public Form1()
{
InitializeComponent();
this.PSN = new Person();
this.imagePath = Application.StartupPath + "/image";
if (Directory.Exists(this.imagePath) == false) Directory.CreateDirectory(this.imagePath);
this.imagePath += "/";
Init();
this.F = new mPublic();
this.CFG = new Config();
this.MAPP = new Mapping();
this.RMap = CFG.rect_map_now;
this.thKill = new Thread(new ThreadStart(AutoKill));
this.stopKill = true;
this.ocrNumber = new OCR();
/*
//初始化数字识别引擎
ocrNumber = new TesseractProcessor();
bInitN = ocrNumber.Init(@".\", "chi_sim", 3);
if (!bInitN) { MessageBox.Show("智能识别引擎初始化失败！"); return; }
ocrNumber.SetVariable("tessedit_char_whitelist", ".,2134567890 ");
* */
//初始化汉字引擎
ocrText = new TesseractProcessor();
bInitT = ocrText.Init(@".\", "chi_sim", 3);
if (!bInitT) { MessageBox.Show("智能识别引擎初始化失败！"); return; }
ocrText.SetVariable("tessedit_char_blacklist", "fi蜘fi'£郜QWERTYUIOPASDFGHJKLZXCVBNM1234567890μqwertyuiopasdfghjklzxcvbnm");
}
public Form1()
{
InitializeComponent();
m_tesseract = new TesseractProcessor();
m_tesseract.Init(m_path, m_lang, (int)TesseractEngineMode.DEFAULT);
m_tesseract.SetVariable("tessedit_pageseg_mode", TesseractPageSegMode.PSM_AUTO.ToString());
}
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 const string m_lang = "eng"; // 识别的语言文件名称
public string read2dig(Image image)
{
TesseractProcessor m_tesseract = new TesseractProcessor();
;
bool succeed = m_tesseract.Init(m_path, m_lang, (int)TesseractEngineMode.DEFAULT);
if (!succeed)
{
return("tesseract初始化失败");
}
m_tesseract.SetVariable("tessedit_char_whitelist", "0123456789"); //设置识别变量，当前只能识别数字。
image.Save("./img/原图.bmp", System.Drawing.Imaging.ImageFormat.Bmp);
Bitmap bmp = ToGray(image);
bmp.Save("./img/灰度图.bmp", System.Drawing.Imaging.ImageFormat.Bmp);
bmp = ConvertTo1Bpp1(bmp, 110);
bmp.Save("./img/二值图.bmp", System.Drawing.Imaging.ImageFormat.Bmp);
bmp = Zoom(bmp, 0.25);
bmp.Save("./img/放大缩小后.bmp", System.Drawing.Imaging.ImageFormat.Bmp);
m_tesseract.Clear();
m_tesseract.ClearAdaptiveClassifier();
return(m_tesseract.Apply(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");
}
}
}
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 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 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 Recognize()
{
int n_images = Workspace.Images.Length;
int i_image = n_images - 1;
//i_image = 0;
i_image = 2;
string fileName = Workspace.Images[i_image];
string imageFile = Path.Combine(Workspace.InputFolder, fileName);
string language = "eng";
int oem = (int)eOcrEngineMode.OEM_DEFAULT;
string name = Path.GetFileNameWithoutExtension(imageFile);
{
using (Bitmap bmp = Bitmap.FromFile(imageFile) as Bitmap)
{
using (GreyImage greyImage = GreyImage.FromImage(bmp))
{
ImageThresholder thresholder = new AdaptiveThresholder();
using (BinaryImage binImage = thresholder.Threshold(greyImage))
{
DateTime started = DateTime.Now;
DateTime ended = DateTime.Now;
Rectangle[] rois = new Rectangle[] {
Rectangle.FromLTRB(807, 43, 1351, 613),
Rectangle.FromLTRB(4, 604, binImage.Width - 15, binImage.Height-35)
};
int nROIs = rois.Length;
string[] texts = new string[nROIs];
#if PARALLEL
Parallel.For(0, nROIs, delegate(int iROI)
#else
using (TesseractProcessor processor = new TesseractProcessor())
for (int iROI = 0; iROI < nROIs; iROI++)
#endif
{
#if PARALLEL
using (TesseractProcessor processor = new TesseractProcessor())
#endif
{
Rectangle roi = rois[iROI];
{
//oem = (int)eOcrEngineMode.OEM_TESSERACT_CUBE_COMBINED;
processor.Init(Workspace.TessdataFolder, language, oem);
processor.UseROI = true;
processor.ROI = roi;
unsafe
{
