FormRecognitionEngine Class

Imports Leadtools 
Imports Leadtools.Barcode 
Imports Leadtools.Codecs 
Imports Leadtools.Forms 
Imports Leadtools.Forms.Ocr 
Imports Leadtools.Forms.Processing 
Imports Leadtools.Forms.Recognition 
Imports Leadtools.Forms.Recognition.Barcode 
Imports Leadtools.Forms.Recognition.Ocr 
 
''' This class contains all Master Form information. 
Friend Class MasterForm 
   Public Image As RasterImage 
   Public Attributes As FormRecognitionAttributes 
   Public Properties As FormRecognitionProperties 
   Public ProcessingPages As FormPages 
 
   ''' This constructor initialize all MasterForm members to zero/null/Empty. 
   Public Sub New() 
      Image = Nothing 
      Attributes = Nothing 
      Properties = FormRecognitionProperties.Empty 
      ProcessingPages = Nothing 
   End Sub 
End Class 
 
''' This class contains all my Form information. My Form is the filled form that is going to be identified and processed by the application. 
Friend Class MyForm 
   Public FileName As String 
   Public Image As RasterImage 
   Public Attributes As FormRecognitionAttributes 
   Public Master As MasterForm 
   Public Result As FormRecognitionResult 
   Public Alignment As IList(Of PageAlignment) 
   Public ProcessingPages As FormPages 
 
   ''' This constructor initialize all MyForm members to zero/null/Empty. 
   Public Sub New() 
      FileName = Nothing 
      Image = Nothing 
      Attributes = Nothing 
      Master = Nothing 
      Result = Nothing 
      Alignment = Nothing 
   End Sub 
End Class 
 
''' TestForms is a class that has all the needed methods to recognize and process unknown forms. Once an instance of it is created StartUp should be  
''' called first to initialize Ocr and barcode engines, besides building the Master Forms set, then Run method should be called to recognize and  
''' process a specific form, finally the ShutDown method should be called once the program is terminated. 
Friend Class TestForms 
   Private RecognitionEngine As FormRecognitionEngine 
   Private ProcessingEngine As FormProcessingEngine 
   Private FormsBarcodeEngine As BarcodeEngine 
   Private FormsOcrEngine As IOcrEngine 
   Private FormSet As List(Of MasterForm) 
   Private RecognizeFirstPageOnly As Boolean 
   Private LoadFromScanner As Boolean 
   Private FormsCodec As RasterCodecs 
 
 
   '=========================================================================== 
   ''' This method loads an image from scanner. 
   Public Function LoadImageScanner(ByVal count As Integer) As RasterImage 
      'TODO 
      Return Nothing 
   End Function 
 
   ''' This method loads an image from scanner for recognition purposes. 
   Public Function LoadRecognitionImageScanner() As RasterImage 
      'TODO 
      Return Nothing 
   End Function 
 
   ''' This method loads an image from a file. 
   Public Function LoadImageFile(ByVal fileName As String, ByVal firstPage As Integer, ByVal lastPage As Integer) As RasterImage 
      ' Load the image and verify that we loaded all the pages  
      Dim image As RasterImage = FormsCodec.Load(fileName, 1, CodecsLoadByteOrder.Bgr, firstPage, lastPage) 
      Return image 
   End Function 
   '=========================================================================== 
   ''' This method loads an image from a file for recognition purposes. 
   Public Function LoadRecognitionImageFile(ByVal fileName As String) As RasterImage 
      Dim image As RasterImage 
      If RecognizeFirstPageOnly Then 
         image = LoadImageFile(fileName, 1, 1) 
      Else 
         image = LoadImageFile(fileName, 1, -1) 
      End If 
      Return image 
   End Function 
 
   ''' This method loads an image for recognition purposes. 
   Public Function LoadRecognitionImage(ByVal fileName As String) As RasterImage 
      Dim image As RasterImage 
      If RecognizeFirstPageOnly Then 
         image = LoadRecognitionImageScanner() 
      Else 
         image = LoadRecognitionImageFile(fileName) 
      End If 
      Return image 
   End Function 
 
