How image preprocessing works in SDK products

Avatar

Sergey Pilipchuk

  • Edited

Question

How image preprocessing works in SDK products?

Answer

Once document pages are loaded, SDK offer a variety of image preprocessing options which allow amending the document images and optimize them for the actual recognition step. The goals of this preprocessing are to enhance the document pages as much as possible and reach the best OCR results during the recognition step.

Image preprocessing options:

  • Cleaning routines to remove noise and garbage from the image
  • Optimization of images taken by digital cameras and smartphones, e.g. straighten curved text lines
  • Dual-page splitting by photographed books
  • Different algorithms for deskew (up to 20 degrees)
  • Built-in adaptive binarization and texture filtering

image-preprocessing.png

 

Also, the improved binarization technologies have an impact on overall recognition quality and processing speed.

The level of control is different, depending on the target audience

  • Several samples allow the users / admins to select some basic options.
    The toolkits like FineReader Engine have a full API for image pre-processing:fre10_samples_image-processing.png

For more details, please see the FineReader Engine sample: Image Preprocessing for OCR

fre10_sample_camera_ocr.png

OCR Processing Steps
OCR Accuracy Measurement

Have more questions? Submit a request

Comments

0 comments

Please sign in to leave a comment.