private void CapacityMark(int newIdx)
{
int idx = _smallMarkList.Count;
for (int i = idx; i < newIdx; i++)
{
FeSmall small = new FeSmall();
small.Idx = i;
_smallMarkList.Add(i, small);
}
}
private void CapacityBox(int newSize)
{
int idx = _smallBoxList.Count;
for (int i = idx; i < newSize; i++)
{
FeSmall small = new FeSmall();
small.Idx = i;
_smallBoxList.Add(i, small);
}
}
private void ProcBoxDoubleClick(FeSmall box)
{
FeSmall mark;
if (_smallMarkList.TryGetValue(box.Idx, out mark))
{
string smallFileName = string.Format("{0}_{1}{2}", _fiName, box.Idx, _fiExt);
string smallFullName = Path.Combine(_fiPath, FeConst.DIR_TEMP_IMG, smallFileName);
mark.GetFromSmall(box);
mark.SaveImage(smallFullName);
}
}
public void GetFromSmall(FeSmall small)
{
_picBmp = small._picBmp.Clone(new Rectangle(0, 0, small._picBmp.Width, small._picBmp.Height), PixelFormat.DontCare);
using (MemoryStream ms = new MemoryStream())
{
_picBmp.Save(ms, ImageFormat.Tiff);
_pictureBox.Image = Image.FromStream(ms);
}
_textBox.Text = small._textBox.Text;
_isEmpty = false;
_value = small._value;
_left = small.Left;
_bottom = small._bottom;
_right = small._right;
_top = small._top;
}
public void PutBox(int smallIdx, string value, int left, int bottom, int right, int top)
{
if (smallIdx >= _smallBoxList.Count)
{
CapacityBox(smallIdx + 1);
}
FeSmall small = _smallBoxList[smallIdx];
small.IsEmpty = false;
small.Idx = smallIdx;
small.Value = value;
small.Left = left;
small.Bottom = bottom;
small.Right = right;
small.Top = top;
_boxCount++;
}
private void btnSaveMark_Click(object sender, EventArgs e)
{
StringBuilder sb = new StringBuilder();
foreach (var train in _trainList)
{
foreach (var item in train.SmallMarkList)
{
FeSmall mark = item.Value;
string line = mark.GetMarkLine(train.FileName);
if (line == "")
{
continue;
}
sb.AppendLine(line);
}
}
SaveFile(sb, FeConst.FILE_MARK);
}
public void PutMark(int smallIdx, string value, int left, int bottom, int right, int top)
{
if (smallIdx >= _smallMarkList.Count)
{
CapacityMark(smallIdx + 1);
}
FeSmall small = _smallMarkList[smallIdx];
small.IsEmpty = false;
small.Idx = smallIdx;
small.Value = value;
small.Left = left;
small.Bottom = bottom;
small.Right = right;
small.Top = top;
_markCount++;
if (_smallBoxList[smallIdx].Value == value)
{
small.IsSame = true;
_sameCount++;
_smallBoxList[smallIdx].IsSame = true;
}
}
private void ProcMarkDoubleClick(FeSmall mark)
{
mark.FullEmpty();
}
private void btnTrain_Click(object sender, EventArgs e)
{
/*
* mark.txt+tmpImg ==> ts.yzm.exp0.tif 合并
* box
* mark.txt 修正box
* unicharset_extractor ts.yzm.exp0.box
* shapeclustering -F font_properties -U unicharset ts.yzm.exp0.tr
* mftraining -F font_properties -U unicharset -O ts.unicharset ts.yzm.exp0.tr
* cntraining ts.yzm.exp0.tr
* rename normproto ts.normproto
* rename inttemp ts.inttemp
* rename pffmtable ts.pffmtable
* rename shapetable ts.shapetable
* combine_tessdata ts
