static void Main(string[] args)
{
String fileToConvert = @"FILE PATH HERE";
RasterSupport.SetLicense(@"C:\LEADTOOLS 20\Common\License\LEADTOOLS.LIC", System.IO.File.ReadAllText(@"C:\LEADTOOLS 20\Common\License\LEADTOOLS.LIC.KEY"));
using (RasterCodecs codecs = new RasterCodecs())
{
using (IOcrEngine ocrEngine = OcrEngineManager.CreateEngine(OcrEngineType.LEAD, false))
{
ocrEngine.Startup(null, null, null, @"C:\LEADTOOLS 20\Bin\Common\OcrLEADRuntime");
using (IOcrPage ocrPage = ocrEngine.CreatePage(ocrEngine.RasterCodecsInstance.Load(fileToConvert, 1), OcrImageSharingMode.AutoDispose))
{
ocrPage.AutoZone(null);
ocrPage.Recognize(null);
string recognizedCharacters = ocrPage.GetText(-1);
BarcodeEngine engine = new BarcodeEngine();
int resolution = 300;
using (RasterImage image = RasterImage.Create((int)(8.5 * resolution), (int)(11.0 * resolution), 1, resolution, RasterColor.FromKnownColor(RasterKnownColor.White)))
{
BarcodeWriter writer = engine.Writer;
QRBarcodeData data = BarcodeData.CreateDefaultBarcodeData(BarcodeSymbology.QR) as QRBarcodeData;
data.Bounds = new LeadRect(0, 0, image.ImageWidth, image.ImageHeight);
QRBarcodeWriteOptions writeOptions = writer.GetDefaultOptions(data.Symbology) as QRBarcodeWriteOptions;
writeOptions.XModule = 30;
writeOptions.HorizontalAlignment = BarcodeAlignment.Near;
writeOptions.VerticalAlignment = BarcodeAlignment.Near;
data.Value = recognizedCharacters;
writer.CalculateBarcodeDataBounds(new LeadRect(0, 0, image.ImageWidth, image.ImageHeight), image.XResolution, image.YResolution, data, writeOptions);
Console.WriteLine("{0} by {1} pixels", data.Bounds.Width, data.Bounds.Height);
writer.WriteBarcode(image, data, writeOptions);
CropCommand cmd = new CropCommand(new LeadRect(0, 0, data.Bounds.Width, data.Bounds.Height));
cmd.Run(image);
codecs.Save(image, "QR.tif", RasterImageFormat.CcittGroup4, 1);
if (RuntimeInformation.IsOSPlatform(OSPlatform.Windows))
{
var process = new Process();
process.StartInfo = new ProcessStartInfo("QR.tif")
{
UseShellExecute = true
};
process.Start();
}
Console.WriteLine();
}
}
}
}
}
private void _btnRecognize_Click(object sender, EventArgs e)
{
try
{
using (WaitCursor wait = new WaitCursor())
{
DocumentFormat format = _documentFormatSelector.SelectedFormat;
String documentFileName = _lblImageFileName.Text;
documentFileName = string.Concat(documentFileName, ".", DocumentWriter.GetFormatFileExtension(format));
_ocrPage.Recognize(null);
using (IOcrDocument _document = _ocrEngine.DocumentManager.CreateDocument(null, OcrCreateDocumentOptions.AutoDeleteFile))
{
_document.Pages.Add(_ocrPage);
_document.Save(documentFileName, format, null);
}
// Engine will correct the page deskew before AutoZone or Recognize, so we need to load the page again form the engine.
_viewer.Image = _ocrPage.GetRasterImage();
_viewer.Refresh();
UpdateMyControls();
// if the "View Final Document" option is checked then no need to show this message since
// it will load the saved document file.
