如何用EGMU + C＃读取虚线字符图像？ [英] How to read dotted character image with EGMU + C#?

问题描述

亲爱的，



我想用EGMU + C阅读点缀的人物形象＃



For参考请看下图：



http://www.armstrongltd.com/cms/Media/media/temp/example.jpg



问候，

Nitin

Dear All,

I want to read dotted character image with EGMU + C#

For reference please see the image as below :

http://www.armstrongltd.com/cms/Media/media/temp/example.jpg

Regards,
Nitin

推荐答案

我想你的意思是Emgu。



这是典型的OCR问题。可能的解决方案路径是：



a）对图像进行垂直阴影投影，从而确定字形的水平位置。



b）计算每个字形的封闭框架



c）用合适的阈值对字形框架内容进行二值化



d）将二值化的字形与一组参考图像进行比较，这些参考图像代表字形集中的字形。找到最适合您当前字形的参考字形。验证与参考字形的相似性是否高于某个识别阈值。还要确认参考集中没有其他符号也足够好。



您的示例图像看起来像是用激光写的。这样的图像受到偏斜和可变水平间距的影响。所以你可能不得不通过一些预处理补偿这些扭曲。

I suppose you mean Emgu.

This is a typical OCR problem. A possible solution path is:

a) Do a vertical shadow projection of the image and determine thereby the horizontal positions of the glyphs.

b) Calculate the enclosing frame for each glyph

c) Binarize the glyph frame contents with a suitable threshold

d) Compare the binarized glyph with a set of reference images that represent the glyphs in your glyph set. Find the reference glyph that matches you current glyph best. Verify that the similarity with the reference glyph is above a certain recognition threshold. Verify also that there is no other glyph in the reference set that matches also good enough.

Your example image looks like it was written with a laser. Such images ofter suffer from skewing and variable horizontal pitch. So you will probably have to compensate for these distortions by some pre-processing.

这篇关于如何用EGMU + C＃读取虚线字符图像？的文章就介绍到这了，希望我们推荐的答案对大家有所帮助，也希望大家多多支持IT屋！