magicsen / p4all_ocr-tables

A module that exports scanned documents (image or .pdf files) to .html, recognizing tabular structures

P4A_OCR-TABLES

A module that exports scanned documents (image or .pdf files) to .html, recognizing tabular structures. A description of the recognition algorithm can be found here: ocrTables.pdf

The repository contains 4 directories

1. ocr_tables : This includes the source code that generates the OCR_TABLES.dll

2. App : This includes the source code for a sample Qt-based app to test the module

3. tessdata : This includes the traindata necessary for the OCR engine. The tessdata folder must be in the same directory as the executable

4. test files : This includes some sample files to test the module

Dependencies

The following libraries were used to build and test the module. Older subversions may also be compatible

[OpenCV 2.4.9] (http://opencv.org/) : Used by the ocr_tables module for image processing

[MuPDF 1.7] (http://mupdf.com/) : Used by the ocr_tables module for pdf processing

[Tesseract-OCR 3.0.4] (https://github.com/tesseract-ocr/tesseract) : Used by the ocr_tables module for OCR

[Leptonica 1.7.1] (http://www.leptonica.com/) : Used by Tesseract-OCR for image processing

[Qt 5.1.0] (http://www.qt.io/download-open-source/) : Used to build the sample App

App usage

Load an image or pdf file using the "LOAD" button. After the processing is finished an html file is create at "filename" + .html which can be opened using the "SHOW" button

Current limitations (to be updated in next version)

The module currently supports single column horizontal text. The input data can be single image or pdf (single and multi page) files. In cases of multi-page files, the module checks for header-footers and removes them. Page segmentation for multiple text columns, non-manhattan layouts and images is not yet implemented

Funding Acknowledgement

The research leading to these results has received funding from the European Union's Seventh Framework Programme (FP7) under grant agreement No.610510