if (!_cbViewFinalDocument.Checked)
{
Messager.ShowInformation(this, String.Format("The output document file was saved at ({0})", documentFileName));
}
else
{
if (File.Exists(documentFileName))
{
try
{
Process.Start(documentFileName);
}
catch
{
Messager.ShowError(this, "Unable to open generated results file with external viewing application");
}
}
else
{
Messager.ShowError(this, "Unable to open generated results file with external viewing application.\nThe system cannot find the file specified");
}
}
}
}
catch (Exception ex)
{
Messager.ShowError(this, ex.Message);
}
}
private IOcrPage DoublePass(RasterImage image)
{
//first pass with default settings
IOcrPage page = _ocrEngine.CreatePage(image.Clone(), OcrImageSharingMode.AutoDispose);
page.Recognize(null);
//second pass with mobile image processing set to true
_ocrEngine.SettingManager.SetBooleanValue("Recognition.Preprocess.MobileImagePreprocess", true);
IOcrPage mobilePage = _ocrEngine.CreatePage(image.Clone(), OcrImageSharingMode.AutoDispose);
mobilePage.Recognize(null);
//get the confidence of both pages
PageResults firstPassResults = GetPageConfidence(page);
PageResults secondPassResults = GetPageConfidence(mobilePage);
double confidenceDif = firstPassResults.Confidence - secondPassResults.Confidence;
IOcrPage highestConfidence;
PageResults pageResultsHighest;
if (confidenceDif > 2)
{
highestConfidence = page;
pageResultsHighest = firstPassResults;
}
else
{
highestConfidence = mobilePage;
pageResultsHighest = secondPassResults;
}
if (pageResultsHighest.TotalWords < 20)
{
IOcrPage thirdPass = highestConfidence.Copy();
thirdPass.Unrecognize();
OcrZone singleZone = new OcrZone()
{
Bounds = new LeadRect(0, 0, image.Width, image.Height)
};
thirdPass.Zones.Add(singleZone);
thirdPass.Recognize(null);
PageResults thirdResults = GetPageConfidence(thirdPass);
double confidencetDifThird = thirdResults.Confidence - pageResultsHighest.Confidence;
if (confidenceDif > 5)
{
highestConfidence = thirdPass;
pageResultsHighest = thirdResults;
}
}
return(highestConfidence);
}
void _rubberBand_RubberBandCompleted(object sender, ImageViewerRubberBandEventArgs e)
{
if (_ocrPage == null)
{
return;
}
try
{
_tsMainZoomComboBox.Enabled = false;
using (WaitCursor cursor = new WaitCursor())
{
if (_viewer.Image != null)
{
_currentHighlightRect = _viewer.ConvertRect(
null,
ImageViewerCoordinateType.Control,
ImageViewerCoordinateType.Image,
LeadRect.FromLTRB(e.Points[0].X, e.Points[0].Y, e.Points[1].X, e.Points[1].Y));
if (_currentHighlightRect.Width > 2 && _currentHighlightRect.Height > 2)
{
OcrZone zone = new OcrZone();
zone.Bounds = _currentHighlightRect;
zone.ZoneType = OcrZoneType.Text;
zone.CharacterFilters = OcrZoneCharacterFilters.None;
_ocrPage.Zones.Clear();
_ocrPage.Zones.Add(zone);
_ocrPage.Recognize(null);
_recognitionResults.Text = _ocrPage.GetText(0);
if (_recognitionResults.Text == "\n" || _recognitionResults.Text == "")
{
Messager.ShowInformation(this, "No text was recognized.");
}
}
}
}
}
catch (Exception ex)
{
Messager.ShowError(this, ex);
}
finally
{
_viewer.Invalidate();
_tsMainZoomComboBox.Enabled = true;
}
}
/// <summary>
/// Uses an instance of the LEADTools OCR Engine to "read" the text
/// in the pre-defined OCR region representing the mailing address
/// area. Is thread safe.
