TechTech Explained

What is OCR – Optical Character Recognition | Feature Detection | TechBuf

Optical Character Recognition: Do you break your teeth by reading your friend’s hand writing? Still, you should feel lucky, because you do not work in any post office. Where thousands of hand-written letters are transmitted every day. Today, our lives are gradually becoming computer-based, so there is a need to have a good understanding of people and computers. We use “keyboard” or “mouse” to talk to almost every type of computer so that they can easily understand what we want them to do. But when a computer is asked to read a book directly or to read a letter written in hand, then the computer will not have to do more to complete it. Here, optical character recognition (OCR) comes to work. It is a computer program or software that can analyze any printed text or handwriting automatically and takes it in a form so that the computer can easily understand it. The handwriting keyboard that you have on your cell phone also works using OCR. But how does it work? Let’s know the details. You may like what is IP address?

What is OCR Optical Character Recognition?

Optical Character Recognition


When you read these texts on a computer screen or the mobile screen, your eyes and your brain are continuously using optical character recognition methods, even if you are not sure about the point. Your eyes are trying to detect different characters such as letters, numbers, and punctuation by using the white, black pattern on this screen, and your brain tries to read the text together with those letters, numbers, and punctuation marks.

Some article you may like

Humans can do this, but computers can not do so easily. The first problem is that the computer does not have any eyes, so if you want to read an old book or your handwriting with a computer, then scan it with an optical scanner or digital camera and display it as an image. The page you create in this method will be set up as a graphic file (usually a JPEG file). But the computer does not have any difference between your picture or the picture of the Sundarbans, that is, the computer can not find any meaning from a pixel pattern of a graphic or a photograph. The computer has no value for having a text-filled page because you can read any page, and the computer can not read it like that. The OCR program changes its text from any image as a direct text-or it can save text from a picture or handwriting as a TXT or DOC file.

How does OCR work?

Let’s think our life is effortless, and our alphabet contains only one letter “A.” Now if you think about it, you will see that OCR will face many problems even after just one character-every person will write this “A” completely differently. Even if you talk about computer-printed text, there are problems too – why are different books and documents written using different printers (fonts), so separately write this letter “A.”

How can OCR recognize the writing? Broadly speaking, there are two ways to work. One is to identify each letter in its possible pattern (pattern recognition), and another is to recognize each letter in each letter, remembering that character with every fine line (feature detection).

Pattern Recognition

If everybody writes a letter like “A” in the same way, the computer will become much easier to recognize it. Then only the computer will scan “A” with its stored “A,” if both are matched, then work is done.

But now the question is, how do you teach the same handwriting to all? In the 1960s, promoting a single phone for writing Blank Check, whose name is “OCR-A.” Then the same font was used in every test, so optical character recognition or OSR could easily understand the font. But currently the same font is not used to print books or documents throughout the world, and never write a handwriting on anyone. The way to s of this problem is to specially train the OCR program to identify each letter so that using an application number to recognize a message. The program shows the number of favorite fonts in the number of a number and decoding these figures when scanning and trying to understand the characters. There is no guarantee that your system can recognize 100 percent of the characters.

Feature Detection

Feature detection is called Intelligent Character Recognition or ICR. It is the most intelligent way to identify a character. Now, assuming you own OCR computer program, you have put together many letters in front of you, where different letters are written using different fonts. How can you recognize these characters? They are all written in various forms!

You can find an excellent rule to make your work easier. When you see two curved lines, and you see that another line is attached to the two lines at the top and tilt the bend line in two halves and another line is leveled together at the top or middle, then you will quickly understand that this character is ” A .” And by using this method, you can recognize uppercase letters, no matter which font is written using a font. Without thinking of a complete pattern of a character directly, each line of that axis is possible to recognize any subtle lines and their position, even if it is easy to identify. Most modern OCR programs use feature detection methods more than pattern recognition methods. Some programs work using neural networks (this is the method, where software can automatically work like a human brain). We will soon write a new post about the neural network.

How Hand Writing Recognition Works?

Hand writing

The computer print with laser printers and it is relatively easy to have a computer reconnect. But when the computer wants to read the handwriting of people’s writing, then 12 times the computer. The human brain works long ago from any computer. In this case, but we can not fully understand the worst handwritten text in the world. But we can assume that we try to identify any character using the feature detection method. That there is a bright idea, about what the writer and what’s written in our brain. But the computer’s intelligence can never work with knowledge.

Some article you may like

When you write something with a tablet computer or smartphone’s handwriting keyboard. It is much easier to identify it to the appliance. If you try to write “A,” then touchscreen feels your drawn line. First, you draw a long curved line, then another long curved line and then ending with another plane line. Here, the computer can understand how a line was drawn and after drawing it would be displayed in the letter “A.” The computer uses feature detection methods. But it is not so easy to recognize any Abrothbro handwriting on any paper. To write a script to a computer, you have to write only after complying with a particular rule, and the rules of this writing are called comic field.

Facilities of OCR

If a printed page is made possible by machine readable. You can do all the things that you could never do before. You can search for any text or character using keywords, edit the text from the printed pages by word processing software, and you can make text into a web page. Compress the entire page text and save space. You can also send mail using text from the print page. The machine can read and understand any screen reader easily.

The last word

You may think that the OCR is no technology to be discovered today! This technology and its working method are not new. This technology has been coming out since 1928. But maybe it possible to improve further in the future. If you do not read my written post about machine learning and artificial intelligence. You can read it, find much interesting information there. Hopefully, today’s post looks great to you, so please comment on any questions and observations below. Help your friends illuminate the light by sharing the post.  You can read . Thank you.

Shushanta Kumar

Founder and CEO of TechBuf blog, also a regular TechBuf contributor. I'm a pretty skilled writer. Try to increase Tech related content.

2 thoughts on “What is OCR – Optical Character Recognition | Feature Detection | TechBuf

Leave a Reply

Your email address will not be published. Required fields are marked *