texts[iROI] = processor.RecognizeBinaryImage(
binImage.BinaryData, binImage.Width, binImage.Height);
}
}
}
}
private void End()
{
if (_ocrProcessor != null)
{
_ocrProcessor.End();
_ocrProcessor = null;
}
}
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);
}
}
static oTesseractRequest __ocrExecute2(oTesseractRequest req, Bitmap bitmap)
{
using (TesseractProcessor processor = new TesseractProcessor())
{
processor.InitForAnalysePage();
using (GreyImage greyImage = GreyImage.FromImage(bitmap))
{
//greyImage.Save(ImageFormat.Bmp, outFile2);
ImageThresholder thresholder = new AdaptiveThresholder();
using (BinaryImage binImage = thresholder.Threshold(greyImage))
{
DocumentLayout doc = null;
switch (req.command)
{
case TESSERACT_COMMAND.GET_TEXT:
//string s = tes.GetText().Trim();
//req.output_text = s;
//req.output_count = s.Length;
req.ok = 1;
break;
default:
unsafe
{
doc = processor.AnalyseLayoutBinaryImage(
binImage.BinaryData, greyImage.Width, greyImage.Height);
}
if (doc != null)
{
var bs = new List <string>();
if (doc.Blocks.Count > 0)
{
for (int i = 0; i < doc.Blocks.Count; i++)
{
for (int j = 0; j < doc.Blocks[i].Paragraphs.Count; j++)
{
bs.AddRange(doc.Blocks[j].Paragraphs[j].Lines
.Select(x => string.Format(
"{0}_{1}_{2}_{3}", x.Left, x.Top, x.Right, x.Bottom)));
}
}
}
req.output_format = "left_top_right_bottom";
req.output_text = string.Join("|", bs.ToArray());
req.output_count = bs.Count;
req.ok = 1;
}
break;
}
}
}
}
return(req);
}
public TesseractOCREngine(Image image)
{
this.image = image;
m_tesseract = new TesseractProcessor();
bool succeed = m_tesseract.Init(m_path, m_lang, (int)TesseractEngineMode.DEFAULT);
m_tesseract.SetVariable("tessedit_char_whitelist", "0123456789-.eExXpP/\\");
m_tesseract.SetVariable("tessedit_pageseg_mode", ((int)TesseractPageSegMode.PSM_SINGLE_LINE).ToString());
System.Environment.CurrentDirectory = System.IO.Path.GetFullPath(m_path);
}
private void ProcessFrame(bool instant)
{
if (!btnScreenshot.Enabled)
{
return;
}
if (!File.Exists(openFileDialogTesseract.FileName))
{
MessageBox.Show(this, "Tesseract-OCR path not configured!", "Screenulate",
MessageBoxButtons.OK, MessageBoxIcon.Error);
return;
}
tesseractProgress.Style = ProgressBarStyle.Marquee;
btnScreenshot.Enabled = false;
var screenshotForm = new ScreenshotForm {
Rectangle = _lastScreenshotRect
};
if (instant)
{
screenshotForm.CaptureImage(_lastScreenshotRect);
}
else
{
screenshotForm.ShowDialog(this);
}
if (screenshotForm.Bitmap == null)
{
tesseractProgress.Style = ProgressBarStyle.Blocks;
btnScreenshot.Enabled = true;
return;
}
_lastScreenshotRect = screenshotForm.Rectangle;
IImage oldImage = previewBox.Image;
previewBox.Image = new Image <Bgr, byte>(screenshotForm.Bitmap);
oldImage?.Dispose();
var processor = new TesseractProcessor(openFileDialogTesseract.FileName, previewBox.DisplayedImage);
processor.Process();
processor.Completed += TesseractProcessor_Completed;
processor.Finished += TesseractProcessor_Finished;
processor.Error += (sender, args) =>
{
MessageBox.Show($@"{args.GetException().Message}", "Screenulate", MessageBoxButtons.OK,
MessageBoxIcon.Error);
};
}
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);
}
static void Simple2_Recognize()
{
int n_images = Images.Length;
int i_image = n_images - 1;
//i_image = 0;
string fileName = Images[i_image];
string imageFile = Path.Combine(InputFolder, fileName);
string language = "eng";
int oem = (int)eOcrEngineMode.OEM_DEFAULT;
using (TesseractProcessor processor = new TesseractProcessor())
{
using (Bitmap bmp = Bitmap.FromFile(imageFile) as Bitmap)
{
using (GreyImage greyImage = GreyImage.FromImage(bmp))
