private static void AnalyseLayout()
{
string imageFile = @"D:\Self-Study\OpenSources\Tesseract\original\phototest.tif";
//imageFile = @"D:\Self-Study\OpenSources\Tesseract\original\eurotext.tif";
TesseractProcessor processor = new TesseractProcessor();
processor.InitForAnalysePage();
processor.SetPageSegMode(ePageSegMode.PSM_AUTO);
using (Bitmap bmp = Bitmap.FromFile(imageFile) as Bitmap)
{
DocumentLayout doc = processor.AnalyseLayout(bmp);
Console.WriteLine(doc.ToString());
using (Image tmp = new Bitmap(bmp.Width, bmp.Height)) // prevents one-byte index format
{
using (Graphics grph = Graphics.FromImage(tmp))
{
Rectangle rect = new Rectangle(0, 0, tmp.Width, tmp.Height);
grph.DrawImage(bmp, rect, rect, GraphicsUnit.Pixel);
grph.SmoothingMode = System.Drawing.Drawing2D.SmoothingMode.HighQuality;
foreach (Block block in doc.Blocks)
{
DrawBlock(grph, block);
}
}
tmp.Save(@"D:\temp\page_layout_test2.bmp");
}
}
}
private static void AnalyseLayout()
{
string imageFile = @"D:\Self-Study\OpenSources\Tesseract\original\phototest.tif";
//imageFile = @"D:\Self-Study\OpenSources\Tesseract\original\eurotext.tif";
TesseractProcessor processor = new TesseractProcessor();
processor.InitForAnalysePage();
processor.SetPageSegMode(ePageSegMode.PSM_AUTO);
using (Bitmap bmp = Bitmap.FromFile(imageFile) as Bitmap)
{
DocumentLayout doc = processor.AnalyseLayout(bmp);
Console.WriteLine(doc.ToString());
using (Image tmp = new Bitmap(bmp.Width, bmp.Height)) // prevents one-byte index format
{
using (Graphics grph = Graphics.FromImage(tmp))
{
Rectangle rect = new Rectangle(0, 0, tmp.Width, tmp.Height);
grph.DrawImage(bmp, rect, rect, GraphicsUnit.Pixel);
grph.SmoothingMode = System.Drawing.Drawing2D.SmoothingMode.HighQuality;
foreach (Block block in doc.Blocks)
{
DrawBlock(grph, block);
}
}
tmp.Save(@"D:\temp\page_layout_test2.bmp");
}
}
}
public override string RecognizeText(IList <Image> images, string lang)
{
string tessdata = Path.Combine(basedir, TESSDATA);
TesseractProcessor processor = new TesseractProcessor();
processor.Init(tessdata, lang, oem);
processor.SetPageSegMode((ePageSegMode)Enum.Parse(typeof(ePageSegMode), PSM));
StringBuilder strB = new StringBuilder();
foreach (Image image in images)
{
//if (rect != Rectangle.Empty)
//{
// processor.UseROI = true;
// processor.ROI = rect;
//}
string text = processor.Recognize(image);
if (text == null)
{
return(String.Empty);
}
strB.Append(text);
}
return(strB.ToString().Replace("\n", Environment.NewLine));
}
private void button1_Click(object sender, EventArgs e)
{
string d = textBox_Path.Text;
if (!Directory.Exists(d))
{
MessageBox.Show("目录" + d + "不存在");
return;
}
if (!d.EndsWith("\\"))
{
d += "\\";
}
string d1 = d + "new\\";
if (!Directory.Exists(d1))
{
Directory.CreateDirectory(d1);
}
Regex reg = new Regex(@"\d+\.\d\d", RegexOptions.Compiled);
foreach (string f in Directory.GetFiles(textBox_Path.Text))
{
if (!f.ToLower().EndsWith(".png") && !f.ToLower().EndsWith(".jpg") && !f.ToLower().EndsWith(".bmp"))
{
continue;
}
Bitmap bmp = new Bitmap(f);
TesseractProcessor process = new TesseractProcessor();
process.SetPageSegMode(ePageSegMode.PSM_SINGLE_LINE);
process.Init(System.Environment.CurrentDirectory + "\\", "chi_sim", (int)eOcrEngineMode.OEM_DEFAULT);
try
{
string result = process.Recognize(bmp);
Match m = reg.Match(result);
if (m.Success)
{
string amount = m.ToString();
File.Copy(f, d1 + amount + Path.GetExtension(f), true);
}
else
{
File.Copy(f, d1 + Path.GetFileName(f), true);
}
}
catch (Exception)
{
MessageBox.Show("识别文件出错:" + f);
File.Copy(f, d1 + Path.GetFileName(f), true);
}
bmp.Dispose();
}
System.Diagnostics.Process.Start(d1);
}
public string UseTesseract(string imgFile)
{
this.ErrMsg = string.Empty;
string defLang = "eng";
string strResult = string.Empty;
try
{
using (TesseractProcessor processor = new TesseractProcessor())
{
processor.Init(this.TESSDATA, defLang, this.oem);
processor.SetPageSegMode((ePageSegMode)Enum.Parse(typeof(ePageSegMode), this.pageSegMode));
#if DEBUG
System.Diagnostics.Debug.WriteLine("processor:");
System.Diagnostics.Debug.WriteLine(processor.GetTesseractEngineVersion());
#endif
string strIndicate = Path.GetFileNameWithoutExtension(imgFile);
strResult = processor.Recognize(imgFile);
if (!string.IsNullOrEmpty(strResult))
{
// correct common errors caused by OCR
strResult = this.CorrectOCRErrors(strResult);
// correct letter cases
strResult = this.CorrectLetterCases(strResult);
}
}
strResult = strResult.Replace("\n", Environment.NewLine);
}
catch (Exception ex)
{
this.ErrMsg = ex.ToString();
strResult = string.Empty;
}
return strResult;
}