I am using tess-two to OCR images in Android. I have pre-processed the input PNG image by deskewing, resizing, binarizing (Bradley threshold), and De-Noising the image. When I run tess-two, the results are pretty poor. When I ran it on command line Tesseract, the message says 0 dpi. The input image is below.
I suspect that each pre-processing step reduces the DPI of the image. How do I boost the DPI in between these image processing calls? Also how do I maintain the tabular structure of the receipt?
