从机器人拍摄的图像中提取信息机器人、图像、信息

2023-09-05 08:02:56 作者:怀抱拥久

这是我的形象:

我用这个链接(tessaract)捕获并处理图像: http://kurup87.blogspot.com/2012 /03/android-ocr-tutorial-image-to-text.html

I used this link(tessaract) to capture and process the image: http://kurup87.blogspot.com/2012/03/android-ocr-tutorial-image-to-text.html

但是,这是问题,如果这整个区域进行扫描,返回值是一些垃圾值,而不是精确的。但是,如果我扫描V516990,2653,并单独的日期。结果是正确的。

But this is the issue, if this entire area is scanned, the return values are some garbage values, not accurate. But if I scan V516990, 2653, and the date separately. results are correct.

我的目的是扫描V516990和2653一气呵成,而无需使用相机的两倍用户。任何意见,欢迎!

My intention is to scan V516990 and 2653 in one go, without the user having to use the camera twice. Any comments are welcome!

推荐答案

让用户只需要一个图像 你可以将其存储在内存中,并设置感兴趣区域的顶部,然后将其解压缩 使用cvSetImageROI

Let the user take one image only You can store it in memory and set region of interest to top portion and then extract it Use cvSetImageROI

https://github.com/mintuhouse/FinMan/ BLOB /主/ src目录/ UNIX / imp.cpp

查线337,当你通过的IplImage到的tesseract 设置你的图像感兴趣区域(顶部或你的情况下部分) 检查297这里

Check line 337, when you pass IplImage to tesseract set your region (top or bottom parts in your case) of interest on image Check 297 here

https://github.com/mintuhouse/FinMan/ BLOB /主/ src目录/ UNIX / main.cpp的

这个功能确实正文提取出来的情况下 使用JNI Java本地接口

This function does main text extraction in out case use JNI java native interface

甚至更好地利用javacv 在Java中OpenCV的端口 我们后来意识到这点。它让生活变得简单许多。

or even better use javacv OpenCV port in Java we realized it later. It make life a lot simpler