   '=========================================================================== 
   ''' This method creates a form attributes for recognition. 
   Public Function CreateForm() As FormRecognitionAttributes 
      Dim attributes As FormRecognitionAttributes = RecognitionEngine.CreateForm(Nothing) 
      RecognitionEngine.CloseForm(attributes) 
      Return attributes 
   End Function 
 
   ''' This method adds a page to form attributes for recognition. 
   Public Sub AddPageToForm(ByVal image As RasterImage, ByVal attributes As FormRecognitionAttributes) 
      RecognitionEngine.OpenForm(attributes) 
      RecognitionEngine.AddFormPage(attributes, image, Nothing) 
      RecognitionEngine.CloseForm(attributes) 
   End Sub 
 
   ''' This method deletes a page from form attributes for recognition. 
   Public Sub DeletePageFromForm(ByVal pagenumber As Integer, ByVal form As FormRecognitionAttributes) 
      RecognitionEngine.OpenForm(form) 
      RecognitionEngine.DeleteFormPage(form, pagenumber) 
      RecognitionEngine.CloseForm(form) 
   End Sub 
 
   ''' This method creates a form with its pages for recognition. 
   Public Sub CreateFormForRecognition(ByVal form As MyForm) 
      form.Attributes = CreateForm() 
      Dim saveCurrentPageIndex As Integer = form.Image.Page 
      Dim i As Integer = 0 
      Do While i < form.Image.PageCount 
         form.Image.Page = i + 1 'page index is a 1-based starts from 1 not zero 
         AddPageToForm(form.Image, form.Attributes) 
         i += 1 
      Loop 
      form.Image.Page = saveCurrentPageIndex 
   End Sub 
 
   '=========================================================================== 
   ''' This method creates a Master Form recognition attributes. 
   Public Function CreateMasterForm(ByVal name As String) As FormRecognitionAttributes 
      Dim attributes As FormRecognitionAttributes = RecognitionEngine.CreateMasterForm(name, Guid.Empty, Nothing) 
      RecognitionEngine.CloseMasterForm(attributes) 
      Return attributes 
   End Function 
 
   ''' This method adds a page to a Master Form recognition attributes. 
   Public Sub AddPageToMasterForm(ByVal image As RasterImage, ByVal attributes As FormRecognitionAttributes) 
      RecognitionEngine.OpenMasterForm(attributes) 
      RecognitionEngine.AddMasterFormPage(attributes, image, Nothing) 
      RecognitionEngine.CloseMasterForm(attributes) 
   End Sub 
 
   ''' This method deletes a page from a Master Form recognition attributes. 
   Public Sub DeletePageFromMasterForm(ByVal pagenumber As Integer, ByVal form As FormRecognitionAttributes) 
      RecognitionEngine.OpenMasterForm(form) 
      RecognitionEngine.DeleteMasterFormPage(form, pagenumber) 
      RecognitionEngine.CloseMasterForm(form) 
   End Sub 
 
   ''' This method creates a Master Form with its pages for recognition purposes. 
   Public Sub CreateMasterFormForRecognition(ByVal name As String, ByVal form As MasterForm) 
      form.Attributes = CreateMasterForm(name) 
      Dim saveCurrentPageIndex As Integer = form.Image.Page 
      Dim i As Integer = 0 
      Do While i < form.Image.PageCount 
         form.Image.Page = i + 1 'page index is a 1-based starts from 1 not zero 
         AddPageToMasterForm(form.Image, form.Attributes) 
         i += 1 
      Loop 
      form.Image.Page = saveCurrentPageIndex 
   End Sub 
 
   '=========================================================================== 
   '''This method compares a Form to a Master Form using all pages. 
   Private Function CompareForm(ByVal master As FormRecognitionAttributes, ByVal form As FormRecognitionAttributes) As FormRecognitionResult 
      Return RecognitionEngine.CompareForm(master, form, Nothing) 
   End Function 
 
