Keren Blau
Posts: 11
Nickname: kerenblau8
Registered: Jul, 2012
|
|
Re: Translate text from photos using ABBYY FotoTranslate mobile OCR software
|
Posted: Jan 30, 2014 10:59 AM
|
|
Jhon, came across the article you referred to during a search.
Since the article was originally published, there's been quite an advancement in OCR technology.
The original article focused on mobile OCR for people in foreign locales, who are scanning random texts, such as brochures, menus and the like.
Now in 2014, the OCR/mobile/traveler combination is quite ubiquitous.
May I offer another angle? OCR/mobile/traveler but for templated text....
In OCT solutions, it's important to differentiate between two types of text sources: structured or unstructured sources. Here's what I mean:
- "Unstructured" documents are unique and random, for example hand-written faxes to a support center.
- "Structured" documents, have a standard layout, e.g. forms filled out by customers (where the fields are always in the same place), various cards (such as business cards, drivers' licenses, credit cards, etc with predefined data such as name, ID number etc).
The tools required for the "template" texts detect and OCR only the specific text fields in the image, clean out "noise", and output the text in a structured manner (e.g. Name = John Smith, ID Number = 123456).
A tourist in a foreign land would certainly find ABBYY FotoTranslate on her smart phone a valuable tool. But what about a tourist who needs to perform a credit check? Pay by bank check? Or hire a car? Or present membership at a trade fair? In those use cases, applying the card to a tablet installed in a kiosk, or point-of-sale, or registration counter and getting immediate confirmation is nuts-and-bolts today.
For mobile OCR solutions for template texts, there are a host of providers out there. I have compiled a short list of ones I know of.
Kofax (http://www.kofax.com/software) web-capture - a web-based OCR app that can take in any type of structured form and process it, as far as I know it requires some "training" as to the data in the forms. They provide not only the OCR engine but also a full web interface for uploading and manipulating the scanned files - this interface can be embedded into a web app using Java or .NET.
CSSN: http://www.card-reader.com - they provide an engine that can take card-type documents like ID cards, driver's licenses, medical cards, bank checks, credit cards, etc, and extract the data to a text file. It's possible to integrate the engine with a web application using the SDK: http://www.cssn-developers.com/sdk.htm.
- Top Image Systems - their eFLOW product does data capture from healthcare forms, driver’s license applications, tax forms, multiple-choice examinations, insurance claims, or any other structured forms. Also enables integrating with existing systems including browser-based interfaces. See http://www.topimagesystems.com/solutions/business-processes/forms-classification/11-solutions/overview/185-eflow-integra
My 2 cents...
|
|
Previous Topic