/// </summary>
/// <param name="ocrEngine"></param>
/// <param name="document"></param>
/// <returns></returns>
private static string GetAddressBlockText(IOcrEngine ocrEngine, PDFDocument document)
{
string returnedText = null;
using (var codecs = new RasterCodecs())
using (var image = new RasterImage(document.GetPageImage(codecs, 1)))
using (IOcrDocument ocrDocument = ocrEngine.DocumentManager.CreateDocument())
using (IOcrPage ocrPage = ocrDocument.Pages.AddPage(image, null))
{
var myZone = new OcrZone();
myZone.Bounds = new LogicalRectangle(0, 2, 4, 1.4, LogicalUnit.Inch);
ocrPage.Zones.Add(myZone);
ocrPage.Recognize(null);
returnedText = ocrPage.GetText(0).ToUpper();
}
return(returnedText);
}
public static void exe_ocr(string filename, IOcrDocument ocrDocument)
{
Console.WriteLine(filename);
RasterCodecs rasterCodecs = null;
RasterImage rasterImage = null;
try
{
rasterCodecs = new RasterCodecs();
rasterCodecs.ThrowExceptionsOnInvalidImages = false;
rasterImage = rasterCodecs.Load(filename);
AutoBinarizeCommand command = new AutoBinarizeCommand();
command.Run(rasterImage);
//AutoLineRemoveCommand commandLine = new AutoLineRemoveCommand();
//commandLine.Run(rasterImage);
IOcrPage page = ocrDocument.Pages.AddPage(rasterImage, null);
if (page != null)
{
page.UpdateNativeFillMethod();
page.Recognize(null);
}
// ocrDocument.Pages.zon(NativeOcrZoneType.AutoGraphic);
}catch (Exception e)
{
// Console.WriteLine("add image faild: " + e);
}
finally
{
if (rasterCodecs != null)
{
rasterCodecs.Dispose();
}
if (rasterImage != null)
{
rasterImage.Dispose();
}
}
}
private void GetOmrReading(IOcrPage ocrPage, FormField field, ImageField imageField, int retry = 1)
{
IOcrPageCharacters pageCharacters = ocrPage.GetRecognizedCharacters();
if (pageCharacters == null)
{
logger.Warn($"could not read OMR for ${field} ");
imageField.FieldResult.Confidence = 0;
imageField.FieldResult.Text = "";
}
else
{
IOcrZoneCharacters zoneCharacters = pageCharacters[0];
if (zoneCharacters.Count > 0)
{
OcrCharacter omrCharacter = zoneCharacters[0];
imageField.FieldResult.Text = omrCharacter.Code.ToString();
imageField.FieldResult.IsFilled = omrCharacter.Code == FilledChar;
imageField.FieldResult.Confidence = omrCharacter.Confidence;
// often on a fill we get the line from the box, so we retry more narrowly
if (imageField.FieldResult.IsFilled)
{
if (retry > 0)
{
var orgZone = ocrPage.Zones[0];
orgZone.Bounds = ChangeBoundsRatio(orgZone.Bounds, 0.66);
ocrPage.Recognize(null);
GetOmrReading(ocrPage, field, imageField, 0);
logger.Info($"FILLED {field.Name}");
}
}
}
else
{
imageField.FieldResult.Text = "";
}
}
}
public List <FilledForm> ProcessOcr(ResultsForPrettyJson formResults,
List <ImageInfo> fileInfos)
{
try
{
var outDir = formResults.OriginalDirectoryName;
var retForms = new List <FilledForm>();
var usedMasters = new HashSet <MasterForm>();
Stopwatch stopWatch = new Stopwatch();
stopWatch.Start();
formResults.PagesInPdf = fileInfos.Count;
foreach (var ofi in fileInfos)
{
FilledForm newForm = new FilledForm();
retForms.Add(newForm);
newForm.ImageInfoMaster.InitialImage = ofi;
newForm.Name = Path.GetFileNameWithoutExtension(ofi.ImageFileInfo.Name);
if (ofi.Image == null)
{
ofi.Image = LoadImageFile(ofi.ImageFileInfo.FullName, 1, -1);
}
//CleanupImage(ofi.Image);
var par = new FormThreadCallParams()
{
ImageInfo = ofi, StopWatch = stopWatch, Form = newForm
};
if (PageTimeoutInSeconds < 50)
{
Thread t = new Thread(this.PrepareNewFormThreader);
t.Start(par);
if (!t.Join(TimeSpan.FromSeconds(PageTimeoutInSeconds)))
{
t.Abort();
formResults.TimedOutPages.Add(newForm.Name);
formResults.BestFormConfidence.Add(-1);
if (formResults.TimedOutPages.Count > 2 && formResults.PagesMappedToForm == 0)
{
formResults.Status =
$"Form abandoned for timeout after {formResults.BestFormConfidence.Count} pages";
logger.Error(formResults.Status);
return(retForms);
}
continue;
}
}
else
{
PrepareNewFormThreader(par);
