private void button1_Click(object sender, EventArgs e)
{
Image mg = Image.FromFile(Application.StartupPath + "\\1.jpg");
////Image mg = Image.FromFile("e:\\test24.tif");
templateViewer1.Initialize(new OCRRenderingData(), new OCRAnalysisRender(templateViewer1));
templateViewer1.Image = mg;
TesseractProcessor _ocrProcessor = null;
_ocrProcessor = new TesseractProcessor();
string _tessData = Application.StartupPath + "\\tessdata\\";
string _lang = "eng";
int _ocrEngineMode = 3;
bool status = _ocrProcessor.Init(_tessData, _lang, _ocrEngineMode);
_ocrProcessor.ROI = templateViewer1.RectA;
rec += templateViewer1.RectA.X + ";" + templateViewer1.RectA.Y + ";" + templateViewer1.RectA.Width + ";" + templateViewer1.RectA.Height + "|";
_ocrProcessor.UseROI = true;
string text = _ocrProcessor.Recognize(mg);
_ocrProcessor.ROI = templateViewer1.RectB;
rec += templateViewer1.RectB.X + ";" + templateViewer1.RectB.Y + ";" + templateViewer1.RectB.Width + ";" + templateViewer1.RectB.Height + "|";
_ocrProcessor.UseROI = true;
text += _ocrProcessor.Recognize(mg);
_ocrProcessor.ROI = templateViewer1.RectC;
rec += templateViewer1.RectC.X + ";" + templateViewer1.RectC.Y + ";" + templateViewer1.RectC.Width + ";" + templateViewer1.RectC.Height + "|";
_ocrProcessor.UseROI = true;
text += _ocrProcessor.Recognize(mg);
_ocrProcessor.ROI = templateViewer1.RectD;
rec += templateViewer1.RectD.X + ";" + templateViewer1.RectD.Y + ";" + templateViewer1.RectD.Width + ";" + templateViewer1.RectD.Height + "|";
_ocrProcessor.UseROI = true;
text += _ocrProcessor.Recognize(mg);
_ocrProcessor.ROI = templateViewer1.RectE;
rec += templateViewer1.RectE.X + ";" + templateViewer1.RectE.Y + ";" + templateViewer1.RectE.Width + ";" + templateViewer1.RectE.Height;
_ocrProcessor.UseROI = true;
text += _ocrProcessor.Recognize(mg);
MessageBox.Show(text);
MessageBox.Show(rec);
}
public static void Main(string[] args)
{
const string language = "eng";
string imageFile = args[0];
TesseractProcessor processor = new TesseractProcessor();
using (Bitmap bmp = Bitmap.FromFile(imageFile) as Bitmap)
{
var success = processor.Init(TessractData, language, (int)eOcrEngineMode.OEM_DEFAULT);
if (!success) {
Console.WriteLine("Failed to initialize tesseract.");
} else {
// use thresholder
processor.UseThresholder();
string text = processor.Recognize(bmp);
Console.WriteLine("Text:");
Console.WriteLine("*****************************");
Console.WriteLine(text);
Console.WriteLine("*****************************");
}
}
Console.WriteLine("Press any key to exit.");
Console.ReadKey();
}
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));
}
public static void Main(string[] args)
{
const string language = "eng";
string imageFile = args[0];
TesseractProcessor processor = new TesseractProcessor();
using (Bitmap bmp = Bitmap.FromFile(imageFile) as Bitmap)
{
var success = processor.Init(TessractData, language, (int)eOcrEngineMode.OEM_DEFAULT);
if (!success)
{
Console.WriteLine("Failed to initialize tesseract.");
}
else
{
// use thresholder
processor.UseThresholder();
string text = processor.Recognize(bmp);
Console.WriteLine("Text:");
Console.WriteLine("*****************************");
Console.WriteLine(text);
Console.WriteLine("*****************************");
}
}
Console.WriteLine("Press any key to exit.");
Console.ReadKey();
}
private void button1_Click(object sender, EventArgs e)
{
Image mg = Image.FromFile(Application.StartupPath + "\\1.jpg");
////Image mg = Image.FromFile("e:\\test24.tif");
imageViewer1.Initialize(new OCRRenderingData(), new OCRAnalysisRender(imageViewer1));
imageViewer1.Image = mg;