   '''This method compares the first page of a Form to the first page of a Master Form. 
   Private Function CompareFirstPage(ByVal master As FormRecognitionAttributes, ByVal form As FormRecognitionAttributes) As FormRecognitionResult 
      Dim resultPage As PageRecognitionResult = RecognitionEngine.ComparePage(master, 1, form, 1) 
      Dim result As FormRecognitionResult = New FormRecognitionResult() 
      result.Confidence = resultPage.Confidence 
      result.LargestConfidencePageNumber = 1 
      result.PageResults.Add(resultPage) 
      result.Reason = FormRecognitionReason.Success 
      Return result 
   End Function 
 
   '''This method identifies the type of the form based on the comparison results. 
   Public Function IdentifyForm(ByVal results As FormRecognitionResult()) As Integer 
      Dim maxIndex As Integer = 0 
      maxIndex = 0 
      Dim i As Integer = 1 
      Do While i < results.Length 
         If results(maxIndex).Confidence < results(i).Confidence Then 
            maxIndex = i 
         End If 
         i += 1 
      Loop 
      If results(maxIndex).Confidence < 30 Then 
         maxIndex = -1 'no match 
      End If 
      Return maxIndex 
   End Function 
 
   '''This method recognizes the type of the unknown form 
   Public Sub RecognizeForm(ByVal form As MyForm) 
      CreateFormForRecognition(form) 
      Dim results As FormRecognitionResult() = New FormRecognitionResult(FormSet.Count - 1) {} 
      Dim i As Integer = 0 
      Do While i < FormSet.Count 
         If RecognizeFirstPageOnly Then 
            results(i) = CompareFirstPage(FormSet(i).Attributes, form.Attributes) 
         Else 
            results(i) = CompareForm(FormSet(i).Attributes, form.Attributes) 
         End If 
         i += 1 
      Loop 
      Dim index As Integer = IdentifyForm(results) 
      If index >= 0 Then 
         form.Master = FormSet(index) 
         form.Result = results(index) 
      Else 
         form.Master = Nothing 
         form.Result = Nothing 
      End If 
   End Sub 
   '=========================================================================== 
   '''This method load images for processing purposes if needed. 
   Private Function LoadProcessingImage(ByVal form As MyForm) As Boolean 
      Dim count As Integer = form.Master.Properties.Pages - form.Image.PageCount 
      If count = 0 Then 
         Return False 
      End If 
 
      Dim image As RasterImage 
 
      If LoadFromScanner Then 
         image = LoadImageScanner(count) 
      Else 
         image = LoadImageFile(form.FileName, form.Image.PageCount + 1, form.Image.PageCount + count) 
      End If 
 
      form.Image.AddPages(image, 1, count) 
      Return True 
   End Function 
   '=========================================================================== 
   '''This method calculates the alignment for the recognized form. 
   Public Sub AlignForm(ByVal form As MyForm, ByVal calculateAlignment As Boolean) 
      If calculateAlignment Then 
         CreateFormForRecognition(form) 
         form.Alignment = RecognitionEngine.GetFormAlignment(form.Master.Attributes, form.Attributes, Nothing) 
      Else 
         form.Alignment = New List(Of PageAlignment)() 
         Dim i As Integer = 0 
         Do While i < form.Result.PageResults.Count 
            form.Alignment.Add(form.Result.PageResults(i).Alignment) 
            i += 1 
         Loop 
      End If 
   End Sub 
 
   '=========================================================================== 
   '''This method processes the recognized form. 
   Public Sub ProcessForm(ByVal form As MyForm) 
      form.ProcessingPages = form.Master.ProcessingPages 
      ProcessingEngine.Pages.Clear() 
      ProcessingEngine.Pages.AddRange(form.ProcessingPages) 
      ProcessingEngine.Process(form.Image, form.Alignment) 
   End Sub 
 
