OCR/Bitmap matching in a PDF

We have build a Finance program and now need a invoice scanner. The scanner we need should take a invoice (PDF) and then make OCR on the invoice and grab information from the invoice. That can be invoice lines, dates, logo, company name, VAT etc etc. The PDF can both be computer generated or scanned by a scanner. The difficult is the invoice layout is not the same, every company have it's own layout. So the process is the user will scan this invoice, then he will via our program teach the system where it can locate the information on the invoice. He will visual point out where the logo, company name, date, invoice lines VAT etc are located on the invoice. Next time he receive a invoice the system recognize the invoice either on the logo or company name and then know how to extract the information. So what I need is : • DLL there can recognize invoice and also teach the DLL about the invoice layout • DLL have a function there take a PDF (both as stream or filepath) and then return the information in XML • DLL Have a function there teach the DLL about the invoice layout. • Then I want a program so I can test the DLL where I can visual teach the program about the invoice layout, give a path where there are 10 different invoice PDF and then it will save the XML information for each found invoice. The requirements are : • It's should be build C# • On a normal computer the recognizing and extract of information must not take over 2 seconds for 1 invoice with 100 layout templates • It should could recognize the invoice by the company (Text) and Logo (Bitmap) so it should also be able to compare bitmaps. There will not be paid any money before I can see the solution is working. Best regards Kim Johansen