* copy ts.traineddata E:\tesseract-ocr\tessdata
*/
string imagePath = tbTrainImagePath.Text;
string lang = tbLang.Text.Trim();
string font = tbFont.Text.Trim();
string tmpBoxPath = Path.Combine(imagePath, FeConst.DIR_TEMP_BOX);
string tmpImgPath = Path.Combine(imagePath, FeConst.DIR_TEMP_IMG);
string tmpTrainPath = Path.Combine(imagePath, FeConst.DIR_TEMP_TRAIN);
Dictionary <int, FeSmall> smallList = new Dictionary <int, FeSmall>();
List <string> fileList = new List <string>();
int idx = 0;
string markFile = Path.Combine(imagePath, FeConst.FILE_MARK);
using (FileStream fs = new FileStream(markFile, FileMode.OpenOrCreate))
{
StreamReader sr = new StreamReader(fs, Encoding.Default);
string lineStr = string.Empty;
while ((lineStr = sr.ReadLine()) != null)
{
string[] arr = lineStr.Split(','); //char left bottom right top filename smallidx
if (arr.Length != 7)
{
continue;
}
string fileName = arr[5];
string fiName = Path.GetFileNameWithoutExtension(fileName);
string fiExt = Path.GetExtension(fileName);
int smallIdx = int.Parse(arr[6]);
FeSmall small = new FeSmall();
small.Idx = idx;
small.FileName = string.Format("{0}_{1}{2}", fiName, smallIdx, fiExt);
small.Value = arr[0];
smallList.Add(idx, small);
fileList.Add(Path.Combine(tmpImgPath, small.FileName));
idx++;
}
}
//合并tiff
string mergeFile = string.Format("{0}.{1}.exp0.tif", lang, font);
MergeTiff(fileList, Path.Combine(tmpTrainPath, mergeFile));
//生成box
string args;
string trainFile = Path.Combine(_tessacertPath, "tessdata", lang + ".traineddata");
if (File.Exists(trainFile))
{
args = string.Format("{0}.{1}.exp0.tif {0}.{1}.exp0 -l {0} -psm 10 batch.nochop makebox", lang, font);
}
else
{
args = string.Format("{0}.{1}.exp0.tif {0}.{1}.exp0 -psm 10 batch.nochop makebox", lang, font);
}
ProcessBat(_tessacertExe, tmpTrainPath, args);
//mark.txt 修正 box
StringBuilder sb = new StringBuilder();
string boxFile = Path.Combine(tmpTrainPath, string.Format("{0}.{1}.exp0.box", lang, font));
using (FileStream fs = new FileStream(boxFile, FileMode.Open))
{
StreamReader sr = new StreamReader(fs, Encoding.Default);
string lineStr = string.Empty;
while ((lineStr = sr.ReadLine()) != null)
{
lineStr = lineStr.Trim();
string[] arr = lineStr.Split(' ');
if (arr.Length == 6)
{
int tmpIdx = int.Parse(arr[5]);
FeSmall small;
if (smallList.TryGetValue(tmpIdx, out small))
{
//修正
if (arr[0] != small.Value)
{
arr[0] = small.Value;
lineStr = string.Format("{0} {1} {2} {3} {4} {5}", arr[0], arr[1], arr[2], arr[3], arr[4], arr[5]);
}
}
}
sb.AppendLine(lineStr);
}
sr.Close();
}
SaveFile(sb, boxFile);
if (File.Exists(trainFile))
{
args = string.Format("{0}.{1}.exp0.tif {0}.{1}.exp0 -l {0} -psm 10 nobatch box.train", lang, font);
}
else
{
args = string.Format("{0}.{1}.exp0.tif {0}.{1}.exp0 -psm 10 nobatch box.train", lang, font);
}
ProcessBat(_tessacertExe, tmpTrainPath, args);
//unicharset_extractor ts.yzm.exp0.box
string cmdExe = Path.Combine(_tessacertPath, "unicharset_extractor.exe");
