public string[] recognizeimage(string imagepath, string lang)
{
ShellExcutor sh = new ShellExcutor();
sh.cmdExcute("tesseract.exe", ShellExcutor.tesseractlocation, " \"" + imagepath + "\" \"" + imagepath.Substring(0, imagepath.LastIndexOf('\\')) + "\\output\" -l " + lang, imagepath.Substring(0, imagepath.LastIndexOf('\\')));
return(File.ReadAllLines(imagepath.Substring(0, imagepath.LastIndexOf('\\')) + "\\output.txt", Encoding.UTF8));
}
public void combineTessDatamethod()
{
if (!Directory.Exists(projectFolder + @"\Tessdata"))
{
Directory.CreateDirectory(projectFolder + @"\Tessdata");
}
ShellExcutor sh = new ShellExcutor();
if (File.Exists(projectFolder + @"\word-list"))
{
sh.cmdExcute("wordlist2dawg.exe", ShellExcutor.tesseractlocation, " word-list " + ShellExcutor.isolang + ".word-dawg " + ShellExcutor.isolang + ".unicharset ", projectFolder);
}
if (File.Exists(projectFolder + @"\unambig-dawg"))
{
sh.cmdExcute("wordlist2dawg.exe", ShellExcutor.tesseractlocation, " unambig-dawg " + ShellExcutor.isolang + ".unicharambigs " + ShellExcutor.isolang + ".unicharset ", projectFolder);
}
if (File.Exists(projectFolder + @"\freq-dawg"))
{
sh.cmdExcute("wordlist2dawg.exe", ShellExcutor.tesseractlocation, " freq-dawg " + ShellExcutor.isolang + ".freq-dawg " + ShellExcutor.isolang + ".unicharset ", projectFolder);
}
try
{
File.Copy(projectFolder + @"\inttemp", projectFolder + '\\' + ShellExcutor.isolang + @".inttemp", true);
File.Copy(projectFolder + @"\shapetable", projectFolder + '\\' + ShellExcutor.isolang + @".shapetable", true);
File.Copy(projectFolder + @"\normproto", projectFolder + '\\' + ShellExcutor.isolang + @".normproto", true);
File.Copy(projectFolder + @"\pffmtable", projectFolder + '\\' + ShellExcutor.isolang + @".pffmtable", true);
sh.cmdExcute("combine_tessdata.exe", ShellExcutor.tesseractlocation, " " + ShellExcutor.isolang + ".", projectFolder);
}
catch (IOException ex)
{
MessageBox.Show(ex.Message, "Error", MessageBoxButtons.OK, MessageBoxIcon.Error);
return;
}
try
{
File.Copy(projectFolder + '\\' + ShellExcutor.isolang + ".traineddata", projectFolder + @"\Tessdata\" + ShellExcutor.isolang + @".traineddata", true);
MessageBox.Show("Creation of Tessdata is Succesfull", "Completed Succesfully", MessageBoxButtons.OK, MessageBoxIcon.Information);
}
catch (IOException ex)
{
MessageBox.Show(ex.Message, "Error", MessageBoxButtons.OK, MessageBoxIcon.Error);
return;
}
}
public void TrainTesseract()
{
ShellExcutor sh = new ShellExcutor();
foreach (var item in images)
{
if (File.Exists(projectFolder + '\\' + item.ToString().Substring(0, item.ToString().LastIndexOf('.')) + ".box"))
{
sh.cmdExcute("tesseract.exe", ShellExcutor.tesseractlocation, " " + item + " " + item.ToString().Substring(0, item.ToString().LastIndexOf('.')) + " nobatch box.train", projectFolder);
}
else
{
MessageBox.Show("Box File is Missing", "Error Cannot continue excution");
return;
}
}
string[] files = Directory.GetFiles(projectFolder);
//Compute the Character Set
string boxfilesingleline = null;
string trfilessigleline = null;
foreach (string item in files)
{
if (item.EndsWith(".box"))
{
boxfilesingleline += " " + item.Substring(item.LastIndexOf('\\') + 1) + " ";
}
if (item.EndsWith(".tr"))
{
trfilessigleline += " " + item.Substring(item.LastIndexOf('\\') + 1) + " ";
}
}
sh.cmdExcute("unicharset_extractor.exe", ShellExcutor.tesseractlocation, boxfilesingleline, projectFolder);
//Clustering
sh.cmdExcute("shapeclustering.exe", ShellExcutor.tesseractlocation, " -F font_properties -U unicharset " + trfilessigleline, projectFolder);
sh.cmdExcute("mftraining.exe", ShellExcutor.tesseractlocation, " -F font_properties -U unicharset -O " + ShellExcutor.isolang + ".unicharset " + trfilessigleline, projectFolder);
sh.cmdExcute("cntraining.exe", ShellExcutor.tesseractlocation, trfilessigleline, projectFolder);
