ABBYY has a long history on offering products and technologies for mobile devices.

• Already in 2002 ¹⁾ the Lingvo dictionaries for PocketPCs²⁾ allowed travelers to look up words digitally on their devices so that they did not have to carry a paper dictionary. At this time, no mobile data plans were available, so the dictionaries had to be very compact, to fit on the devices.

• In 2006 ³⁾ the first Mobile OCR toolkit was launched. The new SDK allowed recognising printed texts and business cards directly on the device where the images were captured with the built-in camera. The SDK started on Symbian, iOS and Android were added when they were released.

• Since 2012 ⁴⁾ the ABBYY Mobile Imaging SDK developers can extend their apps with technologies that allow to check and enhance the image quality before the documents are sent to a data extraction back-end, that is powered e.g. by FlexiCapture (Engine).

• In November 2016 ABBYY's new Real-Time Recognition SDK was made available. It combines intelligent text-recognition out of a video stream with translation capabilities based on dictionaries. Developers can now build the next generation of their apps, that “reads” text out viewfinder of the camera. The recognised words can be used in the app, translated or merged to a new augmented video stream.

	Mobile Imaging SDK	Mobile OCR Engine	Real-Time Recognition SDK
Technology Usage	- Image enhancement - Image quality checks - Creation of high quality (document) images	- Load/take images and run optical character recognition on the mobile device - Extract text/data for further processing	- “Reads” text on smartphone camera preview screen - Use the recognize text to trigger other tasks
Usage/Scenario	- Build or extend your app with intelligent document capture features - Add quality check and image enhancement capabilities to the (document) images before they are sent to the processing back-end	- Edit and email the text - Send it to an online translation service - Apply business card reading and add the details to your address book - Barcode Recognition - Personal information management apps - Many others	- New user experience through real-time interaction - On device, instant recognition and (word by word) translation - Capture texts from real-life natural scenes, e.g. from guide signs - Combined with the user's location data, an app can “read” what the user sees, ideal for special promotion campaigns
Market Availability	2012	2006	2016
Operating Systems	- Android - iOS	- Android - iOS - Windows, macOS, other 5)	- Android - iOS
Processing
Input	- one or multiple images	- one image	- series of video frames
Image Preprocessing	- Multiple features to analyze and manipulate the image, e.g. cropping, color & contrast adjustments	- Preprocessing to prepare the image for good OCR results, e.g. detection of the page orientation, deskewing , geometric transformation, binarisation	- The input is a live video stream, there is no explicit image pre-processing API - Internal processing is aligned with the recognition technologies
Layout Analysis	- The overall image quality is analyzed (sharpness, blurring, etc. - No specific layout elements are detected	- Text, image and barcode areas are detected - Reading order of text blocks is restored	- No traditional layout analysis that expects a layouted document - Text recognition areas are detected flexibly on any background and at any angle, this allows “finding” text in natural scenes and dynamic detection of text lines
Recognition Technologies	- none	- Text recognition (images) for Latin, Cyrillic or Greek alphabets - 4 Asian complex script languages: Chinese (Simplified and Traditional), Japanese and Korean - Barcode recognition - Business card recognition	- Text recognition (video) for Latin, Cyrillic or Greek alphabets - 4 Asian complex script languages: Chinese (Simplified and Traditional), Japanese and Korean
Languages & Dictionary Support	- none	- Support for 60+ languages - OCR dictionaries for 23 languages - Business card recognition for 21 languages	- OCR Support for 60+ languages - OCR dictionaries for 24 languages 6)
Translation	- none	- none built in, but the recognized text can be sent to an online translation service, e.g. Bing or Google Translate - An internet connection is required	- Built-in word by word translation option - 18 Language combinations available for translation - No internet connection needed
Export	- Images - Multi-page PDFs (image only)	- Plain text	- No text export - API-Access to the recognized and translated text
Further Details
Processing Power	- Low CPU requirements, no OCR applied	- Processing speed directly related to CPU speed - Parallel recognition with up to 4 threads, ideal for Asian documents	- CPU intense - Quad-core mobile processor recommended (available on all latest phones)
Related Articles	- Mobile Imaging SDK	- Mobile OCR Engine - Cloud vs. On Device OCR Processing for Mobile Applications	- Real-Time Recognition SDK
Illustrations

Stand: 12/2016

With ABBYY's mobile tool-kits you can extend your existing or new apps with powerful imaging and text recognition features:

• Mobile OCR Engine: Eliminate manual text typing, enable text recognition, business card and barcode recognition on the device.
• Mobile Imaging SDK: Image quality check and enhancement for professional document capture apps
• Real-Time Recognition SDK: New user experience to recognize and translate texts, ideal for for modern, innovative apps that engage their users

• Cloud vs. On Device OCR Processing for Mobile Applications
• Business Card Reading

Back to: Comparisons