TesseractProcessor _ocrProcessor = null;
_ocrProcessor = new TesseractProcessor();
//string _tessData = "E:\\tesseract\\tesseract-ocr-dotnet-3.01\\tessdata\\";
string _tessData = Application.StartupPath + "\\tessdata\\";
string _lang = "eng";
int _ocrEngineMode = 3;
bool status = _ocrProcessor.Init(_tessData, _lang, _ocrEngineMode);
_ocrProcessor.ROI = imageViewer1.RectA;
_ocrProcessor.UseROI = true;
string text = _ocrProcessor.Recognize(mg);
//string text1 = text.Replace(" ", "");
MessageBox.Show(text);
}
public static string Recognize(string url)
{
WebRequest request = WebRequest.Create(url);
WebResponse response = request.GetResponse();
Stream st = response.GetResponseStream();
if (st == null)
{
return(string.Empty);
}
try
{
Bitmap bitmap = ((Bitmap)Bitmap.FromStream(st)).toGray().biLinear(2);
//var newBitmap= KiCut(bitmap, 12, 22, bitmap.Width - 12, 22);
string tessdata = Environment.CurrentDirectory + "\\tessdata\\";
string language = "eng";//设置训练文件的名称，后缀traineddata之前的名称
int oem = 3;
// http://www.lixin.me/blog/2012/05/26/29536 训练
using (TesseractProcessor processor = new TesseractProcessor())
{
//初始化
bool initFlage = processor.Init(tessdata, language, oem);
//processor.GetTesseractEngineVersion();//获取版本号
/*
* Set the current page segmentation mode. Defaults to PSM_SINGLE_BLOCK.
* The mode is stored as an IntParam so it can also be modified by ReadConfigFile or SetVariable("tessedit_pageseg_mode", mode as string). */
//processor.SetPageSegMode(ePageSegMode.PSM_SINGLE_BLOCK);
//设置ROI（图像的感兴趣区域）
processor.UseROI = true;
processor.ROI = new Rectangle(24, 0, bitmap.Width - 24, bitmap.Height);
//设置识别的变量 如果是自定义培训的文件 可以不用设置
//必须在初始化后调用
processor.SetVariable("tessedit_char_whitelist", "0123456789.");
//processor.SetVariable("tessedit_thresholding_method", "1"); //图像处理阀值是否打开
//processor.SetVariable("save_best_choices", "T");
using (Bitmap bmp = bitmap)
{
int i = 3;
oem = i;
string text = processor.Recognize(bmp);
char[] charsToTrim = { '\\', 'n', '\\', 'n' };
return(text.TrimEnd(charsToTrim));
//Console.WriteLine(
// string.Format(
// "RecognizeMode: {1}\nText:\n{0}\n++++++\n", text, ((eOcrEngineMode)oem).ToString()));
}
}
}
catch (Exception)
{
return(string.Empty);
}
}
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 void StartOCR(Bitmap theBitmap)
{
string language = "eng";
string languageData = Application.StartupPath + @"\tessdata\";
using (TesseractProcessor processor = new TesseractProcessor())
{
if (processor.Init(languageData, language, (int)eOcrEngineMode.OEM_DEFAULT))
{
string text = processor.Recognize(theBitmap);
richTextBox1.AppendText(text + "\n");
}
}
}
public static void Recognize()
{
int n_images = Workspace.Images.Length;
int i_image = n_images - 1;
i_image = 0;
string fileName = Workspace.Images[i_image];
string imageFile = Path.Combine(Workspace.InputFolder, fileName);
string language = "eng";
int oem = (int)eOcrEngineMode.OEM_DEFAULT;
using (TesseractProcessor processor = new TesseractProcessor())
{
using (Bitmap bmp = Bitmap.FromFile(imageFile) as Bitmap)
{
DateTime started = DateTime.Now;
DateTime ended = DateTime.Now;
for (int i = 0; i < 4; i++)
//for (int i = 3; i < 4; i++)
{
oem = i;
processor.Init(Workspace.TessdataFolder, language, oem);
string text = "";
unsafe
{
started = DateTime.Now;
text = processor.Recognize(bmp);
ended = DateTime.Now;
Console.WriteLine("Duration recognition: {0} ms\n\n", (ended - started).TotalMilliseconds);
}
Console.WriteLine(
string.Format("RecognizeMode: {1}\nRecognized Text:\n{0}\n++++++++++++++++++++++++++++++++\n", text, ((eOcrEngineMode)oem).ToString()));
}
}
}
}
static void Simple1_Recognize()
{
