I am trying to get an image of the OCR readable blackboard. Naturally, most OCR software does not like dirty images. What image processing should I try to clean the image?
Have you tried OCR software yet? It is likely that OCR software is actually suited to read a black and white image already.
However, if you need to do this, you can try the following:
- Threshold image image
- Originally a grayscale image Take the version and turn it into black / white pixels
- Make binary deletion to increase the remaining objects
- Performance binary erosion < / Ol>
- Algorithms for image processing and computer vision - JR Parker
- Digital Image Processing - Gonzalez / Woods < / Li>
Then by eroding you will remove any thick / noisy edges and then you pass the odd image to the OCR can do.
There are probably so many ways to get the same result, given that computer has been completely devoted to the books in view, the answer will hardly be judged by them.
I have only one book that is from 1997, but definitely this has been written more on this topic.
Comments
Post a Comment