1

I am trying to recognise numbers in a Sudoku board. I'm using a sample printed in Times new Roman, and I have trained the system using that font, and limited the characters to numbers by setting recognizableCharacters = "0123456789 testImages.zip

For my sample set of test images, which are all about 177 x 184 pixels, the only number that is ever recognised is '1'. Everything else is unrecognised.

I have converted all of the images to black & white only to increase contrast, but I was getting the same result with natural colours.

Here are some of the images I'm trying to recognise

sample image - 4

sample image - 2

I know that there will be suggestions that I should use Tesseract - but I'm trying to get this one working as it should be faster.

There must be something simple that I'm missing...

Russell
  • 5,436
  • 2
  • 19
  • 27
  • "SwiftOCR is optimized for recognizing short, one line long alphanumeric codes..." - One line only. VIA https://github.com/garnele007/SwiftOCR/blob/master/README.md#swiftocr – OhadM Jul 31 '17 at 12:30
  • The images I'm working with are one character only. I'm finding edges and parsing the sudoku board to generate 81 individual images which I then try to identify. But thanks for the link to the original source, which is - of course - where I started – Russell Jul 31 '17 at 13:11
  • If you want, you can get the Tesseract and train it for your data/font but I am afraid it might take a while.. – OhadM Jul 31 '17 at 13:27
  • If you are still interested try this library that I wrote, it uses Tesseract but it is quite fast: https://github.com/robertoferraz/OCRSlicer – Roberto Ferraz Jun 29 '18 at 07:07

0 Answers0