using (TesseractProcessor processor = new TesseractProcessor())
{
using (Bitmap bmp = Bitmap.FromFile("phototest.tif") as Bitmap)
{
DateTime started = DateTime.Now;
DateTime ended = DateTime.Now;
int oem = 0;
for (int i = 0; i < 4; i++)
//for (int i = 3; i < 4; i++)
{
oem = i;
bool ok = processor.Init(TessdataFolder, "eng", i);
if (ok)
{
string text = "";
unsafe
{
started = DateTime.Now;
text = processor.Recognize(bmp);
ended = DateTime.Now;
Console.WriteLine("Duration recognition: {0} ms\n\n", (ended - started).TotalMilliseconds);
}
Console.WriteLine(
string.Format("RecognizeMode: {1}\nRecognized Text:\n{0}\n++++++++++++++++++++++++++++++++\n", text, ((eOcrEngineMode)oem).ToString()));
}
else
{
Console.WriteLine("FAIL " + i.ToString());
}
}
}
}
}
private static void Recognize()
{
string imageFile = @"D:\Self-Study\OpenSources\Tesseract\original\phototest.tif";
imageFile = @"D:\Self-Study\OpenSources\Tesseract\original\eurotext.tif";
string tessdata = @"D:\Self-Study\OpenSources\Tesseract\original\tessdata\";
string language = "eng";
int oem = 3;
TesseractProcessor processor = new TesseractProcessor();
processor.Init(tessdata, language, oem);
using (Bitmap bmp = Bitmap.FromFile(imageFile) as Bitmap)
{
string text = processor.Recognize(bmp);
Console.WriteLine(
string.Format("Text:\n{0}\n", text));
}
}
public override string RecognizeText(IList <Image> images, string lang)
{
string tessdata = Path.Combine(basedir, TESSDATA);
TesseractProcessor processor = new TesseractProcessor();
processor.Init(tessdata, lang, oem);
StringBuilder strB = new StringBuilder();
foreach (Image image in images)
{
string text = processor.Recognize(image, rect);
if (text == null)
{
return(String.Empty);
}
strB.Append(text);
}
return(strB.ToString().Replace("\n", Environment.NewLine));
}
static void Simple_Recognize()
{
string imageFile = Images[0];
TesseractProcessor processor = new TesseractProcessor();
using (var bmp = Bitmap.FromFile(imageFile) as Bitmap)
{
var success = processor.Init(TessdataFolder, "eng", (int)eOcrEngineMode.OEM_DEFAULT);
if (!success)
{
Console.WriteLine("Failed to initialize tesseract.");
}
else
{
string text = processor.Recognize(bmp);
Console.WriteLine("Text:");
Console.WriteLine("*****************************");
Console.WriteLine(text);
Console.WriteLine("*****************************");
}
}
}
private static void Recognize()
{
string imageFile = @"D:\Self-Study\OpenSources\Tesseract\original\phototest.tif";
imageFile = @"D:\Self-Study\OpenSources\Tesseract\original\eurotext.tif";
string tessdata = @"D:\Self-Study\OpenSources\Tesseract\original\tessdata\";
string language = "eng";
int oem = 3;
TesseractProcessor processor = new TesseractProcessor();
processor.Init(tessdata, language, oem);
using (Bitmap bmp = Bitmap.FromFile(imageFile) as Bitmap)
{
string text = processor.Recognize(bmp);
Console.WriteLine(
string.Format("Text:\n{0}\n", text));
}
}
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;
}
private void button3_Click(object sender, EventArgs e)
{
OpenFileDialog opn = new OpenFileDialog();
opn.ShowDialog();
StreamReader sr = new StreamReader(opn.FileName);
String line = sr.ReadToEnd();
char[] delimiterChars = { '|' };
char[] delimiterChars1 = { ';' };
string[] str = line.Split(delimiterChars);
Image mg = Image.FromFile(Application.StartupPath + "\\1.jpg");
TesseractProcessor _ocrProcessor = null;
_ocrProcessor = new TesseractProcessor();
string _tessData = Application.StartupPath + "\\tessdata\\";
string _lang = "eng";
int _ocrEngineMode = 3;
bool status = _ocrProcessor.Init(_tessData, _lang, _ocrEngineMode);