}
Debug.Assert(par.Attributes != null);
var filledFormAttributes = par.Attributes;
//List<FormRecognitionResult> results = new List<FormRecognitionResult>();
MasterForm currentMasterBlockForm = null;
int bestConfidence = -1;
int currentConfidence = 85;
foreach (var master in BlockMasterForms)
{
if (usedMasters.Contains(master))
{
continue;
}
var result = RecognitionEngine.CompareForm(master.Attributes, filledFormAttributes, null, null);
//logger.Debug($"Check {master} for {newForm} {stopWatch.ElapsedMilliseconds} {result.Confidence}");
if (result.Confidence > currentConfidence)
{
currentMasterBlockForm = master;
bestConfidence = currentConfidence = result.Confidence;
}
else if (result.Confidence > bestConfidence)
{
bestConfidence = result.Confidence;
}
}
formResults.BestFormConfidence.Add(bestConfidence);
if (currentMasterBlockForm != null)
{
formResults.MasterFormPages.Add(currentMasterBlockForm.Properties.Name);
formResults.PagesMappedToForm++;
logger.Info($"FilledForm matched {newForm.Name} {newForm.Status} {stopWatch.ElapsedMilliseconds} ");
newForm.ImageInfoMaster.InitialImage = ofi;
var centeredImage = ofi.Image.CloneAll();
CleanupImage(centeredImage);
newForm.ImageInfoMaster.CenteredImage = new ImageInfo()
{
Image = centeredImage
};
var omrImage = centeredImage.CloneAll();
PrepareOmrImage(omrImage);
newForm.ImageInfoMaster.OmrImage = new ImageInfo()
{
Image = omrImage
};
newForm.Status = "Matched";
newForm.Master = currentMasterBlockForm;
var alignment =
RecognitionEngine.GetFormAlignment(newForm.Master.Attributes, newForm.Attributes, null);
var fields = currentMasterBlockForm.ProcessingPages[0];
var scaler = currentMasterBlockForm.Resolution;
var fieldsOnlyImage = RasterImage.Create(centeredImage.Width, centeredImage.Height,
centeredImage.BitsPerPixel, 300, RasterColor.White);
//fieldsOnlyImage = new RasterImage(RasterMemoryFlags.Conventional, centeredImage.Width, centeredImage.Height, centeredInage.BitsPerPixel, RasterByteOrder.Rgb, RasterViewPerspective.TopLeft, null, null, 0);
var subDirField = Path.Combine(outDir, "fields");
var fileNameFieldOnly = Path.Combine(subDirField, newForm.Name + "_fields.jpg");
var googleResultsFile = Path.Combine(subDirField, newForm.Name + "_google.json");
var combined = false;
foreach (var field in fields)
{
var isBlock = field.Name.Contains("block");
var rect200 = alignment[0].AlignRectangle(field.Bounds);
scaler = 300;
int fudge = isBlock ? 30 : 1;
var rect300 = new LeadRect(rect200.Left * 300 / scaler - fudge, rect200.Top * 300 / scaler - fudge,
rect200.Width * 300 / scaler + fudge,
rect200.Height * 300 / scaler + fudge);
try
{
var imageInfoToUse = newForm.ImageInfoMaster.CenteredImage;
var zoneType = OcrZoneType.Text;
if (field.GetType() == typeof(OmrFormField))
{
imageInfoToUse = newForm.ImageInfoMaster.OmrImage;
zoneType = OcrZoneType.Omr;
}
else if (field.GetType() == typeof(ImageFormField))
{
zoneType = OcrZoneType.Graphic;
}
var image = imageInfoToUse.Image.CloneAll();
var subDir = Path.Combine(outDir, isBlock ? "blocks" : "fields");
var fileName = Path.Combine(subDir, newForm.Name + "_" + field.Name + ".jpg");
var imageField = new ImageField
{
Field = field,
FieldResult =
{
FieldName = field.Name,
IsBlock = isBlock,
ImageFile = fileName,
Bounds = rect300.ToString(),
FieldType = zoneType.ToString(),
Error = "None"
}
};
imageField.Rectangle = new Rectangle(rect300.X, rect300.Y, rect300.Width, rect300.Height);
try
{
EnsurePathExists(subDir);
CropCommand command = new CropCommand
{
Rectangle = rect300
};
command.Run(image);
RasterCodecs.Save(image, fileName, RasterImageFormat.Jpeg, bitsPerPixel: 8);
if (!isBlock && zoneType == OcrZoneType.Text && !combined)
{
try
{
;
var combiner = new CombineCommand();
//combiner.DestinationImage = fieldsOnlyImage;
combiner.SourceImage = image.Clone();
combiner.DestinationRectangle = rect300;
var regionBounds = image.GetRegionBounds(null);
