I’m trying to train an OCR model, but I need to get my extracted images to look more like the training images. However I’m not sure of how to go about it and what techniques I can use. For instance the images that I extracted are tightly cropped, so I would like to pad them, but the background in the training image is full black whereas in my extracted image it is light black, so I cannot pad it with black directly. I need to make the brighter pixels thicker and cover more area as well.
Training image - For reference, this is an example of the image used for testing data.
And this is an example of the image I extracted -