Hello
From what I have understood about the project, there are multiple ways this can be done depending upon quality of the text in the image and the font of the text.
We can load the image using openCV, crop the area and pass it to some OCR engine to get text from it.
We can also use some image processing algorithm instead of machine learning if the patterns are quite fixed. (fonts and sizes do not vary too much).
And we can also train a machine learning algorithm by ourselves if we have enough of those kind of pictures.
Databases and rest is regular stuff and can be done easily.
Feel free to reach me anytime and we can discuss the path to go.
Thanks Man