0

I am currently working on license plate recognition.For the same purpose I have trained the tesseract with extracted and preprocessed images of alphanumeric characters.

In letters of 0,Q,D and B,8, ambiguity is occurring.I have attached the images of 0,Q,D.Kindly help me on what to do.I have already used unicharambigs but of no use.Is it because I have lesser training images for Q?

Shweta Sharma
  • 5
  • 4

1 Answers1

0

You can do some image-processing before sending image to Tesseract. You can check this link:

image processing to improve tesseract OCR accuracy

Also you can check Tesseract-Ocr GitHub and their Improve Quality document:

https://github.com/tesseract-ocr/tesseract/wiki/ImproveQuality

Lesser training can be a reason, but you can make great difference with image-processing too.

Also, you can check other license plate recognition questions:

Using tesseract to recognize license plates

Community
  • 1
  • 1
bodoroman
  • 278
  • 2
  • 16