combiner.SourcePoint = new LeadPoint(regionBounds.X, regionBounds.Y);
//combiner.Flags = CombineCommandFlags.OperationAdd | CombineCommandFlags.Destination0 | CombineCommandFlags.Source1 | CombineCommandFlags.Destination0 ;
combiner.Flags = CombineCommandFlags.OperationOr | CombineCommandFlags.Destination0;; // |CombineFastCommandFlags.OperationAverage;
combiner.Run(fieldsOnlyImage);
//combined = true;
}
catch (Exception exCombine)
{
logger.Error(exCombine, $"error combining field {field.Name} {rect300}");
}
}
var imageInfo = new ImageInfo()
{
Image = image, ImageFileInfo = new FileInfo(fileName)
};
imageField.ImageInfo = imageInfo;
if (!isBlock && zoneType != OcrZoneType.Graphic)
{
using (IOcrPage ocrPage = OcrEngine.CreatePage(image, OcrImageSharingMode.AutoDispose))
{
OcrZone ocrZone = new OcrZone
{
ZoneType = zoneType,
Bounds = new LeadRect(fudge, fudge, image.ImageSize.Width - fudge,
image.ImageSize.Height - fudge)
};
ocrPage.Zones.Add(ocrZone);
ocrPage.Recognize(null);
if (zoneType == OcrZoneType.Omr)
{
if (field.Name.Contains("C2NGVD1929"))
{
logger.Info(ocrZone.Bounds);
}
GetOmrReading(ocrPage, field, imageField);
}
else if (zoneType == OcrZoneType.Text)
{
var resultsPage = GetPageConfidence(ocrPage);
imageField.FieldResult.Confidence = resultsPage.Confidence;
char[] crlf = { '\r', '\n' };
imageField.FieldResult.Text = ocrPage.GetText(0).TrimEnd(crlf);
}
}
}
logger.Info(
$"field {field.Name} {rect300} [{imageField.FieldResult.Text}] confidence: {imageField.FieldResult.Confidence}");
}
catch (Exception exField)
{
logger.Error(exField, $"Error processing {field.Name}");
formResults.FieldsWithError++;
imageField.FieldResult.Error = exField.Message;
}
newForm.ImageFields.Add(imageField);
formResults.OcrFields.Add(imageField.FieldResult);
formResults.Status = "FormMatched";
}
catch (Exception ex)
{
logger.Error(ex, $"Error on field {field.Name} {rect300}");
newForm.Status = $"Error|Field {field.Name} {rect300}: [{ex.Message}]";
}
}
RasterCodecs.Save(PrepareOmrImage(fieldsOnlyImage), fileNameFieldOnly, RasterImageFormat.Jpeg, bitsPerPixel: 8);
var googleResults = GoogleOcr(fileNameFieldOnly);
if (googleResults.Count > 0)
{
var json = JsonConvert.SerializeObject(googleResults, Formatting.Indented);
File.WriteAllText(googleResultsFile, json);
MergeGoogleOcr(newForm, googleResults);
}
usedMasters.Add(currentMasterBlockForm);
}
else
{
newForm.Status = "Unmatched|No MasterForm match";
}
logger.Info($"FilledForm processed {newForm.Name} {newForm.Status} {stopWatch.ElapsedMilliseconds} ");
if (usedMasters.Count == BlockMasterForms.Count)
{
logger.Info("found all master forms");
break;
}
}
stopWatch.Stop();
return(retForms);
}
catch (Exception ex)
{
logger.Error(ex, "Untrapped error found");
return(null);
}
}
private void DoLoadAndRecognizeDocument(OcrProgressDialog dlg, Dictionary <string, object> args)
{
// Perform load and recognize here
OcrProgressCallback callback = dlg.OcrProgressCallback;
IOcrDocument ocrDocument = null;
try
{
string documentFileName = args["documentFileName"] as string;
ocrDocument = _ocrEngine.DocumentManager.CreateDocument("", OcrCreateDocumentOptions.InMemory);
IOcrPage ocrPage = null;
if (!dlg.IsCanceled)
{
// If we are not using a progress bar, update the description text
if (callback == null)
{
dlg.UpdateDescription("Loading the document (first page only)...");
}
ocrPage = ocrDocument.Pages.AddPage(documentFileName, callback);
}
if (!dlg.IsCanceled)
{
// If we are not using a progress bar, update the description text
if (callback == null)
{
dlg.UpdateDescription("Recognizing the page(s) of the document...");
}
ocrPage.Recognize(callback);
}
if (!dlg.IsCanceled)
{
// We did not cancel, use this document
SetDocument(ocrDocument, documentFileName);
ocrDocument = null;
}
}
catch (Exception ex)
{
ShowError(ex);
}
finally
{
if (callback == null)
{
dlg.EndOperation();
}
// Clean up
if (ocrDocument != null)
{
ocrDocument.Dispose();
}
}
}
