Updated the preprocessImage to add different types of filters for bet…

[Wilsonilo]

I found better results using this filters.
I found the previous approach was causing some of the letters to disappear, after this adjustments 3 lines and 2 lines passports / ids started to work.

[Mattijah]

Do you happen to have an equally taken images to showcase the difference? Also, it seem the traineddata has been changed too? It's been a while since I did this, but the current solution was able to adjust the image dynamically based on its luminance (if taken in a darker environment) and also it was optimised well since it was running on iOS 11+ devices, but that shouldn't be an issue anymore, I guess.

[Wilsonilo]

This is an example. I'm not sure how the traineddata changed, I think I only used cocoapods, installed and unlocked to overwrite that specific method. Currently using iPhone XS on 17.4.1, without the adjustment i was getting really hard luminance contrast before the tesseract.performOCR. Is ok if you test it and still works fine. Maybe just leaving my input just in case someone else find this problem on their end? Also let me know if this pull request method of forking, update, etc is correct, trying to really follow good practices on pull request / helping on open source. Thanks!