Overview OCR & Capture SDKs NLP, Semantic & Linguistic SDKs Imaging

Mobile Imaging SDK

ABBYY Mobile Imaging SDK is a toolkit for Android and iOS that allows to creating/extending existing mobile applications with image pre-processing optimized for paper documents. The new toolkit makes it much easier to create professional, but light “document capture” apps that are linked to mobile on-device OCR or server/cloud based OCR back-ends.

Key Features

  • Optimization of (document) images, that were taken with a mobile device
  • Pre-process, enhance and compress the images for OCR (optical character recognition), before they are sent to recognition (on device or back-end)
  • Special analysis of the image, to check if the text quality is suitable for OCR, before submitting it for processing.
    If the quality is not good enough, the user can try to make a better picture.
  • It is (of course) also possible to enhance images “only” for a better appearance (visual quality) before they are stored in image library. ;-)

ABBYY Mobile Imaging SDK provides a set of functions:

  • Image Pre-processing to enhance the quality of an image for end-user visualization, not for recognition.
  • Image Verification to verify if an image can be recognized with OCR.
  • Image Export to convert/export an image from JPEG to either PDF or JPEG.

Why does ABBYY release a Mobile Imaging SDK?

  • Image quality is extremely important to achieve good OCR results!
    Developers known the “programming proverb”: Garbage in, garbage out! , it can be applied here as well. (more details: GIGO)
  • High text recognition accuracy is key for the extracted data that will be used in the business processing steps!
    Example:
    • Data that is extracted from a receipt will then be used to simplify the travel expense process

Image Quality & Text Extraction

  • Traditionally paper documents are digitized with a scanner.
    The benefits are
    • Option to select the scanning resolution
    • Perfect lightning for the page
    • Fast scanning speed
    • Duplex scanning option
    • Multi- format out put possible e.g.: grey-scale & color
  • Digital cameras in mobile phones are standard and the quality of has increased over the last years:
    • auto-focus and 5+ Mega pixel are standard, so the devices can be used as a mobile scanner.
    • technically it is possible to take a high quality picture of a document with a phone camera.
    • …. but, in reality the images very often have lower quality than scanned documents.
  • To get the best achievable OCR results, special image pre-processing has to be applied, because
    • the images can be much more rotated (than on a scanner)
    • geometric distortions are more likely (than on a scanner)
    • often the lightning conditions are not the same all over the page (shades and glare)
    • the image can be un-sharp or blurred
    • the imaging chip in the camera adds “noise” (= digital dirt) to the image - especially when there is only less light.
    • this might end up in low contrast
    • → all these “defects” make it harder for the OCR technology to “read” the text

Resume

  • Image Quality does matter!
  • Special, tuned image pre-processing is needed to archive good OCR results, that at the end leads in good applications and user-friendly services.

Further Information & Image samples:

Features Mobile Imaging SDK