public void ReadOCRResults(string dir)
{
DirectoryInfo dirinfo = new DirectoryInfo(dir);
foreach (FileInfo NextFile in dirinfo.GetFiles()) //
{
if (NextFile.Extension == ".docx"
&& !NextFile.Name.Contains('~'))
{
string fn = Path.GetFileNameWithoutExtension(NextFile.Name);
string[] token = fn.Split('_');
int num = Convert.ToInt16(token[0]);
int ch = Convert.ToInt16(token[2]);
int x = Convert.ToInt16(token[3]);
int y = Convert.ToInt16(token[4]);
double slope = Convert.ToInt16(token[6]);
int bbxx = Convert.ToInt16(token[7]);
int bbxy = Convert.ToInt16(token[8]);
int bbxw = Convert.ToInt16(token[9]);
int bbxh = Convert.ToInt16(token[10]);
textlabel_list.Add(new TextLabel(num,ch,x,y,slope,fn, bbxx,bbxy,bbxw,bbxh));
}
}
for (int i = 0; i < textlabel_list.Count; i++)
{
TextLabel textlabel = textlabel_list[i];
DocxToText dtt = new DocxToText(dir + textlabel.fn + ".docx");
textlabel.text = dtt.ExtractText();
textlabel.susp_char_count = dtt.susp_char_count;
//if (textlabel.text == "" || textlabel.text == null ||suscharratio(textlabel))// || line_counter > 1)
//{
// textlabel_list.RemoveAt(i);
// i--;
//}
}
ResultMerger();
}
public void ApplyABBYYOCR(string dir, string fn)
{
try
{
ABBYYSingleStringResultParser assrp = new ABBYYSingleStringResultParser();
StreamWriter sw = new StreamWriter(dir + "OCR_results.txt");
assrp.ReadOCRResults(dir);
//sw.WriteLine("Text ID" + ";"
// + "Recognized Text" + ";" + "Suspicious Text" + ";" + "BoundingBox Top-Left X" + ";" + "BoundingBox Top-Left Y" + ";" + "BoundingBox Width" + ";" + "BoundingBox Height" + ";" + "Position X" + ";" + "Position Y" + ";" + "Orientation");
sw.WriteLine("ID" + ","
+ "Text" + "," + "BoundingBoxTopLeftX" + "," + "BoundingBoxTopLeftY" + "," + "BoundingBoxWidth" + "," + "BoundingBoxHeight");
Bitmap srcimg = new Bitmap(dir+fn);
srcimg = ImageUtils.AnyToFormat24bppRgb(srcimg);
Graphics g = Graphics.FromImage(srcimg);
for (int i = 0; i < assrp.textlabel_list.Count; i++)
{
TextLabel textlabel = assrp.textlabel_list[i];
Font font = new Font("Arial", 10);
try
{
SolidBrush sb = new SolidBrush(Color.FromArgb(50, 0, 0, 0));
g.DrawRectangle(
new Pen(Color.Black, 3), new Rectangle(textlabel.bbxx, textlabel.bbxy, textlabel.bbxw, textlabel.bbxh));
g.FillRectangle(
sb, new Rectangle(textlabel.bbxx, textlabel.bbxy, textlabel.bbxw, textlabel.bbxh));
//g.DrawString(textlabel.id.ToString(), font, Brushes.Black, textlabel.bbxx, textlabel.bbxy);
g.DrawString(textlabel.text, font, Brushes.Black, textlabel.bbxx, textlabel.bbxy + textlabel.bbxh);
}
catch { }
if (i < assrp.textlabel_list.Count - 1)
sw.WriteLine(textlabel.id + ","
+ textlabel.text + "," + textlabel.bbxx + "," + -1 * textlabel.bbxy + "," + textlabel.bbxw + "," + textlabel.bbxh);
else
sw.Write(textlabel.id + ","
+ textlabel.text + "," + textlabel.bbxx + "," + -1 * textlabel.bbxy + "," + textlabel.bbxw + "," + textlabel.bbxh);
//srcimg.Save(CurrentState.output_dir + CurrentState.text_recognition_results_fn, ImageFormat.Png);
}
sw.Close();
g.Dispose();
//ShapefileUtils sut = new ShapefileUtils();
//sut.WrtieText2Shp(assrp, dir, "OCR_results");
}
catch (Exception exception)
{
Log.WriteLine("GenerateOCRResults: " + exception.Message);
}
}