}
public string[] recognizeimage(string imagepath,string lang)
{
ShellExcutor sh = new ShellExcutor();
sh.cmdExcute("tesseract.exe",ShellExcutor.tesseractlocation," \""+imagepath+"\" \""+imagepath.Substring(0,imagepath.LastIndexOf('\\'))+"\\output\" -l "+lang,imagepath.Substring(0,imagepath.LastIndexOf('\\')));
return (File.ReadAllLines(imagepath.Substring(0,imagepath.LastIndexOf('\\'))+"\\output.txt",Encoding.UTF8));
}
public void combineTessDatamethod()
{
if (!Directory.Exists(projectFolder + @"\Tessdata"))
{
Directory.CreateDirectory(projectFolder + @"\Tessdata");
}
ShellExcutor sh = new ShellExcutor();
if (File.Exists(projectFolder+@"\word-list"))
{
sh.cmdExcute("wordlist2dawg.exe", ShellExcutor.tesseractlocation, " word-list " + ShellExcutor.isolang + ".word-dawg " + ShellExcutor.isolang + ".unicharset ", projectFolder);
}
if (File.Exists(projectFolder + @"\unambig-dawg"))
{
sh.cmdExcute("wordlist2dawg.exe", ShellExcutor.tesseractlocation, " unambig-dawg " + ShellExcutor.isolang + ".unicharambigs " + ShellExcutor.isolang + ".unicharset ", projectFolder);
}
if (File.Exists(projectFolder + @"\freq-dawg"))
{
sh.cmdExcute("wordlist2dawg.exe", ShellExcutor.tesseractlocation, " freq-dawg " + ShellExcutor.isolang + ".freq-dawg " + ShellExcutor.isolang + ".unicharset ", projectFolder);
}
try
{
File.Copy(projectFolder + @"\inttemp", projectFolder + '\\' + ShellExcutor.isolang + @".inttemp",true);
File.Copy(projectFolder + @"\shapetable", projectFolder + '\\' + ShellExcutor.isolang + @".shapetable",true);
File.Copy(projectFolder + @"\normproto", projectFolder + '\\' + ShellExcutor.isolang + @".normproto",true);
File.Copy(projectFolder + @"\pffmtable", projectFolder + '\\' + ShellExcutor.isolang + @".pffmtable",true);
sh.cmdExcute("combine_tessdata.exe", ShellExcutor.tesseractlocation, " " + ShellExcutor.isolang + ".",projectFolder);
}
catch (IOException ex)
{
MessageBox.Show(ex.Message, "Error", MessageBoxButtons.OK, MessageBoxIcon.Error);
return;
}
try
{
File.Copy(projectFolder + '\\' + ShellExcutor.isolang + ".traineddata", projectFolder + @"\Tessdata\" + ShellExcutor.isolang + @".traineddata",true);
MessageBox.Show("Creation of Tessdata is Succesfull", "Completed Succesfully", MessageBoxButtons.OK, MessageBoxIcon.Information);
}
catch (IOException ex)
{
MessageBox.Show(ex.Message, "Error", MessageBoxButtons.OK, MessageBoxIcon.Error);
return;
}
}
public void TrainTesseract()
{
ShellExcutor sh = new ShellExcutor();
foreach (var item in images)
{
if (File.Exists(projectFolder+'\\'+item.ToString().Substring(0,item.ToString().LastIndexOf('.'))+".box"))
{
sh.cmdExcute("tesseract.exe", ShellExcutor.tesseractlocation," "+item+ " " +item.ToString().Substring(0,item.ToString().LastIndexOf('.')) + " nobatch box.train", projectFolder);
}
else
{
MessageBox.Show("Box File is Missing", "Error Cannot continue excution");
return;
}
}
string[] files = Directory.GetFiles(projectFolder);
//Compute the Character Set
string boxfilesingleline = null ;
string trfilessigleline = null;
foreach (string item in files)
{
if (item.EndsWith(".box"))
{
boxfilesingleline += " " + item.Substring(item.LastIndexOf('\\')+1) + " ";
}
if (item.EndsWith(".tr"))
{
trfilessigleline += " " + item.Substring(item.LastIndexOf('\\') + 1) + " ";
}
}
sh.cmdExcute("unicharset_extractor.exe", ShellExcutor.tesseractlocation, boxfilesingleline, projectFolder);
//Clustering
sh.cmdExcute("shapeclustering.exe", ShellExcutor.tesseractlocation, " -F font_properties -U unicharset " + trfilessigleline, projectFolder);
sh.cmdExcute("mftraining.exe", ShellExcutor.tesseractlocation, " -F font_properties -U unicharset -O " + ShellExcutor.isolang + ".unicharset " + trfilessigleline, projectFolder);
sh.cmdExcute("cntraining.exe", ShellExcutor.tesseractlocation, trfilessigleline, projectFolder);
}