args = string.Format("{0}.{1}.exp0.box", lang, font);
ProcessBat(cmdExe, tmpTrainPath, args);
//echo ts 0 0 0 0 0 >> font_properties
sb = new StringBuilder();
sb.AppendLine(string.Format("{0} 0 0 0 0 0", font));
SaveFile(sb, Path.Combine(tmpTrainPath, "font_properties"));
//shapeclustering -F font_properties -U unicharset ts.yzm.exp0.tr
cmdExe = Path.Combine(_tessacertPath, "shapeclustering.exe");
args = string.Format("-F font_properties -U unicharset {0}.{1}.exp0.tr", lang, font);
ProcessBat(cmdExe, tmpTrainPath, args);
//mftraining -F font_properties -U unicharset -O ts.unicharset ts.yzm.exp0.tr
cmdExe = Path.Combine(_tessacertPath, "mftraining.exe");
args = string.Format("-F font_properties -U unicharset -O {0}.unicharset {0}.{1}.exp0.tr", lang, font);
ProcessBat(cmdExe, tmpTrainPath, args);
//cntraining ts.yzm.exp0.tr
cmdExe = Path.Combine(_tessacertPath, "cntraining.exe");
args = string.Format("{0}.{1}.exp0.tr", lang, font);
ProcessBat(cmdExe, tmpTrainPath, args);
string tmpFile;
FileInfo fi;
tmpFile = Path.Combine(tmpTrainPath, string.Format("{0}.normproto", lang));
if (File.Exists(tmpFile))
{
File.Delete(tmpFile);
}
tmpFile = Path.Combine(tmpTrainPath, string.Format("{0}.inttemp", lang));
if (File.Exists(tmpFile))
{
File.Delete(tmpFile);
}
tmpFile = Path.Combine(tmpTrainPath, string.Format("{0}.pffmtable", lang));
if (File.Exists(tmpFile))
{
File.Delete(tmpFile);
}
tmpFile = Path.Combine(tmpTrainPath, string.Format("{0}.shapetable", lang));
if (File.Exists(tmpFile))
{
File.Delete(tmpFile);
}
tmpFile = Path.Combine(tmpTrainPath, "normproto");
fi = new FileInfo(tmpFile);
fi.MoveTo(Path.Combine(tmpTrainPath, string.Format("{0}.normproto", lang)));
tmpFile = Path.Combine(tmpTrainPath, "inttemp");
fi = new FileInfo(tmpFile);
fi.MoveTo(Path.Combine(tmpTrainPath, string.Format("{0}.inttemp", lang)));
tmpFile = Path.Combine(tmpTrainPath, "pffmtable");
fi = new FileInfo(tmpFile);
fi.MoveTo(Path.Combine(tmpTrainPath, string.Format("{0}.pffmtable", lang)));
tmpFile = Path.Combine(tmpTrainPath, "shapetable");
fi = new FileInfo(tmpFile);
fi.MoveTo(Path.Combine(tmpTrainPath, string.Format("{0}.shapetable", lang)));
//combine_tessdata ts
cmdExe = Path.Combine(_tessacertPath, "combine_tessdata.exe");
args = string.Format("{0}", lang);
ProcessBat(cmdExe, tmpTrainPath, args);
string tmpTrainDataFile = string.Format("{0}.traineddata", lang);
tmpFile = Path.Combine(tmpTrainPath, tmpTrainDataFile);
fi = new FileInfo(tmpFile);
fi.CopyTo(Path.Combine(_tessacertPath, "tessdata", tmpTrainDataFile), true);
/*
* string _fullName = @"E:\tesseract-train\tesseract-train\tesseract-train\bin\Debug\1.tiff";
* string fileName = Path.GetFileName(_fullName);
* string path = Path.GetDirectoryName(_fullName);
* string ext = Path.GetExtension(fileName);
* string name = Path.GetFileNameWithoutExtension(fileName);
* this.Text = string.Format("name={0},ext={1},path={2}", name, ext, path);
*/
}
