从图像中提取文字

Extracting Texts from Images

提出一种从图像中检测目标的思路 ,并利用它提出了从图像中提取文字的新方法。该方法可以用于图像数据库内容检索 ,视频检索 ,以及将纸张出版物中的文本转化为电子出版物。新方法分为如下两个基本步骤 ,首先检测出可能存在文字的候选区 ;其次 ,利用Hough变换方法分析候选内扫描线灰度分布对候选区进行证实 ,并对候选区的形状进行修剪 ,使其尽可能紧凑地包含文字。该方法适用于检测出不同大小、字体、排列方向、灰度 (颜色 )的文字。由于该方法所有操作都是基于像元进行的 ,所以便于硬件并行实现。

A new way for detecting the targets through images is proposed, and then using it, a new method for extracting texts is proposed This method can be used not only to index the image database and the video, but also to invert the documents in papers to electrical text format It can be divided into two steps. In the first step, the candidate regions, which can include texts, are extracted, in the second step the candidate regions are verified by analyzing the distribution of gray value of the scan lines by Hough transform method The shape of the extracted regions is also pruned to contain texts compactly This new method can be used to extract texts with different sizes, orientations, fonts and gray values or colors Because all operations of the new method are based on pixels, the new method can be realized by hardware easily The experiments we have done justify its validity