Converting image to text, mobile-style
- By Thomas McKinney
The test text
When you need to copy something down, you probably do just that. You grab a piece of paper or a notebook and you write it down, verbatim or shorthand. Today, many people are carrying handheld devices, smartphones, which are proving more and more useful and multifaceted as technology improves. Today we’ll be taking a look at OCR (optical character recognition) software built for smartphones that allows you to convert an image of text into actual editable text.
Five years ago the concept of taking a picture and running OCR software on one’s phone was inconceivable. Advancements in two smartphone features have made it possible now. The first aspect that needed upgrading was the camera. The original iPhone (which recently celebrated its fifth birthday) has a 2-megapixel camera that pales in comparison to today’s iPhone 4S, which boasts 8-megapixels and rivals the quality of consumer-level digital cameras. The second thing that now allows more powerful software is the development of faster mobile processor chips. Smartphones today are beginning to ship with quad-core processors, something computers only began doing in the last decade.
Today’s review covers two apps for Apple’s iOS, ABBYY Textgrabber + Translator and Ricoh Innovation’s Image to Text – OCR. Testing was done on an iPhone 4 and an iPhone 4S, with identical documents scanned. Textgrabber + Translator costs $0.99 while Image to Text – OCR is free.
These apps operate similarly in their functionality. Both provide a camera feature built-in as well as allowing the user to select an already existing image on the phone. While the apps’ cameras are convenient, I noticed right away that neither allows using the iPhone’s volume up button as a shutter-release. Only Apple’s own Camera app has that feature, and I’ve found that squeezing a tactile button produces better stability than pressing the screen. For this reason, I recommend taking whatever images you intend to OCR using Apple’s Camera app, and then loading them for OCR separately. This stability is of paramount importance when trying to convert an image to text because any blurriness or discrepancies in the image will lower accuracy of the recognition software. I also recommend taking more than one photo in quick succession as it increases your chances of getting a crystal-clear picture.
|
Converting image to text, mobile-style
- By Thomas McKinney
Our winner
Where Textgrabber + Translator and Image to Text – OCR differ is the mechanism for delivering the converted text. ABBYY’s Textgrabber does the character recognition directly on the phone, and immediately offers the text. One then has many options for moving it: email, text-message, and Facebook are three of the eight options. Image to Text, on the other hand, functions completely differently. Once a photo has been taken or selected, the user is presented with two options: Send by E-Mail or Send to Evernote (Evernote is a free note taking and organizing app for smartphones and computers). No matter which option is chosen, there is about a 30 second delay until you receive the converted text. There is also some awkwardness when choosing the email option as it presents you with a new email window where you must type in the recipient’s email address and send what appears to be a blank email (the subject is filled in by Image to Text). I give Textgrabber the advantage here as it offers more choices for sharing/moving the text, as well as keeping the entire process more streamlined and neat inside the app.
None of this is important if the quality of the conversion is poor, however. There’s no point to running OCR if you still have to go back and edit every other word. I went into this expecting the paid app to be significantly more effective than the free alternative. I was wrong. With identical situations replicated between two phones, Image to Text – OCR actually proved to be more accurate overall, albeit by a minor margin. Another bonus for Image to Text was that the emailed scans remained formatted similarly to how the actual photograph looked. Textgrabber maintained the formatting on the phone itself, but when the text is sent elsewhere, it becomes a solid block of text. Overall, for both conversion quality and converted text usefulness, Image to Text wins.
So why is this article featured in AE Monthly? The next time you’re at a book fair, instead of writing down notes of everything you’re interested in, why not snap a picture of the item as well as the seller’s business card? Or, if you’re more into the research side of the field, OCR the texts you need to cite and save yourself some time not retyping word for word. Books and technology aren’t peas in a pod, but there’s no reason not to take advantage when they work together.
A photograph of the text used for this review is provided, as well as links to the original plain text conversion files produced by the apps:
iPhone 4S Image to Text - OCR
iPhone 4 Image to Text - OCR
iPhone 4S Textgrabber + Translator
iPhone 4 Textgrabber + Translator
To download Ricoh Innovation’s Image to Text – OCR, visit this link on your iPhone or synced computer.
To buy ABBYY’s Textgrabber + Translator, visit this link on your iPhone or synced computer.
|