   '=========================================================================== 
   '''This method recognizes the unknown form then processes it. 
   Public Sub RunFormRecognitionAndProcessing(ByVal form As MyForm) 
      form.Image = LoadRecognitionImage(form.FileName) 
      RecognizeForm(form) 
      If form.Master Is Nothing Then 
         Console.WriteLine("Unknown form") 
         Return 
      End If 
      Dim calculateAlignment As Boolean = LoadProcessingImage(form) 
      AlignForm(form, calculateAlignment) 
      ProcessForm(form) 
      PrintOutResults(form) 
   End Sub 
 
   '=========================================================================== 
   ''' This method loads the specified master form attributes, fields, and image. 
   Public Function LoadMasterForm(ByVal attributesFileName As String, ByVal fieldsFileName As String, ByVal imageFileName As String) As MasterForm 
      Dim formData As Byte() 
      Dim form As MasterForm = New MasterForm() 
      formData = File.ReadAllBytes(attributesFileName) 
      form.Attributes = New FormRecognitionAttributes 
      form.Attributes.SetData(formData) 
      form.Properties = RecognitionEngine.GetFormProperties(form.Attributes) 
      ProcessingEngine.LoadFields(fieldsFileName) 
      form.ProcessingPages = New FormPages() 
      form.ProcessingPages.AddRange(ProcessingEngine.Pages) 
      form.Image = FormsCodec.Load(imageFileName, 1, CodecsLoadByteOrder.Bgr, 1, -1) 
      Return form 
   End Function 
 
   ''' This method saves master form attributes to the specified filename. 
   Public Sub SaveMasterFormAttributes(ByVal form As MasterForm, ByVal attributesFileName As String) 
      Dim formData As Byte() = form.Attributes.GetData() 
      File.WriteAllBytes(attributesFileName, formData) 
   End Sub 
 
   ''' This method saves master form fields to the specified filename. 
   Public Sub SaveMasterFormFields(ByVal form As MasterForm, ByVal fieldsFileName As String) 
      ProcessingEngine.Pages.Clear() 
      ProcessingEngine.Pages.AddRange(form.ProcessingPages) 
      ProcessingEngine.SaveFields(fieldsFileName) 
   End Sub 
 
   '=========================================================================== 
   ''' This method builds the Form Set. 
   Public Sub BuildFormsSet() 
      FormSet = New List(Of MasterForm)() 
 
      Dim Form107 As MasterForm = LoadMasterForm(Path.Combine(LEAD_VARS.ImagesDir, "Forms\MasterForm Sets\OCR\FFC-107.bin"), 
                                                 Path.Combine(LEAD_VARS.ImagesDir, "Forms\MasterForm Sets\OCR\FFC-107.xml"), 
                                                 Path.Combine(LEAD_VARS.ImagesDir, "Forms\MasterForm Sets\OCR\FFC-107.tif")) 
      FormSet.Add(Form107) 
 
      Dim Form180 As MasterForm = LoadMasterForm(Path.Combine(LEAD_VARS.ImagesDir, "Forms\MasterForm Sets\OCR\SF-180.bin"), 
                                                 Path.Combine(LEAD_VARS.ImagesDir, "Forms\MasterForm Sets\OCR\SF-180.xml"), 
                                                 Path.Combine(LEAD_VARS.ImagesDir, "Forms\MasterForm Sets\OCR\SF-180.tif")) 
      FormSet.Add(Form180) 
 
      Dim Form3881 As MasterForm = LoadMasterForm(Path.Combine(LEAD_VARS.ImagesDir, "Forms\MasterForm Sets\OCR\SF-3881.bin"), 
                                                  Path.Combine(LEAD_VARS.ImagesDir, "Forms\MasterForm Sets\OCR\SF-3881.xml"), 
                                                  Path.Combine(LEAD_VARS.ImagesDir, "Forms\MasterForm Sets\OCR\SF-3881.tif")) 
      FormSet.Add(Form3881) 
   End Sub 
 
