Recognition of previously known text. What methods exist?

Question

Pattern recognition

There is a task of determining pre-known words and symbols in the photo. As I understand it, the "usual" recognition algorithm in short and in almost ideal conditions looks like the following steps:
1. Making the photo black and white. Remove noise as much as possible.
2. Find the borders where the text is present
3. Cut out this part with the text. Normalize (i.e. if the text is at an angle, we make it horizontal)
4. Find the boundaries of the letters, divide the image into letters, and then the problem turns into a letter recognition problem. Let's say that this problem is not solvable with the help of neural networks.
Is it possible to simplify the process when the words / phrases / symbols that need to be detected are known in advance?
For example, a list of predefined words and phrases is defined as A = {discount, discount, gift, 1+1, 50% discount, 25%, 10%, $}, where n = |A| for 0

Answer 1

Instead of it is necessary to directly load resources from another's sites. Download this font for yourself and give it away from your site.