string[] str1 = str[0].Split(delimiterChars1);
Rectangle recA = new Rectangle(Convert.ToInt32(str1[0]), Convert.ToInt32(str1[1]), Convert.ToInt32(str1[2]), Convert.ToInt32(str1[3]));
_ocrProcessor.ROI = recA;
_ocrProcessor.UseROI = true;
templateViewer1.DrawRect(recA);
templateViewer1.RectA = recA;
string text = _ocrProcessor.Recognize(mg);
string[] str2 = str[1].Split(delimiterChars1);
Rectangle recB = new Rectangle(Convert.ToInt32(str2[0]), Convert.ToInt32(str2[1]), Convert.ToInt32(str2[2]), Convert.ToInt32(str2[3]));
_ocrProcessor.ROI = recB;
_ocrProcessor.UseROI = true;
templateViewer1.DrawRect(recB);
templateViewer1.RectB = recB;
text += _ocrProcessor.Recognize(mg);
string[] str3 = str[2].Split(delimiterChars1);
Rectangle recC = new Rectangle(Convert.ToInt32(str3[0]), Convert.ToInt32(str3[1]), Convert.ToInt32(str3[2]), Convert.ToInt32(str3[3]));
_ocrProcessor.ROI = recC;
_ocrProcessor.UseROI = true;
templateViewer1.DrawRect(recC);
templateViewer1.RectC = recC;
text += _ocrProcessor.Recognize(mg);
string[] str4 = str[3].Split(delimiterChars1);
Rectangle recD = new Rectangle(Convert.ToInt32(str4[0]), Convert.ToInt32(str4[1]), Convert.ToInt32(str4[2]), Convert.ToInt32(str4[3]));
_ocrProcessor.ROI = recD;
_ocrProcessor.UseROI = true;
templateViewer1.DrawRect(recD);
templateViewer1.RectD = recD;
text += _ocrProcessor.Recognize(mg);
string[] str5 = str[4].Split(delimiterChars1);
Rectangle recE = new Rectangle(Convert.ToInt32(str5[0]), Convert.ToInt32(str5[0]), Convert.ToInt32(str5[0]), Convert.ToInt32(str5[0]));
_ocrProcessor.ROI = recE;
_ocrProcessor.UseROI = true;
templateViewer1.DrawRect(recE);
templateViewer1.RectE = recE;
text += _ocrProcessor.Recognize(mg);
MessageBox.Show(text);
}
public static string Recognize(string url,Rectangle rectangle)
{
WebRequest request = WebRequest.Create(url);
WebResponse response = request.GetResponse();
Stream st = response.GetResponseStream();
if (st == null)
{
return string.Empty;
}
Bitmap bitmap = ((Bitmap)Bitmap.FromStream(st)).toGray().biLinear(2);
//var newBitmap= KiCut(bitmap, 12, 22, bitmap.Width - 12, 22);
string tessdata = Environment.CurrentDirectory + "\\tessdata\\";
string language = "eng";//设置训练文件的名称，后缀traineddata之前的名称
int oem = 3;
// http://www.lixin.me/blog/2012/05/26/29536 训练
using (TesseractProcessor processor = new TesseractProcessor())
{
//初始化
bool initFlage = processor.Init(tessdata, language, oem);
//processor.GetTesseractEngineVersion();//获取版本号
/*
Set the current page segmentation mode. Defaults to PSM_SINGLE_BLOCK.
The mode is stored as an IntParam so it can also be modified by ReadConfigFile or SetVariable("tessedit_pageseg_mode", mode as string). */
//processor.SetPageSegMode(ePageSegMode.PSM_SINGLE_BLOCK);
//设置ROI（图像的感兴趣区域）
if (rectangle != null)
{
processor.UseROI = true;
processor.ROI = rectangle;
}
//设置识别的变量 如果是自定义培训的文件 可以不用设置
//必须在初始化后调用
processor.SetVariable("tessedit_char_whitelist", "0123456789.");
//processor.SetVariable("tessedit_thresholding_method", "1"); //图像处理阀值是否打开
//processor.SetVariable("save_best_choices", "T");
using (Bitmap bmp = bitmap)
{
string text = processor.Recognize(bmp);
char[] charsToTrim = { '\\', 'n', '\\', 'n' };
return text.TrimEnd(charsToTrim);
}
}
}
/// <summary>
/// Search a string into an image
/// </summary>
/// <param name="image">the image</param>
/// <param name="inputString">the string to search</param>
/// <returns>return true if the string was found</returns>
private static bool checkStringByOCR(Bitmap image, string inputString)
{
Stopwatch a = new Stopwatch();
a.Start();
using (var bmp = image)
{
try
{
//init the OCR engine, I use tessaract as OCR engine.
