5 Cool Things Made Possible By OCR Software

 Understanding OCR Technology  No Responses »
Nov 082011
 

Sure, we can use OCR software to make a digital archive of our documents, notes, and even old love letters, if so inclined. However, optical character recognition can also be used for loftier goals. Here are just a few examples of what the technology has enabled us to do so far.

library at your fingertips1. A Universal Library at Your Fingertips

Who doesn’t love books? Without them, we’d be missing out on interesting stories and lots of useful information. The problem is that books are fragile and wither away with age. We’ve already lost countless precious works in libraries due to wars, fires, floods, and earthquakes. It’s a shame, really, but thanks to OCR software, every book can now be preserved in electronic form and made accessible to anyone with Internet access. This is exactly what is being undertaken by Google Books and Project Gutenberg, among others. Together, they have scanned and digitized millions of books in just a few years, with much more on the way. Time to update your reading list!

2. Crime Prevention and Investigation

OCR software has found a place in the field of surveillance, particularly in vehicle identification. Strategically placed cameras zero in on license plates and attempt to read them for things like toll collection, traffic violations, and stolen car recognition. Since vehicle plates tend to vary from country to country, systems are typically modified to suit the existing format. The cameras may be mounted on mobile platforms like roving car patrols to cover more ground.

3. Uber-Fast Forms Processing

Banks deal with a ton of paperwork day in and day out. Manually checking all of them would be a headache for both the bank workers and the impatient customers (that is, me and you). OCR software makes things easier for everyone by automating things like check processing. A machine can read the written amount and facilitate the transaction in a snap. The same goes for other industries like healthcare, wherein patients records are captured through OCR software and the data continually fed into databases. When insurance issues come up, all the pertinent data are on hand.

find cases

OCR technology helps professionals avoid being frustrated by not finding what they want.

4. Research Made Easy

The typical lawyer’s office has walls covered with books and other documents. Sure, some of that is for show, but they really do need to wade through piles of cases, laws, motions, and decisions to do their job. With OCR software, legal research is now more efficient and less like looking for a needle in a haystack. Digitized texts are easily searchable so what used to take hours or even days can now be completed in a few minutes. That leaves more time for coffee breaks, which is always a good thing.

5. From Print to Sound

Don’t think that OCR software technology is limited to letters and numbers. They can also be programmed to detect symbols in different configurations and interpret them depending on the context. For instance, it’s possible for some software packages like Audiveris and SmartScore to convert scanned pages of a music sheet into an electronic format. This can then be edited or played note for note by the computer, with the sound file saved for future use. Quite nifty for following along during piano practice or for discovering an obscure piece should sound like. Now that’s music to my ears!

Posted by doups3 at 4:39 pm  Tagged with: ,

Best Free OCR Software Packages

 Understanding OCR Technology  No Responses »
Nov 082011
 

Are you itching to see how OCR software works? Do you want to try digitizing a few of your documents? Great news for the curious but noncommittal: it’s possible to do all these without spending a dime. Free OCR software packages typically have limited feature sets but they do get the job done. For casual users, that’s the bottom line. Here are a few examples to get you started.

1. FreeOCR

free OCR

FreeOCR is like a plain-looking car with a monster engine. Indeed, its claim to fame is that it uses the Tesseract optical character recognition software underneath the hood. Tesseract is a project sponsored by Google that has an uncanny ability to produce extremely accurate results even in less than ideal conditions. Since it uses a barebones command line interface (think DOS) in its original form, other developers have seen fit to blanket it in a more user-friendly GUI like FreeOCR.

The program’s main window is divided into two panels. On the left is the staging area for the digital image and on the right is where the processed text is shown. Note that it outputs plain text with no formatting. In addition to English, users can install language packs for Brazilian Portuguese, Fraktur, Dutch, Spanish, German, Italian, Vietnamese, and French. FreeOCR accepts direct scans, image files, and PDFs for input. Installation requires Microsoft’s .NET Framework 2.0 or higher.

2. JOCR or GOCR

gocr

JOCR works by letting you select what you see on your computer screen and converting it almost instantly into text. Pick one of the capture modes from the left side then wait for your chosen image block to appear on the staging area. Once satisfied with the selection, click the “Recognize” button to begin the process. You’ll find that JOCR works well with protected PDFs, error messages, and other hard-to-pin-down content.

The program supports multiple Western and Asian languages. It requires Microsoft Office Document Imaging (MODI) which comes with MS Office 2003 and up to function properly.

3. SimpleOCR

simple ocr

Ironically, SimpleOCR is the most complex of the bunch. In addition to basic optical character recognition, it allows the user to mark off parts of the page where pictures or extraneous details lie so that it can ignore them. Doing this avoids spurious results in the form of garbled text. It also provides a proofreading screen wherein unrecognized words by the internal dictionary are highlighted, making it easier for the user to correct them one by one.

Another great thing about SimpleOCR is that it allows decent conversion of handwritten text through its SoftWriting recognition engine. Each user has to create his own profile and train the program to learn his style of penmanship. Unfortunately, this feature is free for only 14 days, though the machine-print recognition is unlimited. SoftWriting works with English by default with other languages like Italian, French, German, Spanish, Dutch, and Portuguese available as add-ons.

Remember that these are just some of the freely available OCR software packages and there’s a lot more out there. Hey, you’ve got nothing to lose so why not try them all? Keep in mind that for the best OCR software you’ll probably have to pay.

