OCR with tesseract, pre-processing image

Question

I need to extract digits from images like the one shown below, I'm using tesseract now, but it isn't working. Can anyone help me in pre-processing the images before feeding it to tesseract?

Answer 1

I don't think tesseract is the right tool for it, Tesseract can only handle very clear letters.
If your numbers are all like those in the picture you can use opencv ORB detector https://opencv-python-tutroals.readthedocs.io/en/latest/py_tutorials/py_feature2d/py_orb/py_orb.html
Or if it don't work, you can use some deeplearning aproch, as a SSD Keras or YOLO.
https://github.com/pierluigiferrari/ssd_keras
Another option is to dismember the numbers (it is easy if is all the same size) and create a very simple convolutional neural network with keras.
https://keras.io/