Mobile Imaging SDK Feature Overview

The new imaging toolkit allows to develop mobile applications with advanced integrated imaging capabilities. Better image quality is required to archive better OCR results.

ABBYY's new SDK contains image analysis algorithms to check if the quality of the document image is suitable for OCR, so developers can make sure that their built in image processing delivers improvements. If this is not the case, the user of the app should be asked to take a new, better image - before it is sent to real OCR processing or data extraction.

  • Image Evaluation
    • Provides routines used to check if an image can be optically recognized. Includes the following evaluation options:
      • Blurred images
      • Uneven Lighting/Glare
      • Poor Lighting
    • These functions can be used to determine whether the image is fit for OCR.

  • Image Pre-processing
    • Provides a set of routines to enhance the quality of an image prior to recognition and viewing:
      • Manual Crop
      • Auto-Crop / Edge Detection
      • Rotation
      • Perspective Correction (corrects trapezium distortions)
      • Binarization
      • Inversion
      • Greyscale
      • Auto Brightness/contrast
      • Manual Contrast/brightness adjustment,
      • Noise Removal
      • Image Sharpening
    • These functions are used to adjust and improve images
      • A well cropped image snippet that contains only relevant information will deliver better text recognition results than a de-skewed image with not relevant text
  • Image Export
    • Provides routines to convert/export an images as
      • (Multi-page) Image PDFs
      • Compressed JPEG.

Related Pages: Further illustrations can be found on http://www.abbyy.com/mobile_imaging_sdk/description/

Imaging Application - ABBYY FineScanner

If you would like to see the Mobile Imaging SDK in a real world app, then get the
ABBYY FineScanner App from iTunes: http://itunes.apple.com/app/id534203582

Image Preprocessing Illustration

This screenshots shows how important image preprocessing is for the Camera OCR scenario:

  • without proper image preprocessing almost no text can be OCRed
  • with right image preprocessing capabilities (before OCR) the recognoition results can be much better
  • The Mobile Imaging SDK will of course not be able to do these “magic tricks”1), but
    • it can be used to design apps so that users make much better document images
    • removing “garbage” just by cropping the image to the area that should be recognized, delivers much better results.

Preprocessing with an ABBYY OCR backend

Notes: The images shows “backend” processing with FineReader Engine (Win), and is delivered as a code sample, FlexiCapture Engine offers similar settings

1) because therefore also OCR is needed and that is not available in the imaging tool kit