Posted by doups3 at 4:37 pm  Tagged with: , , ,

How Does OCR Software Work? The Technology Behind Optical Character Recognition

 Understanding OCR Technology  No Responses »
Nov 082011
 

how does ocr software work

There’s a staggering amount of printed information out there but managing and organizing it can be difficult.  Fortunately, things can be made much easier thanks to modern OCR software. Optical character recognition enables us to convert texts from their physical form into electronic data, making way for efficient storage, search, editing, and a host of extended applications. If you’ve ever wondered how this technology works, fret not. The following is your step-by-step guide to the process.

ocr scanner1. Scanning

It all starts with capturing the text’s digital image. A page is usually scanned with the output set to black and white at 300 dpi. You can increase the resolution for better results, especially if the text is quite small. However, this will result in much larger files which will take more time to process. Conversely, you can decrease the resolution to speed it up, but this might lead to a higher number of artifacts and lower accuracy rates.

2. Pre-processing and Feature Extraction

After scanning, the digital image is then cleaned up by removing elements considered as noise, like grains, paper markings, ink blots, and the like. After that, corrections in terms of skew get underway to ensure proper orientation and uniform text size. The OCR software must be intelligent enough to differentiate text blocks from any images on the page (the best OCR software programs can do this). These text blocks are subdivided into lines of text, then to words, and finally to individual characters. Each character is then analyzed through edge detection techniques to find tell-tale features like strokes and curves.

3. Character Modeling

Now that their features are known, the characters are matched to an extensive database of fonts to find the best fit. Each OCR software has its own unique set of algorithms for this process. Several candidates may be found which are then given a confidence vote. The one with the highest score makes the cut.

4. Language Modeling

After the all the characters have been selected, accuracy is further enhanced by cross-checking all the words formed with an internal dictionary to correct spelling errors. Some OCR software programs have built-in grammar checking as well. Through this, sentences are examined for proper construction to ensure that they form a coherent thought.

5. Training

To speed up character recognition and boost accuracy rates, OCR software may have a means to train itself and learn. Humans can quickly spot repeating patterns and artificial intelligence is used to mimic this ability. In a batch of documents, for example, the software will automatically take note that a certain pattern correlates to a specific character and use that information to expand its internal database. The next time it encounters this pattern, the memory will enable it to draw conclusions more swiftly. This process may also involve human intervention wherein the user has to make a manual decision on difficult characters or words. It’s a time-consuming process at the start but for those using complex fonts like old gothic texts, this might be the best way to achieve decent accuracy. Just a note of caution: one tiny mistake could be replicated several times over until you’re left with gibberish output and a splitting headache. Train carefully!

Posted by doups3 at 4:36 pm  Tagged with: ,

What You Should Know about OCR Software Reliability and Accuracy

 Understanding OCR Technology  No Responses »
Nov 082011
 
gibberish

Some OCR programs put out gibberish.

If there is one thing that OCR software should aim for, it is accuracy. Imagine running your documents through the process then finding a ton of gibberish in the results. You’d be forced you to make manual corrections to the digitized text, wasting precious time and effort in doing so. There is no need to go through such stress. With highly accurate algorithms, these incidents are minimized to acceptable levels, ensuring that your eyebrows don’t shoot up and your blood pressure remains stable.

Measuring OCR Software Accuracy

Most OCR software developers advertise extremely high accuracy rates up to around 99%. It’s important to know how these figures are derived in order to make an informed choice. They often refer to character accuracy, which is the number of correct characters over the total. Let’s take for example a 1000-character text composed of 200 words. If we found 20 character errors, then the remaining 980 are correct for a character accuracy rate of 98%.

It would be more instructive to know the word accuracy which is the number of correct words over the total. If the 20 character errors were all concentrated in a few words then it would be easy to correct them, but chances are that these will be randomly scattered all over the text. Going back to our example, having 20 word errors in 200 words means a word accuracy rate of just 90%.

Other Factors That Affect OCR Software Accuracy

Aside from the algorithms of the software itself, external factors have a profound effect on the results. Knowing them will allow you to set your expectations and, in some cases, make adjustments to get better outcomes.

1. Scanner Settings

Since the scanned image is where it will all start, settings must be optimized to make it as clean as possible. This entails having a high contrast and a minimum resolution of 300 dpi. Some say that using gray scale instead of simply black and white for the text can also help, but tests show that this produces mixed results.

quality of paper

The quality of the paper you are working with makes a difference.

2. Paper Quality

The advertised 99% accuracy rates are indeed achievable but only in ideal conditions. Texts must be laser-printed on immaculately white paper to considerably lessen noise. In the real world, paper quality is something out of our control, as in the case of archiving old books. The pages may already have yellowed and ink has faded, with creases and markings throughout. Look for OCR software that produces decent results despite these conditions.

3. Printing Process

Good accuracy rates are achieved when the text is printed in a uniform manner in terms of size and font, which is why machine-printed texts are easier to process than handwritten ones. Each character should also be in good horizontal alignment. Some OCR software can detect skewing and correct them automatically during pre-processing. The layout should also be considered as recognition algorithms can have difficulty finding the blocks of text if there are multiple columns or if certain paragraphs wrap around images. Words within pictures such as street signs can also be mistaken as part of the text.

4. Nature of the Document

Straightforward essays can be dealt with without much fuss but pages with technical notations like science textbooks can be more complicated. Unusual character sequences such as model numbers from product manuals and catalogues may also present problems. The OCR software may think they are spelling errors and try to make unnecessary corrections. The same things may happen for archaic words and some uncommon names not found in the internal dictionary. Training the software to recognize these words can help. Repeat after me: patience is a virtue.

Posted by doups3 at 4:35 pm  Tagged with: ,
© 2011 Best OCR Software Sitemap Suffusion theme by Sayontan Sinha