{
ImageThresholder thresholder = new AdaptiveThresholder();
using (BinaryImage binImage = thresholder.Threshold(greyImage))
{
DateTime started = DateTime.Now;
DateTime ended = DateTime.Now;
int i = 3;
//for (i = 0; i < 4; i++)
//for (i = 3; i < 4; i++)
{
oem = i;
processor.Init(TessdataFolder, language, oem);
string text = "";
unsafe
{
started = DateTime.Now;
text = processor.RecognizeBinaryImage(
binImage.BinaryData, greyImage.Width, greyImage.Height);
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()));
}
}
}
}
}
}
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 = 3;
TesseractProcessor processor = new TesseractProcessor();
using (Bitmap bmp = Bitmap.FromFile(imageFile) as Bitmap)
{
GreyImage greyImage = GreyImage.FromImage(bmp);
DateTime started = DateTime.Now;
AdaptiveThresholder thresholder = new AdaptiveThresholder();
BinaryImage binImage = thresholder.Threshold(greyImage);
DateTime ended = DateTime.Now;
Console.WriteLine("Duration thresholding: {0} ms", (ended - started).TotalMilliseconds);
binImage.Invert();
//for (int i = 0; i < 4; i++)
for (int i = 3; i < 4; i++)
{
oem = i;
oem = (int)eOcrEngineMode.OEM_TESSERACT_CUBE_COMBINED;
processor.Init(Workspace.TessdataFolder, language, oem);
string text = "";
unsafe
{
started = DateTime.Now;
//string text = processor.Recognize(bmp);
text = processor.RecognizeBinaryImage(
binImage.BinaryData, binImage.Width, binImage.Height);
ended = DateTime.Now;
Console.WriteLine("Duration recognition: {0} ms\n\n", (ended - started).TotalMilliseconds);
}
Console.WriteLine(
string.Format("RecognizeMode: {1}\nText:\n{0}\n++++++++++++++++++++++++++++++++\n", text, ((eOcrEngineMode)oem).ToString()));
}
}
}
protected override void OnLoad(EventArgs e)
{
base.OnLoad(e);
this.LoadGeneralSettings();
_ocrProcessor = new TesseractProcessor();
_ocrProcessor.DoMonitor = true;
bool status = _ocrProcessor.Init(_tessData, _lang, _ocrEngineMode);
Console.WriteLine(string.Format("[DEBUG] Init status: {0}", status));
}
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);
}
private void OCR_Load(object sender, EventArgs e)
{
m_tesseract = new TesseractProcessor();
bool succeed = m_tesseract.Init(m_path, m_lang, 3);
if (!succeed)
{
MessageBox.Show("Tesseract initialization failed. The application will exit.");
Application.Exit();
}
m_tesseract.SetVariable("tessedit_char_whitelist", "ABCDEFHKLMNPQRSTUVXYZ-.1234567890").ToString(); //Training Sets //abcdefghijklmnopqrstuvxyzABCDEFGHIJKLMNOPQRSTUVXYZ-.1234567890
System.Environment.CurrentDirectory = System.IO.Path.GetFullPath(m_path);
}
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 TesseractEngine(string langFilesPath, string lang)
{
this.tp = new TesseractProcessor();
this.lang = lang;
this.langFilesPath = langFilesPath;
if (this.langFilesPath.get_Length() == 0)
{
this.langFilesPath = string.Concat(Application.get_StartupPath(), "\\tessdata\\");
}
if (!this.langFilesPath.EndsWith("\\"))
{
TesseractEngine tesseractEngine = this;
tesseractEngine.langFilesPath = string.Concat(tesseractEngine.langFilesPath, "\\");
}
}
void CheckInPage_Loaded(object sender, RoutedEventArgs e)
{
full_tesseract = new TesseractProcessor();
bool succeed = full_tesseract.Init(m_path, m_lang, 3);
if (!succeed)
{
System.Windows.MessageBox.Show("Lỗi thư viện Tesseract. Chương trình cần kết thúc.");
}
full_tesseract.SetVariable("tessedit_char_whitelist", "ACDFHKLMNPRSTVXY1234567890").ToString();
ch_tesseract = new TesseractProcessor();
succeed = ch_tesseract.Init(m_path, m_lang, 3);
if (!succeed)
{
System.Windows.MessageBox.Show("Lỗi thư viện Tesseract. Chương trình cần kết thúc.");
