private string processImage(string imageInput, string outputFilePath)
{
var debugMessage = new StringBuilder();
try
{
debugMessage.Append("Process Image output file path " + outputFilePath);
var ocr = new Ocr();
var preProcessor = new PreProcessor();
debugMessage.Append("Objects created");
preProcessor.Deskew = true;
preProcessor.Autorotate = false;
preProcessor.RemoveLines = true;
preProcessor.Binarize = 96;
preProcessor.Morph = "c2.2";
ocr.License =
"MD1BZHZhbmNlZDsxPWtlZXJ0aSAtIGt2a2lydGh5QGdtYWlsLmNvbTsyPTUyNDY4Njg5OTE5MTMwMjg5NTI7Mz1rZWVydGkgLSBrdmtpcnRoeUBnbWFpbC5jb207ND05OTk7NT1UcnVlOzUuMT1GYWxzZTs3PTYzNTE4ODYwODAwMDAwMDAwMDs4PTQxRDA3NEFFODJFQjI3QjM3RDdGMTUzQ0REQjVEQkNFNEVGRjdGREU5MEIwOTg1MjkwQ0JDREFCQTM3MEFBNzU7OT0xLjQxLjAuMA";
ocr.ResourceFolder = @"C:\Aquaforest\OCRSDK\bin";
ocr.EnableConsoleOutput = true;
ocr.EnableTextOutput = true;
//ocr.ReadBMPSource(@"C:\Users\KotaruV\Documents\Visual Studio 2012\Projects\Playground\OCRConsoleApp\OCRConsoleApp\images\3.jpg");
//ocr.ReadTIFFSource(imageInput);
ocr.ReadBMPSource(imageInput);
debugMessage.Append("Read from bmp source. ");
ocr.Recognize(preProcessor);
debugMessage.Append("Recognize executed. ");
ocr.SaveTextOutput(outputFilePath, true);
var ocrText = getOcrTextFromFile(outputFilePath);
debugMessage.Append("Saved text output. ");
ocr.DeleteTemporaryFiles();
debugMessage.Append("Deleted temporary files. ");
return(ocrText);
}
finally
{
System.Diagnostics.EventLog.WriteEntry("Application", debugMessage.ToString(), System.Diagnostics.EventLogEntryType.Error);
}
}
private string processImage(string imageInput, string outputFilePath)
{
var debugMessage = new StringBuilder();
try
{
debugMessage.Append("Process Image output file path " + outputFilePath);
var ocr = new Ocr();
var preProcessor = new PreProcessor();
debugMessage.Append("Objects created");
preProcessor.Deskew = true;
preProcessor.Autorotate = false;
preProcessor.RemoveLines = true;
preProcessor.Binarize = 96;
preProcessor.Morph = "c2.2";
ocr.License =
"MD1BZHZhbmNlZDsxPWtlZXJ0aSAtIGt2a2lydGh5QGdtYWlsLmNvbTsyPTUyNDY4Njg5OTE5MTMwMjg5NTI7Mz1rZWVydGkgLSBrdmtpcnRoeUBnbWFpbC5jb207ND05OTk7NT1UcnVlOzUuMT1GYWxzZTs3PTYzNTE4ODYwODAwMDAwMDAwMDs4PTQxRDA3NEFFODJFQjI3QjM3RDdGMTUzQ0REQjVEQkNFNEVGRjdGREU5MEIwOTg1MjkwQ0JDREFCQTM3MEFBNzU7OT0xLjQxLjAuMA";
ocr.ResourceFolder = @"C:\Aquaforest\OCRSDK\bin";
ocr.EnableConsoleOutput = true;
ocr.EnableTextOutput = true;
//ocr.ReadBMPSource(@"C:\Users\KotaruV\Documents\Visual Studio 2012\Projects\Playground\OCRConsoleApp\OCRConsoleApp\images\3.jpg");
//ocr.ReadTIFFSource(imageInput);
ocr.ReadBMPSource(imageInput);
debugMessage.Append("Read from bmp source. ");
ocr.Recognize(preProcessor);
debugMessage.Append("Recognize executed. ");
ocr.SaveTextOutput(outputFilePath, true);
var ocrText = getOcrTextFromFile(outputFilePath);
debugMessage.Append("Saved text output. ");
ocr.DeleteTemporaryFiles();
debugMessage.Append("Deleted temporary files. ");
return ocrText;
}
finally
{
System.Diagnostics.EventLog.WriteEntry("Application", debugMessage.ToString(), System.Diagnostics.EventLogEntryType.Error);
}
}
private List <LineData> ProcessDocument(COB_ParseType stepId, string sourceFile, string extension)
{
List <LineData> lines = new List <LineData>();
using (Ocr ocr = new Ocr(GetConfigResource()))
{
try
{
var preProcessor = Configure(ocr);
switch (extension.ToLower())
{
case ".jpg":
case ".jpeg":
case ".png":
case ".bmp":
ocr.ReadJPEGSource(sourceFile);
break;
case ".tif":
case ".tiff":
ocr.ReadTIFFSource(sourceFile);
break;
case ".pdf":
//ocr.EnablePdfOutput = true;
//ocr.EnableTextOutput = false;
ocr.EndPage = 2; //process only 2 pages as currently nothing is being read from 3rd page
