Scratch Card Reader is my attempt in understanding how OCR works/performs on Android (the does and dont's of OCR regardless of approach used). Some of the code in this repo is therefore not of the highest quality and is also undocumented. Feel free to document any piece of code you think needs documentation.
Scratch Card Reader utilizes Tesseract for OCR. You therefore need to setup Tesseract as a seperate android project (on Eclipse) and import that project to this one as a library. Note that Scratch Card Reader is also an eclipse project. You can learn how to setup and how to compile the Tesseract project by reading [this post] (http://gaut.am/making-an-ocr-android-app-using-tesseract)
I used Tesseract 2 with this project. I however noticed that you can use training data generated by any version of Tesseract that is >= version 2
If you are interested in how training on Tesseract is done check [Training Tesseract 2] (https://code.google.com/p/tesseract-ocr/wiki/TrainingTesseract2) and [Training Tesseract 3] (https://code.google.com/p/tesseract-ocr/wiki/TrainingTesseract3) out.
In Scratch Card Reader, the resultant tesseract training file (eng.trainedata) is an android project asset. The file is in assets/tessdata/
I have created another repository that contains all the files used in training tesseract for this project.