   '=========================================================================== 
 
   ''' This method prints the image field results 
   Public Sub PrintOutImage(ByVal field As ImageFormField) 
      Console.WriteLine() 
      Console.WriteLine("*** Image Field") 
      Dim result As ImageFormFieldResult = TryCast(field.Result, ImageFormFieldResult) 
      FormsCodec.Save(result.Image, Path.Combine(LEAD_VARS.ImagesDir, "ImageField.tif"), RasterImageFormat.Tif, 1) 
   End Sub 
 
   ''' This method prints the barcode field results 
   Public Sub PrintOutBarcode(ByVal field As BarcodeFormField) 
      Console.WriteLine() 
      Console.WriteLine("*** Barcode Field") 
      Dim result As BarcodeFormFieldResult = TryCast(field.Result, BarcodeFormFieldResult) 
      For Each barcode As BarcodeFieldData In result.BarcodeData 
         Console.WriteLine(" *- Barcode Type: " & barcode.Symbology) 
         Console.WriteLine("  - Barcode Data: ") 
         Console.WriteLine(barcode.GetData()) 
      Next barcode 
   End Sub 
 
   ''' This method prints the Omr field results 
   Public Sub PrintOutOmr(ByVal field As OmrFormField) 
      Console.WriteLine() 
      Console.WriteLine("*** Omr Field") 
      Dim result As OmrFormFieldResult = TryCast(field.Result, OmrFormFieldResult) 
      Console.WriteLine(" Is Checked?[1 = yes, 0 = No] = " & result.Text) 
   End Sub 
 
   ''' This method prints the text field results 
   Public Sub PrintOutText(ByVal field As TextFormField) 
      Console.WriteLine() 
      Console.WriteLine("*** Text Field") 
      Dim result As TextFormFieldResult = TryCast(field.Result, TextFormFieldResult) 
      Console.WriteLine("Text: " & result.Text) 
   End Sub 
 
   ''' This method prints the processing results 
   Public Sub PrintOutResults(ByVal form As MyForm) 
      If form.Result.Confidence < 30 Then 
         Console.WriteLine("No match found") 
         Return 
      End If 
      Console.WriteLine("=======================Type===========================") 
      Console.WriteLine("Form Type is: " & form.Master.Properties.Name) 
      Console.WriteLine("Confidence = " & form.Result.Confidence) 
      Console.WriteLine() 
      Console.WriteLine("====================Fields Result=======================") 
 
      Dim pageIndex As Integer = 0 
      Do While pageIndex < form.ProcessingPages.Count 
         Console.WriteLine() 
         Console.WriteLine("=====Page # " & form.ProcessingPages(pageIndex).PageNumber & "=====") 
         Console.WriteLine() 
         For Each field As FormField In form.ProcessingPages(pageIndex) 
            If TypeOf field Is ImageFormField Then 
               PrintOutImage(TryCast(field, ImageFormField)) 
            ElseIf TypeOf field Is BarcodeFormField Then 
               PrintOutBarcode(TryCast(field, BarcodeFormField)) 
            ElseIf TypeOf field Is OmrFormField Then 
               PrintOutOmr(TryCast(field, OmrFormField)) 
            ElseIf TypeOf field Is TextFormField Then 
               PrintOutText(TryCast(field, TextFormField)) 
            End If 
         Next field 
         pageIndex += 1 
      Loop 
   End Sub 
 
   '=========================================================================== 
   '''This method starts up the Ocr engine 
   Private Sub StartUpOcrEngine() 
      Try 
         FormsOcrEngine = OcrEngineManager.CreateEngine(OcrEngineType.Advantage, False) 
         FormsOcrEngine.Startup(FormsCodec, Nothing, Nothing, LEAD_VARS.OcrAdvantageRuntimeDir) 
      Catch exp As Exception 
         Console.WriteLine(exp.Message) 
      End Try 
   End Sub 
 
   '''This method shuts down the Ocr engine 
   Private Sub ShutDownOcrEngine() 
      FormsOcrEngine.Shutdown() 
      FormsOcrEngine.Dispose() 
   End Sub 
 