TesseractProcessor processor = new TesseractProcessor();
//verify if tessaract has been successfully loaded
var success = processor.Init(_ocrLangData, _ocrLanguageSelected, (int)eOcrEngineMode.OEM_DEFAULT);
if (!success)
{
//if tessaract has not been successfully loaded then write the error
LogUtils.Write(new StackFrame(0, true), LogUtils.ErrorLevel.Error, "Failed to start OCR engine");
Program.Finish(true);
return false;
}
else
{
//extract the string from the image
string textInImage = processor.Recognize(bmp);
//textInImage = textInImage.Replace('\n', ' ');
//textInImage = textInImage.Replace('\r', ' ');
//write a debug message
if (_debugLogLevel)
LogUtils.Write(new StackFrame(0, true), LogUtils.ErrorLevel.Debug, "text found: " + textInImage);
//check if the string extracted from the image is equal to the string that we want to find.
//NOTE that tesseract occasionally swaps a letter with another. For example, it may change
//an "m" with two letters "rm" or an "l" with a "|" and so on. So i try to solve some of these errors.
//You can set a debug level on the log option and search into the log file the string "text found: " and
//then see text that tesseract has found
if (Regex.IsMatch(textInImage, inputString) ||
Regex.IsMatch(textInImage.Replace("m", "rn"), inputString) || Regex.IsMatch(textInImage.Replace("rn", "m"), inputString) ||
Regex.IsMatch(textInImage.Replace("l", "1"), inputString) || Regex.IsMatch(textInImage.Replace("1", "l"), inputString) ||
Regex.IsMatch(textInImage.Replace("l", "i"), inputString) || Regex.IsMatch(textInImage.Replace("i", "l"), inputString) ||
Regex.IsMatch(textInImage.Replace("t", "l"), inputString) || Regex.IsMatch(textInImage.Replace("l", "t"), inputString) ||
Regex.IsMatch(textInImage.Replace("1", "i"), inputString) || Regex.IsMatch(textInImage.Replace("i", "1"), inputString) ||
Regex.IsMatch(textInImage.Replace("I", "l"), inputString) || Regex.IsMatch(textInImage.Replace("l", "I"), inputString) ||
Regex.IsMatch(textInImage.Replace("I", "1"), inputString) || Regex.IsMatch(textInImage.Replace("1", "I"), inputString) ||
Regex.IsMatch(textInImage.Replace("I", "t"), inputString) || Regex.IsMatch(textInImage.Replace("t", "I"), inputString) ||
Regex.IsMatch(textInImage.Replace("0", "o"), inputString) || Regex.IsMatch(textInImage.Replace("o", "0"), inputString) ||
Regex.IsMatch(textInImage.Replace("0", "O"), inputString) || Regex.IsMatch(textInImage.Replace("O", "0"), inputString) ||
Regex.IsMatch(textInImage.Replace("l", "|"), inputString) || Regex.IsMatch(textInImage.Replace("|", "l"), inputString) ||
Regex.IsMatch(textInImage.Replace("I", "|"), inputString) || Regex.IsMatch(textInImage.Replace("|", "I"), inputString) ||
Regex.IsMatch(textInImage.Replace("t", "|"), inputString) || Regex.IsMatch(textInImage.Replace("|", "t"), inputString) ||
Regex.IsMatch(textInImage.Replace("i", "|"), inputString) || Regex.IsMatch(textInImage.Replace("|", "i"), inputString) ||
Regex.IsMatch(textInImage.Replace("M", "II"), inputString) || Regex.IsMatch(textInImage.Replace("II", "M"), inputString) ||
Regex.IsMatch(textInImage.Replace("ni", "m"), inputString) || Regex.IsMatch(textInImage.Replace("m", "ni"), inputString))
{
a.Stop();
return true;
}
else
return false;
}
}
catch (Exception ex)
{
//write the error
LogUtils.Write(ex);
Program.Finish(true);
return false;
}
}
}
