c# - Image OCR - Filtering unwanted data -
basically i'm reading vehicle license plates using tessract ocr, despite being able emphasise text enough via changing contrast, reducing noise on, 'parts' of vehicle remain on image cause ocr throw bad results.
for example take:
i can change enough, such as:
i'm looking eliminate edges off each plate, here's example:
i remove edges using pixel manipulation algorithm, don't feel it's right method, , cause quite lot of problems.
i've been using following application test various methods such morphology , eliminating unwanted data, far haven't been successful.
http://www.codeproject.com/kb/gdi-plus/image_processing_lab.aspx
however knowledge of use application on article above achieve want i'm trying, feel free give try.
thanks
please try use stroke width transformation concept.
this concept use segment text natural images.....
Comments
Post a Comment