Image processing functions for a better OCR quality

Image pre-processing

Advanced image pre-processing functions

After receiving images, ABBYY FineReader Engine performs a range of image processing functions to improve the quality of document images for the actual recognition process. The technology uses a set of pre-processing functions and increase the quality of images. This way, even images with a low quality or documents photographed with smartphones can be efficiently processed and deliver high recognition results.

Basic image processing functions

FineReader Engine offers a number of image processing functions, such as:

Image scaling
Image cropping
Image clipping
Creating previews
Image rotation (90, 180, and 270 degrees)
Lines straightening
Mirroring and inverting
Remove noise
Enhance local contrast

Advanced image processing functions

ABBYY Camera OCR technology
Pre-processing for documents with stamps and written notes
Auto-splitting of double-pages
Autodetection of page orientation (90, 180, and 270 degrees)
Automated image de-skewing (up to +/- 20 degrees)
Image despeckling (image clean-up)
Despeckling images in individual blocks Texture filtering and Adaptive Binarization
Adjusting text and background color
Data extraction from fields with various borders and frames

Unrivalled photo processing (Camera OCR)

Documents photographed by digital cameras, smartphones or tablets provide suitable resolution and image quality; however they carry some device-specific distortions. The intelligent technology identifies images captured by a digital camera and implements special image processing algorithms to eliminate distortion on digital photos, such as blur, curved text lines, missing information on image resolution or errors caused by insufficient light.

ABBYY FineReader Engine contains the following Camera OCR features:

Correct image resolution
Straighten curved lines
Automatic 3D perspective distortions correction

Сorrection of 3D perspective distortions

3D distortions cause multiple problems with recognized text: OCR errors might occur, page splitting fails and font size changes (from top to bottom). After correction of the original image, the OCR process provides excellent results:

without correction

Сorrection of 3D persrective distortions

with correction

Blurred images correction

If a camera is not fixed during exposure, an image could be slightly blurred. This is not visible on camera screen, but might cause OCR problems. After transformation of the source image, the binary image becomes better readable:

without correction

with correction

ISO noise correction

This type of noise looks like the picture is filled with a great number of pixels of different colors. This can lead to wrong binarization and some characters could be lost. By applying a special filter, ABBYY FineReader Engine smoothens image background and prevents text loss:

without correction