Hello all!
I wonder if this is the right approach: I want to do OCR of my dashboard cam videos, to extract sign and car plates in text form.
So, using avconv to convert mkv to png or any other preferred format works fine, but then my imagination runs dry. Since the signs can be distorted
in many ways (fisheye lens, signs and plates not immediately in front of camera, difference in light/shading).
Example images from video: imgur 14 images
My idea was to first remove anything with non-black, non-white and non-gray and then try different rotation angles until the ocr program either picked up the sign/plate text.
I have though no clue where to start or if it is even worth the effort doing it this way (avconv -> imagemagick -> ocr). Anyone got a better suggestion?
Great forum, btw!
Edit: OpenALPR does a good job, but fails too often. I am still tweaking it to see if it performs better.