}
ch_tesseract.SetVariable("tessedit_char_whitelist", "ACDEFHKLMNPRSTUVXY").ToString();
num_tesseract = new TesseractProcessor();
succeed = num_tesseract.Init(m_path, m_lang, 3);
if (!succeed)
{
System.Windows.MessageBox.Show("Lỗi thư viện Tesseract. Chương trình cần kết thúc.");
}
num_tesseract.SetVariable("tessedit_char_whitelist", "1234567890").ToString();
System.Environment.CurrentDirectory = System.IO.Path.GetFullPath(m_path);
for (int i = 0; i < box.Length; i++)
{
box[i] = new PictureBox();
}
string folder = System.AppDomain.CurrentDomain.BaseDirectory + "\\ImageTest";
foreach (string fileName in Directory.GetFiles(folder, "*.bmp", SearchOption.TopDirectoryOnly))
{
lstimages.Add(System.IO.Path.GetFullPath(fileName));
}
foreach (string fileName in Directory.GetFiles(folder, "*.jpg", SearchOption.TopDirectoryOnly))
{
lstimages.Add(System.IO.Path.GetFullPath(fileName));
}
pic1 = (hostPic1.Child as System.Windows.Forms.PictureBox);
pic2 = (hostPic2.Child as System.Windows.Forms.PictureBox);
img1 = (hostImg1.Child as Emgu.CV.UI.ImageBox);
}
private void MainForm_Load(object sender, EventArgs e)
{
int soluong;
int tong;
tong = int.Parse(txt_tong.Text);
soluong = int.Parse(bss.so_luong_xe());
txt_conlai.Text = (tong - soluong).ToString();
dataGridView1.DataSource = bss.Danhsachxe();
capture = new Emgu.CV.Capture();
timer1.Enabled = true;
full_tesseract = new TesseractProcessor();
bool succeed = full_tesseract.Init(m_path, m_lang, 3);
if (!succeed)
{
MessageBox.Show("Tesseract initialization failed. The application will exit.");
Application.Exit();
}
full_tesseract.SetVariable("tessedit_char_whitelist", "ABCDEFHKLMNPRSTVXY1234567890").ToString();
ch_tesseract = new TesseractProcessor();
succeed = ch_tesseract.Init(m_path, m_lang, 3);
if (!succeed)
{
MessageBox.Show("Tesseract initialization failed. The application will exit.");
Application.Exit();
}
ch_tesseract.SetVariable("tessedit_char_whitelist", "ABCDEFHKLMNPRSTUVXY").ToString();
num_tesseract = new TesseractProcessor();
succeed = num_tesseract.Init(m_path, m_lang, 3);
if (!succeed)
{
MessageBox.Show("Tesseract initialization failed. The application will exit.");
Application.Exit();
}
num_tesseract.SetVariable("tessedit_char_whitelist", "1234567890").ToString();
m_path = System.Environment.CurrentDirectory + "\\";
string[] ports = SerialPort.GetPortNames();
for (int i = 0; i < box.Length; i++)
{
box[i] = new PictureBox();
}
}
//private string _Caption = "http://www.laptrinhvb.net";
#endregion
private void FrmMain_Load(object sender, EventArgs e)
{
full_tesseract = new TesseractProcessor();
bool succeed = full_tesseract.Init(m_path, m_lang, 3);
if (!succeed)
{
MessageBox.Show("Lỗi thư viện Tesseract. Chương trình cần kết thúc.");
Application.Exit();
}
full_tesseract.SetVariable("tessedit_char_whitelist", "ACDFHKLMNPRSTVXY1234567890").ToString();
ch_tesseract = new TesseractProcessor();
succeed = ch_tesseract.Init(m_path, m_lang, 3);
if (!succeed)
{
MessageBox.Show("Lỗi thư viện Tesseract. Chương trình cần kết thúc.");
Application.Exit();
}
ch_tesseract.SetVariable("tessedit_char_whitelist", "ACDEFHKLMNPRSTUVXY").ToString();
num_tesseract = new TesseractProcessor();
succeed = num_tesseract.Init(m_path, m_lang, 3);
if (!succeed)
{
MessageBox.Show("Lỗi thư viện Tesseract. Chương trình cần kết thúc.");
Application.Exit();
}
num_tesseract.SetVariable("tessedit_char_whitelist", "1234567890").ToString();
System.Environment.CurrentDirectory = System.IO.Path.GetFullPath(m_path);
for (int i = 0; i < box.Length; i++)
{
box[i] = new PictureBox();
}
string folder = Application.StartupPath + "\\ImageTest";
