I am using the SDK version 10.4 and cannot get OCR to work with pages or PDF files. The documents are already created and have pages/PDFs assigned to them.

I run the following code against a valid page with text:

   try
   {
     OcrEngine ocr = OcrEngine.LoadEngine();
     ocr.Run(lfDocNew, new PageSet(pageNum.ToString()));
   }
   catch (Exception ex)
   {
     result.ErrorMessage = "OCR Error on page " + pageNum + ": " + ex.Message;
   }

And get this error thrown:

Error: 

"Interface not registered"

Stack Trace:

    at BPIClientLib.IBPProcessEx.Process(IBPPageEx pUnkPage)
   at Laserfiche.DocumentServices.OcrEngine.ProcessImage(String imagePath, DocumentInfo document, PageInfo page)
   at Laserfiche.DocumentServices.OcrEngine.ProcessPage(DocumentInfo document, PageInfo page)
   at Laserfiche.DocumentServices.OcrEngine.Run(DocumentInfo document, PageSet pages)
   at mviLFPost.LFRAPostEngine.AssignPages(LFDocumentInfo DocInfo, DocumentInfo lfDocNew, DocumentImporter importer, PostResult& result) in C:\vs\mvidev\mvi\Dev\Laserfiche Integration\SVR\mviLFPost\mviLFPost\LFRAPostEngine.cs:line 1381

When I run the following code with a PDF electronic document, it does not extract text nor does it throw any error.

      try
      {  
        using (TextExtractor te = TextExtractor.LoadExtractor())
        {
          te.ExtractFrom(lfDoc);
        }
      }
      catch (Exception ex)
      {
        result.ErrorMessage = "Error extracting text for electronic document: " + ex.Message; // Don't fail on entire import if text is not extracted.
      }

Am I missing a required component to run OCR when importing with the 10.4 SDK?  Relying on Workflow after the fact really slows down the server when importing many documents.