ocr.ReadPDFSource(sourceFile);
break;
}
if (stepId == COB_ParseType.TradeLicense)
{
preProcessor.ForceTableZones = true;
}
if (!ocr.Recognize(preProcessor))
{
if (ocr.LastException == null)
{
throw new Exception("Unable to parse document");
}
else
{
throw ocr.LastException;
}
}
for (int i = 1; i <= ocr.NumberPages && i <= ocr.EndPage; i++)
{
var pageLines = ocr.ReadPageLines(i);
if (pageLines?.Count > 0)
{
lines.AddRange(pageLines);
}
}
}
catch (Exception)
{
throw;
}
finally
{
try
{
File.Copy(ocr.LogFilePath, Path.Combine(logsPath, Path.GetFileNameWithoutExtension(sourceFile) + ".log"));
}
catch (Exception exp)
{
}
ocr.DeleteTemporaryFiles();
}
}
return(lines);
}
public string getValuesbyStepId(COB_ParseType stepId, string sourceFile, string extension)
{
string result = string.Empty;
using (Ocr ocr = new Ocr(GetConfigResource()))
{
try
{
var preProcessor = Configure(ocr);
if (stepId == COB_ParseType.EmiratesId)
{
using (Bitmap map = new Bitmap(sourceFile))
{
//using (var img = Helper.ConvertBlackWhiteImage(map))
{
var img = map;
//Helper.SetContrast(img, 80);
ocr.ReadImageSource(img);
if (ocr.Recognize(preProcessor))
{
result = this.getValuesByEid(ocr);
}
else
{
if (ocr.LastException == null)
{
throw new Exception("Unable to parse document");
}
else
{
throw ocr.LastException;
}
}
}
}
}
else
{
switch (extension.ToLower())
{
case ".jpg":
case ".jpeg":
case ".png":
case ".bmp":
ocr.ReadJPEGSource(sourceFile);
break;
case ".tif":
case ".tiff":
ocr.ReadTIFFSource(sourceFile);
break;
case ".pdf":
//ocr.EnablePdfOutput = true;
//ocr.EnableTextOutput = false;
ocr.EndPage = 2; //process only 2 pages as currently nothing is being read from 3rd page
ocr.ReadPDFSource(sourceFile);
break;
}
if (stepId == COB_ParseType.TradeLicense)
{
preProcessor.ForceTableZones = true;
}
if (!ocr.Recognize(preProcessor))
{
if (ocr.LastException == null)
{
throw new Exception("Unable to parse document");
}
else
{
throw ocr.LastException;
}
}
switch (stepId)
{
case COB_ParseType.TradeLicense:
var licenseLines = new List <LineData>();
for (int i = 1; i <= ocr.NumberPages && i <= ocr.EndPage; i++)
{
var pageLines = ocr.ReadPageLines(i);
if (pageLines?.Count > 0)
{
licenseLines.AddRange(pageLines);
}
}
// currently 'dubai' is hardcoded, required changes to fetch it from Application
result = TradeLicenseParserFactory.Parse(licenseLines, "DUBAIECONOMY", "LLC");
break;
case COB_ParseType.Passport:
StringBuilder build = new StringBuilder();
for (int pageNo = 0; pageNo < ocr.NumberPages; pageNo++)
{
var lines = ocr.ReadPageLines(pageNo + 1);
if (lines?.Count > 0)
{
int i = 0;
for (i = 0; i < lines.Count; i++)
{
Regex re = new Regex(@"(P|V|C)(<|«)(.*)");
if (re.IsMatch(lines[i].LineWords))
{
break;
}
}
int counti = 0;
while (i < lines.Count && counti < 2) //only two lines to be read as passport has only two line MRZ as per standard
{
string temp = lines[i].LineWords;
temp = temp.TrimEnd().Replace("«", "<<").Replace("&", "<").Replace("\n", "").Replace(" ", "");
temp = System.Text.RegularExpressions.Regex.Replace(temp, "[']+", "");
if (counti == 0)
{
temp = Regex.Replace(temp, ".*P<", "P<");
}
temp = temp.Substring(0, Math.Min(temp.Length, 44));
temp = temp.PadRight(44, '<');
build.Append(temp);
i++;
}
}
}
result = build.ToString();
MRZParser pareser = new MRZParser();
result = pareser.Parse(result);
break;
case COB_ParseType.EmiratesId:
result = getValuesByEid(ocr);
break;
default:
return(string.Empty);
}
}
}
catch (Exception exp)
{
throw;
}
finally{
try
{
File.Copy(ocr.LogFilePath, Path.Combine(logsPath, Path.GetFileNameWithoutExtension(sourceFile) + ".log"));
}
catch (Exception exp) {
}
ocr.DeleteTemporaryFiles();
}
}
return(result);
}