foreach (string fileName in Directory.GetFiles(folder, "*.bmp", SearchOption.TopDirectoryOnly))
{
lstimages.Add(Path.GetFullPath(fileName));
}
foreach (string fileName in Directory.GetFiles(folder, "*.jpg", SearchOption.TopDirectoryOnly))
{
lstimages.Add(Path.GetFullPath(fileName));
}
}
/* private void button3_Click(object sender, EventArgs e)
* {
* if (capture != null)
* {
* timer1.Enabled = false;
* pictureBox_XeRA.Image = null;
* IF.pictureBox2.Image = null;
* capture.QueryFrame().Save("aa.bmp");
* FileStream fs = new FileStream(m_path + "aa.bmp", FileMode.Open, FileAccess.Read);
* Image temp = Image.FromStream(fs);
* fs.Close();
* pictureBox_XeRA.Image = temp;
* IF.pictureBox2.Image = temp;
* pictureBox_XeRA.Update();
* IF.pictureBox2.Update();
* Image temp1;
* string temp2, temp3;
* /*Reconize(m_path + "aa.bmp", out temp1, out temp2, out temp3);
* pictureBox_XeVAO.Image = temp1;
* if(temp3 == "")
* text_BiensoVAO.Text = "ko nhận dạng dc biển số";
* else
* text_BiensoVAO.Text = temp3;
*
* timer1.Enabled = true;
* }
*
* }*/
private void MainForm_Load_1(object sender, EventArgs e)
{
capture = new Emgu.CV.Capture();
timer2.Enabled = true;
timer4.Start();
IF = new ImageForm();
full_tesseract = new TesseractProcessor();
bool succeed = full_tesseract.Init(m_path, m_lang, 3);
if (!succeed)
{
MessageBox.Show("Tesseract initialization failed. The application will exit.");
Application.Exit();
}
full_tesseract.SetVariable("tessedit_char_whitelist", "ABCDEFHKLMNPRSTVXY1234567890").ToString();
ch_tesseract = new TesseractProcessor();
succeed = ch_tesseract.Init(m_path, m_lang, 3);
if (!succeed)
{
MessageBox.Show("Tesseract initialization failed. The application will exit.");
Application.Exit();
}
ch_tesseract.SetVariable("tessedit_char_whitelist", "ABCDEFHKLMNPRSTUVXY").ToString();
num_tesseract = new TesseractProcessor();
succeed = num_tesseract.Init(m_path, m_lang, 3);
if (!succeed)
{
MessageBox.Show("Tesseract initialization failed. The application will exit.");
Application.Exit();
}
num_tesseract.SetVariable("tessedit_char_whitelist", "1234567890").ToString();
m_path = System.Environment.CurrentDirectory + "\\";
string[] ports = SerialPort.GetPortNames();
for (int i = 0; i < box.Length; i++)
{
box[i] = new PictureBox();
}
}
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()));
}
}
}
}
protected void Page_Load(object sender, EventArgs e)
{
var image = System.Drawing.Image.FromFile(@"D:\Image\Capture1T.tif");
m_tesseract = new TesseractProcessor();
bool succeed = m_tesseract.Init(m_path, m_lang, (int)TesseractEngineMode.DEFAULT);
if (!succeed)
{
}
m_tesseract.SetVariable("tessedit_pageseg_mode", ((int)TesseractPageSegMode.PSM_SINGLE_LINE).ToString());
m_tesseract.Clear();
m_tesseract.ClearAdaptiveClassifier();
string outValue = m_tesseract.Apply(image);
Response.Write(outValue);
}
static void Main()
{
Application.EnableVisualStyles();
Application.SetCompatibleTextRenderingDefault(false);
IOcrProcessor ocrProcessor = new TesseractProcessor();
IBlobDim gameBlobDef = new GameBlobDef
{
MinWidth = 42,
MaxWidth = 170,
MinHeight = 11,
MaxHeight = 13
};
IBlobCounter blobCounter = new AccordCounter();
blobCounter.SetBlobDim(gameBlobDef);
Application.Run(new Form1(ocrProcessor, blobCounter));
}
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 void button1_Click(object sender, RoutedEventArgs e)
{
const string language = "eng";
const string TessractData = @"C:\Users\Salman\Documents\GitHub\project-pinnacle\Research\Salman\OCRTest\tessdata\";
const string Path = @"C:\Users\Salman\NUS\EE4001\Research\trial.tif";
