OCR Text Comparer

Allows user to select two image files. Then after pressing 'Compare' button texts from two images are compared and differences highlighted.

How it works

Imges are preprocessed and texts are extracted from both images using Tesseract ocr engine. These texts are then compared using google's diff-match-path algo. Result is put into text box below images. When user clicks on line in text box, this line is highlighted (+ pan & zoom) in both original images allowing quick peek into text's original location.

Quality of results is highly dependent on quality of ocr, for best results use:

• images with at least 400 DPI (600 DPI and more is optimal)
• perform comparison selecting appropriate language
  • to keep size of project in check only 'English' language is included, to extend add tesseract *.traineddata to src/OcrTextComparer/OcrTextComparer/tessdata/

Code organization

Solution contains two projects - ImageCompare and OcrTextComparer. ImageCompare project contains everything related to comparison, including compare control - ImageCompareControl. OcrTextComparer project basically acts as a host for compare control, it contains components required to provide inputs for compare control.

Dependencies

• .NET Framework 4
• Nuget packages:
  • EmguCV
  • Magick.NET-Q8-AnyCPU
  • OpenTK
  • OpenTK.GLControl
  • Tesseract
  • ZedGraph