Implementing my own OCR
I wanted to try my hand as part of my fun adventure implementing solutions to everyday things.
OCR is everywhere
I was a bit interested when Apple introduced a way to copy text natively some iOS releases back and I just kept thinking, how they got it to work so quick and easily. From my Applied Machine Learning class, I know there is a ML way of doing feature extraction and such. I wanted to explore how the solution for OCR build up to where it is now.
Recently I wanted to implement a to scan receipts for another project I am not in a rush to finish, so I took Wikipedia. https://en.wikipedia.org/wiki/Optical_character_recognition#Techniques.
Now I am going to document my journey so far. First of I chose Rust, I found the Image crate to be straightforward to comprehend (so far at least). I have so far read some papers on Implementing deskewing To rotate a scanned image to compensate for skewing. (https://en.wiktionary.org/wiki/deskew) as part of the steps for an OCR.