   '''This method starts up the Barcode engine 
   Private Sub StartUpBarcodeEngine() 
      Try 
         FormsBarcodeEngine = New BarcodeEngine() 
      Catch exp As Exception 
         Console.WriteLine(exp.Message) 
      End Try 
   End Sub 
 
   '''This method sets the object managers. 
   Public Sub SetObjectManagers(ByVal enableDefault As Boolean, ByVal enableOcr As Boolean, ByVal enableBarcode As Boolean) 
      If RecognitionEngine Is Nothing Then 
         Return 
      End If 
      If enableDefault Then 
         Dim defaultObjectManager As DefaultObjectsManager = New DefaultObjectsManager() 
         RecognitionEngine.ObjectsManagers.Add(defaultObjectManager) 
      End If 
 
      If enableOcr Then 
         Dim ocrObejectManager As OcrObjectsManager = New OcrObjectsManager(FormsOcrEngine) 
         ocrObejectManager.Engine = FormsOcrEngine 
         RecognitionEngine.ObjectsManagers.Add(ocrObejectManager) 
      End If 
 
      If enableBarcode Then 
         Dim barcodeObjectManager As BarcodeObjectsManager = New BarcodeObjectsManager(FormsBarcodeEngine) 
         barcodeObjectManager.Engine = FormsBarcodeEngine 
         RecognitionEngine.ObjectsManagers.Add(barcodeObjectManager) 
      End If 
   End Sub 
 
   '''This method starts up all engines. 
   Public Sub StartUpEngines() 
      StartUpOcrEngine() 
      StartUpBarcodeEngine() 
      RecognitionEngine = New FormRecognitionEngine() 
      SetObjectManagers(False, True, False) 
      ProcessingEngine = New FormProcessingEngine() 
      ProcessingEngine.OcrEngine = FormsOcrEngine 
      ProcessingEngine.BarcodeEngine = FormsBarcodeEngine 
   End Sub 
 
   '''This method shuts down all engines. 
   Public Sub ShutDownEngines() 
      ShutDownOcrEngine() 
   End Sub 
 
   '''This method starts up RasterCodec, initialize Form Codec, starts up all engines, and Build the form set 
   Public Sub StartUp() 
      '1) 
      FormsCodec = New RasterCodecs() 
      '2) 
      StartUpEngines() 
      '3) 
      RecognizeFirstPageOnly = False 
      LoadFromScanner = False 
      '4) 
      BuildFormsSet() 
   End Sub 
 
   '''This method shuts down engines, dispose all images, and shut down the RasterCodec. 
   Public Sub ShutDown() 
      '1) 
      ShutDownEngines() 
      '2) 
      For Each master As MasterForm In FormSet 
         master.Image.Dispose() 
      Next master 
      '3) 
      FormsCodec.Dispose() 
   End Sub 
 
 
   ''' Here all starts for to recognize and process the unknown form that is stored in a fiel with passed fileName. 
   Public Sub Run(ByVal fileName As String) 
      Dim form As MyForm = New MyForm() 
      form.FileName = fileName 
      RunFormRecognitionAndProcessing(form) 
      If Not form.Image Is Nothing Then 
         form.Image.Dispose() 
      End If 
   End Sub 
End Class 
 
Friend Class Program 
   ''' The main entry to the program. 
   Shared Sub Main(ByVal args As String()) 
      If args Is Nothing OrElse args.Length = 0 Then 
         Return 
      End If 
 
      Dim test As TestForms = New TestForms() 
 
      test.StartUp() 
 
      test.Run(args(0)) 
 
      test.ShutDown() 
   End Sub 
End Class 
 
Public NotInheritable Class LEAD_VARS 
   Public Const ImagesDir As String = "C:\Users\Public\Documents\LEADTOOLS Images" 
   Public Const OcrAdvantageRuntimeDir As String = "C:\LEADTOOLS 19\Bin\Common\OcrAdvantageRuntime" 
End Class