Selecting Pre-Processing Method

[rohit raj]

I have to run tesseract ocr on tens of thousands of snippets of videos having varying characteristics. 

I have been reading up on pre-processing methods for tesseract. It seems to various pre-processing method works best in separate cases. 

Apart from converting image into grayscale which helps in all cases. I hope.

How to programmatically chose pre-processing methods in other cases.