How to improve recognition for the given image?

[TrueWodzu]

Hi guys,

I am testing tesseract (4.0.0), I am a new user of the library. I've read the documentation about improving quality but it did not helpin this particular case.

The tiff image produced by tesseract is in the attachement.

The command I am running is tesseract.exe "155_2.jpg" output.txt --psm 7 -c tessedit_write_images=1

I've tried various psm values, without psm I get nothing, with psm 7 my reading is "195",with psm 8 there is no result.

I appreciate any advice you can give me.

Thanks!

[Ha Duong]

Dear TrueWodzu,

Use file datatrain best for digits

https://github.com/Shreeshrii/tessdata_shreetest/blob/master/digits.traineddata 

And before you set image to OCR by Tesseract, you can pre processing image, your image is low quality.

You can use function Cv2.Erode() with size (5,5) and SetSetResolution(xDpi = 300, yDpi = 300) ( it's best to OCR) then use tesseract to OCR.

- This is image after process:

- Command: tesseract Capture.PNG output --psm 6 -l digits

- Result: 

[TrueWodzu]

Hi,

Thank you for your answer.

I think it works kinda randomly. When I've rescaled image and decreased its resolution it was able to give me a proper read.