Thursday, 25 June 2009

OcrNow ! Lite for iphone 3GS

Whilst trip trapping through the app store yesterday, I happened upon a free trial of an OCR program for the iphone 3gs called OcrNow ! Lite. There are others, but this one had a free version which gives you ten goes to see if it works for you. "I'll have some of that", I thought.

OCR, for those who may not know, stands for "optical character recognition", or reading words, to put it simply. Us visually impaired types might have more experience with this than most because it's software which allows us to read printed letters and the like which we maybe wouldn't normally be able to read. You pop your letter on a scanner, scan it, and the software takes a picture of the letter and converts the words in the image into plain text, which you can then read with a screen reader, edit, or whatever else you like - put straight into the trash bin (as is the case for me with junk mail) or run around the place screaming in horror (as is the case for me with household bills).
The iphone 3gs has an improved camera over the previous models (although the app does work on the previous iphones too) and finding this app gave me a thought about what use I could put it to, being a nearly blind bint and all.
Basically, the software takes a picture of your real world item containing the text you want to convert, uploads it to an online server, and then emails it back to you as an attached file in whatever format you choose - options being, pdf, plain text or rich text. You can also have the image itself included if you want. There are similar things floating about for Symbian phones, so I gather, though I've not played with them. Obviously to use this you need an active data connection.

The ocrNow ! app is very accessible via VoiceOver on the iphone 3gs. Hooray !

On opening the app, flicking from the top gets you the app title, an "add" button (this is where you go when you want to get started), and then a set of instructions which tell you what happens and how to do it.
At the bottom we have an "ocrnow" button (the "go" button for when you're taken your picture, basically), a delete button, and an options button.
The options button takes you to a screen with a "using ocr now tab", a "connection details" button, and a "jobs" button.
The using ocr now section gives you some tips about how to get the best results out of the program, more on that later. The connection settings department is where you will find the nominated email address for converted text to be sent, and the ocr server address (I advise not to fiddle with this bit, lol). The jobs button shows you a list of your jobs, how they were interpreted - for example how many characters and words the program recognised, and how many words of those were in the dictionary, plus any "suspicious characters", lol, the program detected. This doesn't mean how many dodgy reprobate types the program might have scanned your surroundings for (though wouldn't that be a triumph of an app ? heh heh) but how many characters the program couldn't quite suss. More about this section in a bit.

Now, here's where I have to point something out. When I started to fiddle with this app, I didn't have the idea of actually reviewing such things. It only occurred to me that my messing about with such things could possibly benefit others when I'd fiddled with a couple of programs and found my way around them already. So when you start this app for the very first time, you may well get a setup screen that tells you what to do and asks you for an email address to send the output to. I can't remember - it was yesterday when I first met the app and I have serious memory problems much of the time. I will try to remember to document the next program I "review" right from the off though.

So, to the meat of the matter.

On the first screen of the app, you have your add button as I said. Press this to get started scanning your text. The camera opens and tells you "viewfinder, image". Here is where the tips the program gives you in the using ocr now section really help. It recommends keeping the item flat, for example on a desk or on a wall. It recommends good light (well, we'll have to chance it on that one I guess). It recommends holding the phone about 13 inches away from the paper if you're doing an a4 letter, for a3 about 20 inches. Try to keep the border of the picture in line with the border of the paper, it says. Well, those of us who can see abit might have better success with this one - but it is worth remembering something that caught me out a few times - the camera on the iphone is on very top right corner of the rear of the phone if you hold it facing you with the home button at the bottom. I guess most of us would imagine it was in the middle of the phone, but it isn't. This makes lining stuff up abit less instinctive but there you go.
Once you think you've lined up your paper, gently double tap the screen somewhere in the middle. This focuses the camera on the item below it. You'll hear a sound when it focuses. I say do it gently because whatever movement you put on the phone will likely mess with the position of it and thus the focus. Double tap the take picture button to take a picture (or use the split tap method if you prefer - I reckon this is better for taking pictures, as it moves the phone less). It's probably worth messing about with the camera if you're so inclined beforehand to find your way around it.
When you've done this, you will get a preview of the image and a "retake" button at the bottom of the screen, and a "use" button. Double tap the use button to use the image (or retake if you've loused it up).
Then you're taken back to the previous screen, which will be showing a preview of your image, and giving you the magic "ocrNow!" button at the bottom left of the screen. Activate it !

