Which software to use for PDF scanning that can extract information and send to an invoice program?

Question

I'm working as an intern at a company where they need accurate scanning of invoices that comes in PDF-forms. The program, or java-library, needs to be able to extract certain parts of the invoice so that the user doesn't have to retype the information manually.

Requirements:

• It cannot run as a Windows application
• It needs to run on Linux servers
• It needs to run using a web interface and integrate with third-party solutions

I have been trying to search, but can only find user-interactive based applications.

Any recommendations?

Please ask if anything in the question is unclear.

Answer 1

The LEADTOOLS Linux SDK has an OCR module in its Java Library. (Disclaimer: I am an employee of this toolkit’s vendor).

If your invoices come in a suitably structured manner, you can define OCR zones that correspond to each field you wish to recognize:

OcrZone ocrZone = new OcrZone();
ocrZone.setBounds(demoData.zoneBounds);
ocrPage.getZones().add(ocrZone);

This can be done for every area in every page that you want to recognize before calling ocrPage.recognize() to begin OCR.

Also, if you wish to try this out for yourself, there is a demo implemented in the form of a command-line console application in the evaluation edition of the Linux toolkit here.

Answer 2

PDF.co Web API (On-Demand) has built-in OCR and also supports template based extraction for your documents.

Below is a source code sample using template based extraction (node.js):

How to parse uploaded file (node for document parser API in JavaScript using PDF.co Web API

If you're interested we're happy to provide you with a free test account!

Disclaimer: I work for ByteScout.