Drawing on a blog:
http://my.oschina.net/yushulx/blog/359468#OSC_h3_4
English version is here:
http://www.codepool.biz/making-an-android-ocr-application-with-tesseract.html
I uploaded an example in the android studio environment, you can take pictures or select pictures in the gallery for identification.
This link ( https://github.com/yuzhiyun/TesseractOCR2/ ) includes a project source code and two compressed packages (language packages), one is Chinese characters and the other is English.
caution:
You must now root the phone's internal storage
(Get this string in the program to understand what this directory is Environment.getExternalStorageDirectory())
Create a new folder tessdata, and then unzip the language pack and put it in this directory. If there is no language pack, the program will crash.
The program is not particularly accurate, because it is directly recognized and does not process the image. I will describe how to process the image (sharpening, etc.) to improve the accuracy in a later blog.