public void combineTessDatamethod()
{
if (!Directory.Exists(projectFolder + @"\Tessdata"))
{
Directory.CreateDirectory(projectFolder + @"\Tessdata");
}
if (File.Exists(projectFolder + @"\word-list"))
{
TesseractExecutor.cmdExcute("wordlist2dawg.exe", " word-list " + Configuration.IsoLang + ".word-dawg " + Configuration.IsoLang + ".unicharset ", projectFolder);
}
if (File.Exists(projectFolder + @"\unambig-dawg"))
{
TesseractExecutor.cmdExcute("wordlist2dawg.exe", " unambig-dawg " + Configuration.IsoLang + ".unicharambigs " + Configuration.IsoLang + ".unicharset ", projectFolder);
}
if (File.Exists(projectFolder + @"\freq-dawg"))
{
TesseractExecutor.cmdExcute("wordlist2dawg.exe", " freq-dawg " + Configuration.IsoLang + ".freq-dawg " + Configuration.IsoLang + ".unicharset ", projectFolder);
}
try
{
File.Copy(projectFolder + @"\inttemp", projectFolder + '\\' + Configuration.IsoLang + @".inttemp", true);
File.Copy(projectFolder + @"\shapetable", projectFolder + '\\' + Configuration.IsoLang + @".shapetable", true);
File.Copy(projectFolder + @"\normproto", projectFolder + '\\' + Configuration.IsoLang + @".normproto", true);
File.Copy(projectFolder + @"\pffmtable", projectFolder + '\\' + Configuration.IsoLang + @".pffmtable", true);
TesseractExecutor.cmdExcute("combine_tessdata.exe", " " + Configuration.IsoLang + ".", projectFolder);
}
catch (IOException ex)
{
MessageBox.Show(ex.Message, "Error", MessageBoxButtons.OK, MessageBoxIcon.Error);
return;
}
try
{
File.Copy(projectFolder + '\\' + Configuration.IsoLang + ".traineddata", projectFolder + @"\Tessdata\" + Configuration.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()
{
foreach (var item in images)
{
if (File.Exists(projectFolder + '\\' + item.ToString().Substring(0, item.ToString().LastIndexOf('.')) + ".box"))
{
TesseractExecutor.cmdExcute(Configuration.TesseractName, " " + 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) + " ";
}
}
TesseractExecutor.cmdExcute("unicharset_extractor.exe", boxfilesingleline, projectFolder);
//Clustering
TesseractExecutor.cmdExcute("shapeclustering.exe", " -F font_properties -U unicharset " + trfilessigleline, projectFolder);
TesseractExecutor.cmdExcute("mftraining.exe", " -F font_properties -U unicharset -O " + Configuration.IsoLang + ".unicharset " + trfilessigleline, projectFolder);
TesseractExecutor.cmdExcute("cntraining.exe", trfilessigleline, projectFolder);
}
private void bootstrapnewchar(object sender, EventArgs e)
{
TesseractExecutor.cmdExcute(Configuration.TesseractName, " " + filename + " " + filename.Substring(0, filename.LastIndexOf('.')) + " -l " + Configuration.IsoLang + " batch.nochop makebox", projectFolder);
}
private void createnewbox(object sender, EventArgs e)
{
TesseractExecutor.cmdExcute(Configuration.TesseractName, " " + filename + " " + filename.Substring(0, filename.LastIndexOf('.')) + " batch.nochop makebox", projectFolder);
}
public string[] recognizeimage(string imagepath, string lang)
{
TesseractExecutor.cmdExcute(Configuration.TesseractName, " \"" + 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));
}
