How does OCR work? Explained in a simple and understandable way
In this article we explain how OCR works. OCR stands for "Optical Character Recognition" and is simply referred to in German as text recognition. This means that a computer can recognize text on a scanned image using OCR and then convert it into a simple text document.
This is how OCR works
Imagine that you received a paper presentation from a colleague. You now want to edit these on the PC because you do not like individual passages. So scan it and go to your OCR program. Now the following happens:
- The software first does a so-called layout analysis. To do this, she looks at the page layout and separates images from text. She also notes their position on the page. Then the number of paragraphs are counted and individual elements such as page numbers are saved.
- Now comes the hard part. The software looks at the individual text blocks and breaks them down into sentences. The sentences are then split into individual words and the words into letters.
- The OCR software contains patterns of letters and characters. The program now compares the scanned letters with these patterns. If they are 99% similar, the algorithm decides that it probably has to be that letter. It is very precise because it can compare many patterns in a short time. In this way he successfully differentiates between an "8" and a "B".
- So the letters and characters are gradually recognized. Then they are combined again as words and put back in their place in the sentence. As soon as the software is finished, the whole thing is saved in a normal document, which you can then edit. Finished!
Do you want to test your software knowledge? Take part in our big software quiz!