科技发展迅速，相机、手机等拍摄设备应用变得十分广泛。人们可以从相片中获取很多信息，相片中的文本信息包含很大的商业价值及应用价值，比如马路上的红绿灯会拍摄到很多违反交通规则的车辆，可以利用文本检测技术识别出车辆的车牌号，从而避免了耗费人力去观察这些车辆，所以，近年来国内外文本检测技术越来越多。本文提出用最大稳定极值区域的方法对自然场景中文字部分进行检测，以便于后续工作的开展，实验证明，该方法有效可行。关键词：自然场景，MSER，文本检测ABSTRACT：In recent years, science and technology have developed rapidly, and cameras, mobile phones and other shooting equipment have become widely used. People can get a lot of information from photos. Text information in photos contains great commercial value and application value. For example, traffic lights on the road can capture many vehicles that violate traffic rules. Text detection technology can be used to identify the license plate number of vehicles, thus avoiding wasting manpower to observe these vehicles. Therefore, text detection technology has become more and more popular at home and abroad in recent years. In this paper, the method of maximum stable extremum region is proposed to detect the text part of natural scene, so as to facilitate the follow-up work. The experiment proves that the method is effective and feasible.Key words: natural scene, MSER, text detection1目录第一章 绪论....................................................................................................................................41.1 自然场景文本检测的意义.................................................................................................41.2 国内外研究现状.................................................................................................................41.3 常用文本检测方法................................................