TesseractProcessor processor = new TesseractProcessor();
processor.DoMonitor = true;
System.Drawing.Image bmp = System.Drawing.Image.FromFile(Path);
bool x = processor.Init(TessractData, language, (int)eOcrEngineMode.TESSERACT_CUBE_COMBINED);
processor.Clear();
processor.ClearAdaptiveClassifier();
string result = processor.Apply(bmp);
List<Word> detectedWords = processor.RetriveResultDetail();
int a = 2;
//this.UpdateImageViewer(detectedWords);
//using (var bmp = Bitmap.FromFile(Path) as Bitmap)
//{
// var success = processor.Init(TessractData, language, (int)eOcrEngineMode.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("*****************************");
// }
//}
//Console.WriteLine("Press any key to exit.");
//Console.ReadKey();
}
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));
}
public static string Ocr(Bitmap image_s, bool isFull, TesseractProcessor full_tesseract, TesseractProcessor num_tesseract, TesseractProcessor ch_tesseract, bool isNum = false)
{
Image <Gray, byte> image = new Image <Gray, byte>(image_s);
while (true)
{
if ((double)CvInvoke.cvCountNonZero((IntPtr)((UnmanagedObject)image)) / (double)(image.Width * image.Height) <= 0.5)
{
image = image.Dilate(2);
}
else
{
break;
}
}
Bitmap bitmap = image.ToBitmap();
TesseractProcessor tesseractProcessor = !isFull ? (!isNum ? ch_tesseract : num_tesseract) : full_tesseract;
int num = 0;
tesseractProcessor.Clear();
tesseractProcessor.ClearAdaptiveClassifier();
string str = tesseractProcessor.Apply((Image)bitmap);
while (str.Length > 3)
{
bitmap = new Image <Gray, byte>(bitmap).Erode(2).ToBitmap();
tesseractProcessor.Clear();
tesseractProcessor.ClearAdaptiveClassifier();
str = tesseractProcessor.Apply((Image)bitmap);
++num;
if (num > 10)
{
str = "";
break;
}
}
return(str);
}
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;
}
}
}
public parser(Image img, Form1 call)
{
baseIMG = img;
caller = call;
ocr = new TesseractProcessor();
ocr.DoMonitor = true;
//MessageBox.Show(Application.StartupPath+@"\");
ocr.Init(Application.StartupPath + @"\tessdata", "deu", 0);
}
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));
}
}
private void End()
{
if (_ocrProcessor != null)
{
_ocrProcessor.End();
_ocrProcessor = null;
}
}
protected override void OnLoad(EventArgs e)
{
base.OnLoad(e);
this.LoadGeneralSettings();
_ocrProcessor = new TesseractProcessor();
_ocrProcessor.DoMonitor = true;
bool status = _ocrProcessor.Init(_tessData, _lang, _ocrEngineMode);
Console.WriteLine(string.Format("[DEBUG] Init status: {0}", status));
}
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 string GetNumber(Bitmap image)
{
UnCodebase ud = new UnCodebase(image);
ud.DealMap2();
image = ud.bmpobj;
pictureBox1.Image = image;
if (!bInitT)
{
ocrNumber = new TesseractProcessor();
bool inited = ocrNumber.Init(@".\", "chi_sim", 3);
if (!inited) { MessageBox.Show("智能识别引擎初始化失败！"); return ""; }
ocrNumber.SetVariable("tessedit_char_whitelist", ".,3214567890 ");
}
ocrNumber.Clear();
ocrNumber.ClearAdaptiveClassifier();
string result = ocrNumber.Recognize(image);
return result;
}
* */
private string GetChiSim(Bitmap image)
{
UnCodebase ud = new UnCodebase(image);
ud.DealText();
image = ud.image;
//pictureBox1.Image = image; textBox3.Text = ud.txt;
if (!bInitT)
{
ocrText = new TesseractProcessor();
bInitT = ocrText.Init(@".\", "chi_sim", 3);
if (!bInitT) { MessageBox.Show("智能识别引擎初始化失败！"); return ""; }
ocrText.SetVariable("tessedit_char_blacklist", "1234567890μqwertyuiopasdfghjklzxcvbnm'I flU]M.H flO");
}
ocrText.Clear();
ocrText.ClearAdaptiveClassifier();
string result = ocrText.Recognize(image);
return result;
}