You're now taken to a screen which allows you to set the options for your output. From top to bottom, you got the subject text field where you will likely want to put the subject of your email.
Then flicking past that gets you a choice of pdf, text or rtf buttons - then you're told this is the output format you're choosing. Ass backwards I know but you can't have it all. Your previous choice of format is remembered and indicates as selected.
Flicking by gets you a yes button and a no button, then a "return image" description - again backwards. Check yes if you want to have the image emailed to you to.
Then you have your ocr now button, which is where the magic happens (or a cancel button next to it if you have given up all hope by this point).
Activating the ocrnow button sends the image to the server. The screen changes to a progress bar but doesn't say anything when it's done so, give it a wee while. If you've got push notification for your email (assuming you're sending it to yourself and want to get the result on your iphone) you'll know when it's done as an email will come in. If not, about thirty seconds should do it. There's a button at the bottom of the screen which changes from "cancel" to "continue" when it's finished, so if you're light of touch this is a good way to find out when the magic's happened. Double tapping continue takes you back to the screen with the add button for further excitement (or boredom, depending on your point of view.)

Fire up your email on the iphone if you want instant gratification and you're not near your pooter, and look for an attachment right at the bottom of the blurb that the company who make the app write you when they send you your output. Flicking through this take a while. Double tap the attached file to get your outputted text, where you can read it at your leisure.
I have to admit I totally missed the attachment hiding at the end of the blurb to begin with and even emailed the company asking if the text could be sent inline with the email as opposed to attached, as I couldn't read attachments using my iphone (though in the back of my mind I recalled that you could - I just couldn't find them). What a silly cow I am ! They were very kind and said they would definitely consider this option in future version of the app, by the way. They must have thought I was a right eejit though.
If you don't want to wait for email, in the jobs section of the ocr now app you get a list of the jobs you've done (in the program, not in real life, heh) and options to delete them or download them. Choosing download gets you an option to view the output - which will then go to a screen which will read it all to you, but is not interactive so it will read it in one big lump (keep up at the back there, lol) or to email the output, and you can choose where to email it this time as opposed to having a default address set up.

So - does it work and how well ?

Well, yes, it does work. I snapped a letter I had been sent on a4 paper, and after a few tries of lining it up I managed to get an output that sent the entire letter back to me in a format I could actually read, with very few mistakes or missed characters. Anyone familiar with ocr will know that sometimes you do get a few letters wrong here and there, but you know what the word is supposed to be by context and common sense. I tried a book cover - pretty successful but not quite as easy to line up as the letter. I got real words out of it though, and they even made sense ! And a miniscule chocolate wrapper than even sighted folks might struggle to read I'd wager. I got enough to know that it was 54 percent cocoa solids and made by Kraft Foods. I have no idea where I had lined it up in the viewfinder though.
I suspect this concept might be more successful for those with some remaining vision than those without. I have some left so I was able to line the letter up ok with a good light and contrasting background, but not the smaller things I tried. All of this will depend on one's level of sight and perseverance. I imagine some enterprising and inventive types could make a frame for the iphone and the items to be scanned which makes the process of lining up and focusing a snap, but that would rather destroy the on the go nature of the app. Still, if you don't have a solution already, that would be an idea worth considering I reckon - because compared to scanning software and a camera for your computer, or a flatbed scanner and software, or a standalone solution (that's my credit card I hear screaming in the background at the mention of such things !), the app is dirt cheap.
The beauty of it is that there is a trial version which will do you ten goes, and the program is fully accessible with VoiceOver on the iphone, so you can try it yourself and see if it works for you. To buy the full version is £7.99.
I'm going to give it a go tonight at the pub and see how it copes with menus and the like.

If you've got some residual vision and don't want to spring for a mobile magnifier, or don't want to carry one around in addition to your iphone, ocrNow ! may well be what you are looking for ... pun intended.


  1. This comment has been removed by the author.

  2. Hi,

    About my last post. Oops, my bad. I thought Ms. Farmer who writes the House of the Scorpians books coined the term Eejit, but I see now that it is an Irish word. My bad. Us Canadian's probabl never use this word and I suppose that is why I was unfamiliar with it. Good post though.