With today's intelligent scanners and advanced user-friendly software, scanning photos and images has become a simple task for even the most novice computer user. What you may not be aware of, however, is that capturing text with your scanner and converting it into an editable document (a tedious process in days past) can now be accomplished with just as much ease.
There are many reasons why you might want to convert a printed page of text into a computer file. Perhaps a newspaper article featured your business, and you would like to post it on your company website. Or maybe you have an old manuscript that you’d like to put in electronic format for editing. Or you've collected a large number of forms that need to be scanned and tabulated. Whatever your need, using a scanner can significantly reduce the amount of time you spend entering information into your computer by hand.
While the process of converting printed text to a computer file is generally called Optical Character Recognition (OCR), in truth the technology goes far beyond simple identification of individual characters. The general term "text recognition" more accurately describes the advanced capabilities of today's software and hardware. Not only can you scan text with accuracy in excess of 99%, but you can also re-create an entire page of text and graphics in electronic form, while maintaining the original layout.
Simple text scanning
HP's Precision Scanning software makes scanning text a simple task. To make a scan of a text alone, follow these steps:
- Align your document on the scanner and click the "Preview" button.
- In the preview window, use the mouse to select the text you wish to scan. The Output Type menu should change automatically to "Text", though if there is a background image you may have to change it manually. HP scanners will often recognize and select the text for you.
- Click the "Scan to" button and select an application as a destination for the scanned text, most likely a word processing program.
- Click "Scan" to complete the scan and send it to the destination application. If the destination supports text, the scanned image will be converted to text and dropped into the application. If not, it will be sent as an appropriate image format.
OCR software uses several different methods to recognize text images and convert them into editable text in electronic format.
- Pattern matching is one of the earliest and simplest methods of OCR. In this method, each character is divided into a matrix of blocks, which are then compared with a library of sample characters in order to find a match. This method can be highly accurate when the source document contains only a limited number of fonts, such as a computer-generated hardcopy. But when the software must try to read an unknown font, accuracy can drop dramatically.
- Feature extraction is a more advanced method of OCR, in which each character is broken down by the features that make up that symbol. For example, the letter “A” is made up of a crossbar and two slanted lines joined at the top. Because this method does not rely upon existing font libraries, it is more flexible than pattern matching methods, and it is even capable of reading carefully handwritten documents.
- Context checking will compare a questionable character with those on either side and determine the most logical translation. Or, an OCR program can use word lists and dictionaries to check and replace questionable characters.
Scanning complex text
If your source contains combination of images and text that you want to capture in electronic form, you can easily capture not only the discrete elements with a single scan, but even the layout of the original document.
- Align your original document on the scanner and click "Preview."
- Use the mouse to select the entire document.
- In the Output Type menu, select "Text and Image."
- From the scanner's Settings or Preferences menu choose a text output mode:
- Flowed Text preserves the layout of the original as closely as possible, including images, text, columns, tables and so forth. Each element of the original is placed within a frame in the text file in the same place as it appeared on the original. This method is advantageous if you want to capture the layout but are not planning on making major edits.
- Flowed Text mode places each of the scanned elements in-line, or consecutively in the document. This method is more useful if you are planning on making major edits, and plan on reformatting the document afterwards.
5. Click "Scan to" button to pick a destination application.
6. Click "Scan" to complete the scan.
OCR scanning tips
The most important thing to remember about getting accurate text scans is to start with a good quality original. Tears, wrinkles and smudges can confuse the OCR software and lead to errors in the final output. Touch up a dirty original with a touch of correction fluid, or make a photocopy to improve the contrast of the original. Here are a few other tips for making your text more readable by your scanner.
- If the text on the original is slanted, or the page shape is irregular, it can be difficult to align the text properly for scanning. Draw a line on the back of the page that corresponds with the baseline of the text, and use that line to align the page with the scanner's guides.
- Translucent papers and newsprint allow text from the opposite side to show through the paper, confusing the scanner. Place a sheet of black paper over the back of the page to block the light.
- Certain types of text are sure to cause trouble for your OCR software. Text that is underlined boldfaced, or italic can make characters unreadable to the scanner. Joined text or kerned text (when the space between letters has been adjusted to fit together more efficiently) can also pose difficulties for OCR programs. Recognizing in advance that these problems might occur will save you time and trouble in scanning.
- Use your word processor's spell checker to go over the scanned text when you are finished scanning. Though OCR programs have built in dictionaries, they are seldom as extensive or precise as the one you've personalized for your own use.
OCR can provide some complex challenges for your computer, and having a fast computer and scanner can ease the process. At least 32MB of RAM is necessary (though 64MB is better) for you computer to carry out the complex processes of differentiating images from text, identifying characters, and translating a document's layout into electronic form. The faster your processor and scanner, the less time you'll spend waiting for the scan to be complete.
Additionally, if you plan on scanning large numbers of pages at a time, you'll find an automatic document feeder (ADF) will greatly ease your task.