OCR (Arabic & English)

OCR is a technology that allows computers to recognize text in physical documents and convert it into data. When we read text on a page, whether it be on paper or on a computer screen, we immediately recognize the letter or other symbols. Computers, on the other hand, are a little more complex. Certain applications utilize optical character recognition (OCR) to allow you to edit the text from a scanned document in just the same way which you would in a word processor. You can use text to highlight it, copy it to other documents or rewrite entire sections. Another application of OCR is to enable full-text searching. Some OCR programs will add the text recognized from a scanned document as metadata to the file, allowing certain programs to search for the document using any text contained within the document.

All OCR systems are slightly different, but they all work by identifying text character by character, word by word, and line by line in the picture of each page. OCR is simply a binary process that recognizes either present or absent objects. If the original scanned picture is flawless, any black in it will be part of a character that must be identified, while any white will be part of the backdrop. As a result, the first step in determining the text that has to be processed is to convert the image to black and white.

OCR software applications work differently depending on the intended purpose but still follow several common principles. No scanner is perfect, so with most modern, commercial scanners, there are bound to be imperfections in the scanned image. The software typically has a preprocessing phase that attempts to make the text in the document clearer and easier to read. It does this by cleaning up the image and isolating the characters from everything else. It makes sure the lines of text are properly aligned and the pixels are smoothed out.

The programme then isolates each individual character, identifying the pixels that make up the characters and the gaps between them. This enables the software to analyse each individual character as well as identify that a word is made up of a cluster of characters.

The following stage is the most difficult, and it is frequently the one that distinguishes different OCR systems. Once the OCR software understands what defines a character that it must recognise, it must determine which character it is in order to give the appropriate metadata. Simple OCR software compares the characters to a library of common fonts to see whether they match, and then the data can be allocated. However, for a text that doesn’t match any recognizable fonts in a library, such as uncommon fonts or handwritten text, more sophisticated techniques are required.

More sophisticated OCR algorithms will keep comparing characters to common patterns to figure out which one they are. They’ll recognize the letter “A” as two diagonal lines with a line in the center. Contextual information will be used by the most sophisticated OCR to discern what letters and words are what. If it can’t figure out whether a character is an “I” or a “1,” it looks at the nearby characters it recognizes and makes an educated guess. The following line is more likely to be interpreted as “Invoice to be delivered,” rather than “1nvoice to be delivered.”