I am working on a project to digitize medical documents, focusing on extracting tables. I have applied a neural network model to detect and crop columns from the tables. After cropping, I preprocess the images with the following steps: