public static string OCR(Bitmap img)
{
TesseractEngine ocr = null;
string sResult = "";
try
{
ocr = new TesseractEngine("./tessdata", "eng"); //初始化 (一定要放在tessdata資料夾下)
ocr.SetVariable("tessedit_char_whitelist", "0123456789"); //強迫Char List，較準確
Page page = ocr.Process(img, PageSegMode.SingleLine);
sResult = page.GetText();//result
page.Dispose();
}
catch (Exception ex)
{
//MessageBox.Show(ex.Message);
sResult = "";
}
finally
{
ocr?.Dispose();
}
return(sResult.Replace(" ", ""));
}
public void Dispose()
{
if (engine != null)
{
engine.Dispose();
engine = null;
}
}
public void Dispose()
{
if (_engine != null)
{
_engine.Dispose();
_engine = null;
}
}
private void button1_Click(object sender, EventArgs e)
{
double width = Math.Floor(img1.Width / img1.HorizontalResolution);
double height = Math.Floor(img1.Height / img1.VerticalResolution);
label1.Text = img1.HorizontalResolution + " " + img1.VerticalResolution;
img1 = Grayscale.CommonAlgorithms.BT709.Apply(img1);
Threshold filter = new Threshold(150);
filter.ApplyInPlace(img1);
DocumentSkewChecker skewChecker = new DocumentSkewChecker();
double angle = skewChecker.GetSkewAngle(img1);
RotateBilinear rotationFilter = new RotateBilinear(-angle);
rotationFilter.FillColor = Color.White;
img1 = rotationFilter.Apply(img1);
pictureBox1.Image = img1;
TesseractEngine engine = new TesseractEngine(@"C:\Users\moham\source\repos\pfe3\tessdata\", "fra", EngineMode.Default);
if (((width == 3) && (height == 2)) || (width == 2 && height == 3))
{
if ((width == 2 && height == 3))
{
rotationFilter = new RotateBilinear(-90);
rotationFilter.FillColor = Color.White;
img1 = rotationFilter.Apply(img1);
Console.WriteLine("if1");
}
string ocrText = engine.Process(img1, new Rect(700, 450, 150, 100)).GetText();
engine.Dispose();
ocrText = ocrText.Replace(" ", " ");
Console.WriteLine(ocrText + " " + ocrText.Length);
if (ocrText.Length <= 9)
{
label1.Text = "carte national " + ocrText;
}
else
{
rotationFilter = new RotateBilinear(180);
rotationFilter.FillColor = Color.White;
img1 = rotationFilter.Apply(img1);
engine = new TesseractEngine(@"C:\Users\moham\source\repos\pfe3\tessdata\", "fra", EngineMode.Default);
ocrText = engine.Process(img1, new Rect(700, 450, 150, 100)).GetText();
engine.Dispose();
ocrText = ocrText.Replace(" ", "");
}
label1.Text = "carte national " + ocrText;
pictureBox1.Image = img1;
engine.Dispose();
}
else if ((width == 8 && height == 11) || (width == 11 && height == 8))
{
}
else
{
}
}
//получение номера серии через с помощью масштабирования страницы с разным шагом(исп-ть, если не находит на исходном масштабе)
//sourceImage = входное изображение, stepScale = шаг масштабирования, amountOfSteps = кол-во проходов рескейлинга
public string getSeriaNumberViaRescaling(Image sourceImage, float stepScale, int amountOfSteps)
{
//подключаем распознаватель
TesseractEngine ocrRus = new TesseractEngine("./tessdata", "rus", EngineMode.Default);
TesseractEngine ocrEng = new TesseractEngine("./tessdata", "eng", EngineMode.Default);
//создаем графический фильтр
TResizeTool resizeTool = new TResizeTool();
//ставим методику распознавания страницы
PageSegMode pSegMode = PageSegMode.SparseText;
string seriaNumber = "";
float scaleValue = 1.0f;//исходный масштаб
List <string> seriaList = new List <string>();
for (int k = 1; k <= amountOfSteps; k++)
{
//по индексу отмасш-й страницы берем исходную(без масштаба)
Image img = sourceImage;
int newWidth = Convert.ToInt32((img.Width) * ((k * stepScale) + scaleValue));
int newHeight = Convert.ToInt32((img.Height) * ((k * stepScale) + scaleValue));
Bitmap imgBitmap = resizeTool.resizeImage(img, newWidth, newHeight);
Image rescaledPage = ((Image)imgBitmap);
string rescaledPageTextRus = "";
string rescaledPageTextEng = "";
//удаляем процесс, чтобы не было ошибки
using (var pageRus = ocrRus.Process(imgBitmap, pSegMode))
{
rescaledPageTextRus = pageRus.GetText();
}
//удаляем процесс, чтобы не было ошибки
using (var pageEng = ocrEng.Process(imgBitmap, pSegMode))
{
rescaledPageTextEng = pageEng.GetText();
}
seriaList.Add(getSeriaNumber(rescaledPageTextRus, rescaledPageTextEng));
}
seriaNumber = highestMatch(seriaList);
ocrRus.Dispose();
ocrEng.Dispose();
return(seriaNumber);
}
protected virtual void Dispose(bool disposing)
{
if (disposing)
{
//remove managed resources
}
if (_tesseractEngine != null)
{
_tesseractEngine.Dispose();
}
}
protected virtual void Dispose(bool disposing)
{
if (!disposed)
{
if (disposing)
{
_engine.Dispose();
}
disposed = true;
}
}
protected override void OnFormClosing(FormClosingEventArgs e)
{
base.OnFormClosing(e);
foreach (Observer observer in Observers)
{
observer.Dispose();
}
ScreenSaverPreventer.Dispose();
SpeechSynthesizer.SpeakAsyncCancelAll();
SpeechSynthesizer.Dispose();
TesseractEngine.Dispose();
}
protected virtual void Dispose(bool disposing)
{
// Only dispose once:
if (!m_disposed)
{
// Don't dispose if called from finalizer or if m_ocrEngine is null:
if (disposing && m_ocrEngine != null)
{
m_ocrEngine.Dispose();
m_ocrEngine = null;
}
m_disposed = true;
}
}
protected virtual void Dispose(bool disposing)
{
if (!is_disposed)
{
if (disposing)
{
ocr?.Dispose();
ocr = null;
streamWriter?.Dispose();
streamWriter = null;
}
this.is_disposed = true;
}
}
public void Recognize()
{
try
{
string path = @"C:\Users\cf200\source\repos\Scanner\Scanner\bin\Debug\";
TesseractEngine _ocr = new TesseractEngine(path, "chi_sim", EngineMode.Default);
Bitmap bit = inImg.Bitmap;
//bit = PreprocesImage(bit);//进行图像处理,如果识别率低可试试
Page page = _ocr.Process(bit);
recoText = page.GetText();//识别后的内容
page.Dispose();
_ocr.Dispose();
Console.WriteLine(recoText);
}
catch (Exception ex)
{
Console.WriteLine(ex.ToString());
}
}
public string LerArquivoImagem(string pathImage)
{
try
{
string directory = Directory.GetCurrentDirectory();
var engine = new TesseractEngine("./tessdata", "por", EngineMode.Default);
var pix = Pix.LoadFromFile(pathImage);
var page = engine.Process(pix, PageSegMode.Auto);
var meanConfidence = page.GetMeanConfidence();
var text = page.GetText();
var indexTrial = text.IndexOf("notice!)") + 8;
string subs = text.Substring(indexTrial + 1);
pix.Dispose();
engine.Dispose();
return(subs);
}
catch (Exception e)
{
Debug.WriteLine(e.Message);
return($"Erro ao tentar ler resultado do documento enviado, erro: {e.Message}");
}
}
/// <summary>
/// 图片文字识别
/// </summary>
/// <param name="imageUrl">图片路径</param>
/// <param name="tessdataUrl">语言包路径</param>
/// <param name="language">语言包名称</param>
/// <param name="variableData">设置识别变量，默认识别数字</param>
/// <param name="pageSegMode">PageSegMode</param>
/// <returns></returns>
public static string Identity(string imageUrl, string tessdataUrl, string language = "eng", string variableData = "0123456789", PageSegMode pageSegMode = PageSegMode.Auto)
{
Bitmap bitmap = null;
TesseractEngine tesseractEngine = null;
Page page = null;
try
{
bitmap = new Bitmap(imageUrl);
tesseractEngine = new TesseractEngine(tessdataUrl, language, EngineMode.Default);
if (!string.IsNullOrEmpty(variableData))
{
tesseractEngine.SetVariable("tessedit_char_whitelist", variableData);
}
page = tesseractEngine.Process(PixConverter.ToPix(bitmap), pageSegMode);
return(page.GetText());
}
catch
{
throw;
}
finally
{
if (bitmap != null)
{
bitmap.Dispose();
}
if (page != null)
{
page.Dispose();
}
if (tesseractEngine != null)
{
tesseractEngine.Dispose();
}
}
}
public void Dispose()
{
lock (this)
{
if (engine != null)
{
//if (cachedPage != null)
//{
// try
// {
// cachedPage.Dispose();
// }
// catch (Exception e)//for some reason: Attempted to read or write protected memory. This is often an indication that other memory is corrupt.
// {
// }
// cachedPageBitmap = null;
// cachedPage = null;
//}
engine.Dispose();
engine = null;
}
}
}
public IEnumerable <string> IdentifyData(IEnumerable <Rectangle> coOrdinates)
{
List <string> data = new List <string>();
try
{
var img = new Bitmap(pngImage);
img.SetResolution(300, 300);
foreach (Rectangle rec in coOrdinates)
{
var ocr = new TesseractEngine(ConfigurationManager.AppSettings.Get("TessarecDirectory"), "eng", EngineMode.Default);
var page = ocr.Process(img, new Rect(rec.X, rec.Y, rec.Width, rec.Height));
data.Add(page.GetText().Trim());
ocr.Dispose();
}
img.Dispose();
}
catch (Exception ex)
{
throw ex;
}
return(data);
}
public void PerformOcr(List <Tuple <int, int> > textRowLocations)
{
int DarkPixels;
var engine = new TesseractEngine(Program.GetDataPath(@"tessdata"), Program.DBCon.getIniValue <String>(IBE.IBESettingsView.DB_GROUPNAME, "TraineddataFile"), EngineMode.Default);
engine.DefaultPageSegMode = PageSegMode.SingleLine;
string Stationname_OCR;
string StationameAnalysisBase;
// delete the old brainerous images - otherwise Brainerous will process older but not relevant images too
if (Directory.Exists(Program.GetDataPath(@"Brainerous\images")))
{
foreach (string file in Directory.GetFiles(Program.GetDataPath(@"Brainerous\\images"), "*.*"))
{
File.Delete(file);
}
}
else
{
Directory.CreateDirectory(Program.GetDataPath(@"\Brainerousimages"));
}
float level;
var text = AnalyseFrameUsingTesseract(_bTrimmedHeader, engine, out level);
Stationname_OCR = StripPunctuationFromScannedText(text);// (text + " {" + page.GetMeanConfidence() + "}\r\n");
string[] StationsInSystem = Program.Data.getStations(SystemAtTimeOfScreenshot);
if (Program.actualCondition.Station.Equals("", StringComparison.InvariantCultureIgnoreCase))
{
StationameAnalysisBase = Stationname_OCR;
}
else
{
StationameAnalysisBase = Program.actualCondition.Station;
}
string headerResult_temp = StationsInSystem.FirstOrDefault(x => x.Equals(StationameAnalysisBase, StringComparison.InvariantCultureIgnoreCase));
if (headerResult_temp == null)
{
// station not found in database
var matchesInStationReferenceList = StationsInSystem.OrderBy(x => _levenshtein.LD2(Stationname_OCR, x)).ToList();
if (matchesInStationReferenceList.Count > 0)
{
var ld = _levenshtein.LD2(Stationname_OCR, matchesInStationReferenceList[0].ToUpper());
// this depends on the length of the word - this factor works really good
double LevenshteinLimit = Math.Round((matchesInStationReferenceList[0].Length * 1.0), 0);
if (ld <= LevenshteinLimit)
{
Stationname_OCR = matchesInStationReferenceList[0];
}
}
}
else
{
Stationname_OCR = headerResult_temp;
}
// show station on GUI
_callingForm.cOcrCaptureAndCorrect.DisplayResults(Stationname_OCR);
var commodityColumnText = new string[textRowLocations.Count(), 8];
var originalBitmaps = new Bitmap[textRowLocations.Count(), 8];
var originalBitmapConfidences = new float[textRowLocations.Count(), 8];
var rowIds = new string[textRowLocations.Count()];
var rowCtr = 0;
var bitmapCtr = 0;
foreach (var row in textRowLocations)
{
int startRow = row.Item1 - 3;
int heightRow = row.Item2 - row.Item1 + 6;
if (startRow < 0)
{
startRow = 0;
}
if (heightRow + startRow > _bTrimmed_4_OCR.Height)
{
heightRow = _bTrimmed_4_OCR.Height - startRow;
}
// We'll use this later to identify the right correction image
rowIds[rowCtr] = Guid.NewGuid().ToString();
using (Bitmap b = RNGraphics.Crop(_bTrimmed_4_OCR, new Rectangle(0, startRow, _bTrimmed_4_OCR.Width, heightRow)))
{
b.Save(Program.GetDataPath(@"OCR Correction Images\" + rowIds[rowCtr] + ".png"));
}
int columnCounter = 0;
while (columnCounter < 8)
{
int left, width;
switch (columnCounter)
{
case 0:
// commodity
left = 0; width = _calibrationPoints[3].X - _calibrationPoints[2].X;
break;
case 1:
// sell
left = _calibrationPoints[3].X - _calibrationPoints[2].X; width = _calibrationPoints[4].X - _calibrationPoints[3].X;
break;
case 2:
//buy
left = _calibrationPoints[4].X - _calibrationPoints[2].X; width = _calibrationPoints[5].X - _calibrationPoints[4].X;
break;
case 3:
// freight
left = _calibrationPoints[5].X - _calibrationPoints[2].X; width = _calibrationPoints[6].X - _calibrationPoints[5].X;
break;
case 4:
// demand
left = _calibrationPoints[6].X - _calibrationPoints[2].X; width = _calibrationPoints[7].X - _calibrationPoints[6].X;
break;
case 5:
// demand level
left = _calibrationPoints[7].X - _calibrationPoints[2].X; width = _calibrationPoints[8].X - _calibrationPoints[7].X;
break;
case 6:
// supply
left = _calibrationPoints[8].X - _calibrationPoints[2].X; width = _calibrationPoints[9].X - _calibrationPoints[8].X;
break;
case 7:
// supply level
left = _calibrationPoints[9].X - _calibrationPoints[2].X; width = _calibrationPoints[10].X - _calibrationPoints[9].X;
break;
default:
left = 0; width = _calibrationPoints[3].X - _calibrationPoints[2].X;
break;
}
var fudgeFactor = 0;// _bOriginal.Height * 6 / 1440;
left = left + fudgeFactor;
width = width - fudgeFactor;
DarkPixels = 0;
if (Program.DBCon.getIniValue <Boolean>(IBESettingsView.DB_GROUPNAME, "CheckNextScreenshotForOne", false.ToString(), false, true))
{
if (PixelTest == null)
{
PixelTest = new EBPixeltest();
}
if (columnCounter == 3)
{
var brainerousOut = RNGraphics.Crop(_bTrimmed_4_OCR, new Rectangle(left, startRow, width, heightRow));
// check how much dark pixels are on the bitmap
for (int i = 0; i < brainerousOut.Height; i++)
{
for (int j = 0; j < brainerousOut.Width; j++)
{
if (brainerousOut.GetPixel(j, i).GetBrightness() < Program.DBCon.getIniValue <Int32>(IBE.IBESettingsView.DB_GROUPNAME, "EBPixelThreshold"))
{
DarkPixels++;
}
}
}
PixelTest.addPicture(brainerousOut, DarkPixels);
}
}
else
{
// RNGraphics.Crop a little bit more form the left border because sometimes if theres
// the line of the table it was recognized as "1" or "7"
left += 10;
width -= 10;
if (columnCounter != 0 && columnCounter != 5 && columnCounter != 7)
{ //If it's a numeric column write it out for Brainerous to process later
var brainerousOut = RNGraphics.Crop(_bTrimmed_4_OCR, new Rectangle(left, startRow, width, heightRow));
if (Program.DBCon.getIniValue <Int32>(IBE.IBESettingsView.DB_GROUPNAME, "EBPixelAmount") > 0)
{
// check how much dark pixels are on the bitmap -> we process only bitmaps
// with something on it (minimum one digit supposed, a "1" hat about 25 pixels in default 1920x1200)
for (int i = 0; i < brainerousOut.Height; i++)
{
for (int j = 0; j < brainerousOut.Width; j++)
{
if (brainerousOut.GetPixel(j, i).GetBrightness() < Program.DBCon.getIniValue <Int32>(IBE.IBESettingsView.DB_GROUPNAME, "EBPixelThreshold"))
{
DarkPixels++;
}
}
}
}
if (DarkPixels >= Program.DBCon.getIniValue <Int32>(IBE.IBESettingsView.DB_GROUPNAME, "EBPixelAmount"))
{
brainerousOut.Save(Program.GetDataPath(@"Brainerous\images\" + bitmapCtr + ".png"));
}
bitmapCtr++;
}
else
{ // It's a text column, we'll use Tesseract
// Prepare some different versions of the bitmap, we will take the best result
var c = new Bitmap[7];
c[0] = (RNGraphics.Crop(_bTrimmed_4_OCR, new Rectangle(left, startRow, width, heightRow)));
c[1] = (RNGraphics.Crop(_bTrimmed_4_OCR, new Rectangle(left + 1, startRow, width, heightRow)));
c[2] = (RNGraphics.Crop(_bTrimmed_4_OCR, new Rectangle(left - 1, startRow, width, heightRow)));
c[3] = (RNGraphics.Crop(_bTrimmed_4_OCR, new Rectangle(left, startRow - 1, width, heightRow)));
c[4] = (RNGraphics.Crop(_bTrimmed_4_OCR, new Rectangle(left + 1, startRow - 1, width, heightRow)));
c[5] = (RNGraphics.Crop(_bTrimmed_4_OCR, new Rectangle(left - 1, startRow - 1, width, heightRow)));
c[6] = (RNGraphics.Crop(_bTrimmed_4_OCR, new Rectangle(left, startRow + 2, width, heightRow - 2)));
var t = new string[c.Length];
var cf = new float[c.Length];
for (int i = 0; i < c.Length; i++)
{
t[i] = AnalyseFrameUsingTesseract((Bitmap)(c[i].Clone()), engine, out cf[i]);
}
int result = 0;
float confidence = cf[0];
for (int i = 1; i < c.Length; i++)
{
if (confidence < cf[i])
{
result = i; confidence = cf[i];
}
}
originalBitmaps[rowCtr, columnCounter] = (Bitmap)(c[result].Clone());
switch (columnCounter)
{
//bodges for number columns
case 1:
case 2:
case 3:
t[result] = t[result].Replace(" ", "").Replace("O", "0").Replace("I", "1").Replace("'", "");
t[result] = System.Text.RegularExpressions.Regex.Replace(t[result], @"[a-zA-Z\s]+", string.Empty); // remove any alphas that remain
break;
case 5:
case 7:
t[result] = t[result].Replace(" ", "").Replace("-", "");
if (t[result] == "HIGH" || t[result] == "MED" || t[result] == "LOW")
{
cf[result] = 1;
}
break;
}
if ((columnCounter == 5 && t[result].Contains("ENTER")) ||
(columnCounter == 6 && (t[result].Contains("NGAR") || t[result].Contains("SURFACE"))))
{
t[result] = "";
cf[result] = 1;
}
commodityColumnText[rowCtr, columnCounter] += t[result];
originalBitmapConfidences[rowCtr, columnCounter] = cf[result];
}
}
columnCounter++;
}
rowCtr++;
}
if (Program.DBCon.getIniValue <Boolean>(IBESettingsView.DB_GROUPNAME, "CheckNextScreenshotForOne", false.ToString(), false, true))
{
PixelTest.StartModal(_callingForm);
}
else
{
if (textRowLocations.Count > 0)
{
// Call out to Brainerous to process the numeric bitmaps we saved earlier
var outputFromBrainerous = "";
var pr = new Process();
pr.StartInfo.UseShellExecute = false;
pr.StartInfo.CreateNoWindow = true;
pr.StartInfo.RedirectStandardOutput = true;
pr.StartInfo.FileName = Program.GetDataPath(@"Brainerous\nn_training.exe");
pr.StartInfo.WorkingDirectory = Program.GetDataPath("Brainerous");
pr.Start();
outputFromBrainerous = pr.StandardOutput.ReadToEnd();
while (outputFromBrainerous.Contains("Failed to pad successfully"))
{
var o2 = outputFromBrainerous.IndexOf("Failed to ");
var o3 = outputFromBrainerous.Substring(0, o2);
var o4 = outputFromBrainerous.Substring(o2).IndexOf(Program.GetDataPath("images"), StringComparison.InvariantCultureIgnoreCase);
// I had a string with "Failed to pad successfully" and only some trash behind but no "./images"
// so "o4" was "-1" and this results in strange behaviour
if (o4 > 0)
{
var o5 = outputFromBrainerous.Substring(o2 + o4);
outputFromBrainerous = o3 + "\r\n" + o5;
}
else
{
outputFromBrainerous = o3;
}
}
pr.WaitForExit();
List <string> splitOutput = ((string[])outputFromBrainerous.Replace("\r", "").Split('\n')).ToList();
for (var i = 0; i < (textRowLocations.Count * 10); i += 2)
{
string Filename = (i / 2).ToString() + ".png";
if ((splitOutput.Count <= i) || (splitOutput[i].Length < 14) || (splitOutput[i].Substring(9) != Filename))
{
splitOutput.Insert(i, Program.GetDataPath(@"images\" + Filename));
splitOutput.Insert(i + 1, "");
}
}
// Load the result from Brainerous into the OCR output
for (var i = 0; i < textRowLocations.Count; i++)
{
commodityColumnText[i, 1] = splitOutput[i * 10 + 1];
originalBitmaps[i, 1] = null;
originalBitmapConfidences[i, 1] = 1;
commodityColumnText[i, 2] = splitOutput[i * 10 + 3];
originalBitmaps[i, 2] = null;
originalBitmapConfidences[i, 2] = 1;
commodityColumnText[i, 3] = splitOutput[i * 10 + 5];
originalBitmaps[i, 3] = null;
originalBitmapConfidences[i, 3] = 1;
commodityColumnText[i, 4] = splitOutput[i * 10 + 7];
originalBitmaps[i, 4] = null;
originalBitmapConfidences[i, 4] = 1;
commodityColumnText[i, 6] = splitOutput[i * 10 + 9];
originalBitmaps[i, 6] = null;
originalBitmapConfidences[i, 6] = 1;
}
}
}
_bOriginal.Dispose();
_bOriginalClone.Dispose();
engine.Dispose();
if (Program.DBCon.getIniValue <Boolean>(IBESettingsView.DB_GROUPNAME, "CheckNextScreenshotForOne", false.ToString(), false, true))
{
Program.DBCon.setIniValue(IBESettingsView.DB_GROUPNAME, "CheckNextScreenshotForOne", false.ToString());
Form1.InstanceObject.cOcrCaptureAndCorrect.clearOcrOutput();
}
else
{
// Send the results for this screenshot back to the Form
_callingForm.cOcrCaptureAndCorrect.DisplayCommodityResults(commodityColumnText, originalBitmaps, originalBitmapConfidences, rowIds, CurrentScreenshot);
}
// ...and if we've got any buffered screenshots waiting to be processed, process the next one
if (ScreenshotBuffer.Count > 0)
{
var screenshot = ScreenshotBuffer[0];
ScreenshotBuffer.Remove(screenshot);
ProcessNewScreenshot(screenshot);
}
Working = false;
Debug.WriteLine("set to " + Working);
}
protected override void Cleanup()
{
engine.Dispose();
MeanConfidences = null;
}
//получение всех данные по странице с помощью масштабирования страницы с разным шагом(исп-ть, если не находит на исходном масштабе)
//sourceImage = входное изображение, stepScale = шаг масштабирования, amountOfSteps = кол-во проходов рескейлинга,
//infoPage = ссылка на структуру для заполнения
public void getData(Image sourceImage, ref Tesseract_OCR_Window.pdfPageInfo infoPage, ref Dictionary <string, string> replaceWords, float stepScale, int amountOfSteps)
{
//подключаем распознаватель
TesseractEngine ocrRus = new TesseractEngine("./tessdata", "rus", EngineMode.Default);
TesseractEngine ocrEng = new TesseractEngine("./tessdata", "eng", EngineMode.Default);
//создаем графический фильтр
TResizeTool resizeTool = new TResizeTool();
//ставим методику распознавания страницы
PageSegMode pSegMode = PageSegMode.SparseText;
string seriaNumber = ""; //серия
string country = ""; //страна
string orgSert = ""; //организация
string shortNumber = ""; //short number
string typeSeria = ""; //тип серии
float scaleValue = 1.0f; //исходный масштаб
for (int k = 0; k <= amountOfSteps; k++)
{
//по индексу отмасш-й страницы берем исходную(без масштаба)
Image img = sourceImage;
int newWidth = Convert.ToInt32((img.Width) * ((k * stepScale) + scaleValue));
int newHeight = Convert.ToInt32((img.Height) * ((k * stepScale) + scaleValue));
Bitmap imgBitmap = resizeTool.resizeImage(img, newWidth, newHeight);
Image rescaledPage = ((Image)imgBitmap);
string rescaledPageTextRus = "";
string rescaledPageTextEng = "";
//удаляем процесс, чтобы не было ошибки
using (var pageRus = ocrRus.Process(imgBitmap, pSegMode))
{
rescaledPageTextRus = pageRus.GetText();
}
//удаляем процесс, чтобы не было ошибки
using (var pageEng = ocrEng.Process(imgBitmap, pSegMode))
{
rescaledPageTextEng = pageEng.GetText();
}
//собираем данные для структуры infoPage
if (infoPage.seriaNumber_ == "")
{
seriaNumber = getSeriaNumber(rescaledPageTextRus, rescaledPageTextEng);//серия
if (seriaNumber != "")
{
infoPage.seriaNumber_ = seriaNumber;
}
}
if (infoPage.country_ == "")
{
country = getCountry(rescaledPageTextEng);
if (country != "")
{
infoPage.country_ = country;
}
}
if (infoPage.orgSert_ == "")
{
orgSert = getOrgSert(rescaledPageTextRus);
if (orgSert != "")
{
infoPage.orgSert_ = orgSert;
}
}
if (infoPage.shortNumber_ == "")
{
shortNumber = getShortNumber(rescaledPageTextRus);
if (shortNumber != "")
{
infoPage.shortNumber_ = shortNumber;
}
}
if (infoPage.typeSeria_ == "")
{
typeSeria = getTypeSeria(rescaledPageTextRus, rescaledPageTextEng, ref replaceWords);
if (typeSeria != "")
{
infoPage.typeSeria_ = typeSeria;
}
}
}
ocrRus.Dispose();
ocrEng.Dispose();
}
public void getDataViaRescaling(Image sourceImage, ref Tesseract_OCR_Window.pdfPageInfo infoPage, ref Dictionary <string, string> replaceWords, float stepScale, int amountOfSteps)
{
//подключаем распознаватель
TesseractEngine ocrRus = new TesseractEngine("./tessdata", "rus", EngineMode.Default);
TesseractEngine ocrEng = new TesseractEngine("./tessdata", "eng", EngineMode.Default);
//создаем графический фильтр
TResizeTool resizeTool = new TResizeTool();
//ставим методику распознавания страницы
PageSegMode pSegMode = PageSegMode.SparseText;
float scaleValue = 1.0f;//исходный масштаб
List <string> seriaNumberList = new List <string>();
List <string> countryList = new List <string>();
List <string> orgSertList = new List <string>();
List <string> shortNumberList = new List <string>();
List <string> typeSeriaList = new List <string>();
for (int k = 0; k <= amountOfSteps; k++)
{
//по индексу отмасш-й страницы берем исходную(без масштаба)
Image img = sourceImage;
int newWidth = Convert.ToInt32((img.Width) * ((k * stepScale) + scaleValue));
int newHeight = Convert.ToInt32((img.Height) * ((k * stepScale) + scaleValue));
Bitmap imgBitmap = resizeTool.resizeImage(img, newWidth, newHeight);
Image rescaledPage = ((Image)imgBitmap);
string rescaledPageTextRus = "";
string rescaledPageTextEng = "";
//удаляем процесс, чтобы не было ошибки
using (var pageRus = ocrRus.Process(imgBitmap, pSegMode))
{
rescaledPageTextRus = pageRus.GetText();
}
//удаляем процесс, чтобы не было ошибки
using (var pageEng = ocrEng.Process(imgBitmap, pSegMode))
{
rescaledPageTextEng = pageEng.GetText();
}
//набираем статистику для последующего поиска наибольшего соответствия
string seriaNumber = getSeriaNumber(rescaledPageTextRus, rescaledPageTextEng);
string country = getCountry(rescaledPageTextEng);
string orgSert = getOrgSert(rescaledPageTextRus);
string shortNumber = getShortNumber(rescaledPageTextRus);
string typeSeria = getTypeSeria(rescaledPageTextRus, rescaledPageTextEng, ref replaceWords);
if (seriaNumber != "")
{
seriaNumberList.Add(seriaNumber);//серия
}
if (country != "")
{
countryList.Add(country);//страна
}
if (orgSert != "")
{
orgSertList.Add(orgSert);//орган по сертификации
}
if (shortNumber != "")
{
shortNumberList.Add(shortNumber);//shortNumber
}
if (typeSeria != "")
{
typeSeriaList.Add(typeSeria);//тип серии
}
}
ocrRus.Dispose();
ocrEng.Dispose();
//собираем данные для структуры infoPage
infoPage.seriaNumber_ = highestMatch(seriaNumberList);
infoPage.country_ = highestMatch(countryList);
infoPage.orgSert_ = highestMatch(orgSertList);
infoPage.shortNumber_ = highestMatch(shortNumberList);
infoPage.typeSeria_ = highestMatch(typeSeriaList);
}
public static void RecognizeBill(object obj)
{
object[] objs = (object[])obj;
MainForm mform = (MainForm)objs[0];
ori_img = (Bitmap)Bitmap.FromFile((string)objs[1]);
RecognizeForm rform = new RecognizeForm();
ShowRecognizeForm(mform, rform);
mform.UpdateText("Clear Edges");
EdgeFilter();
rform.UpdateImage(show_img);
mform.UpdateLog("Clear Edges");
mform.UpdateText("Execute Sobel Filter");
SobelFilter();
rform.UpdateImage(show_img);
mform.UpdateLog("Execute Sobel Filter");
mform.UpdateText("Find target Rectangles");
FindRectangle(10, 20);
rform.UpdateImage(show_img);
mform.UpdateLog("Find target Rectangles");
mform.UpdateText("Cut image");
CutImage();
mform.UpdateLog("Cut image");
mform.UpdateText("Recognize address");
TesseractEngine ocr = new TesseractEngine(@"C:\Users\Allen Chou\Documents\Visual Studio 2013\Projects\Finder\packages\Tesseract.3.0.2.0\tessdata", "chi_tra+eng", EngineMode.Default);
Pix img = PixConverter.ToPix(targets[1]);
Page addpage = ocr.Process(img);
string address = addpage.GetText().Trim().Replace(" ", String.Empty);
ocr.Dispose();
mform.UpdateLog("Recognize address");
mform.UpdateText("Recognize eid, date, price");
Pix idpimg = PixConverter.ToPix(targets[2]);
TesseractEngine ocre = new TesseractEngine(@"C:\Users\Allen Chou\Documents\Visual Studio 2013\Projects\Finder\packages\Tesseract.3.0.2.0\tessdata", "eng", EngineMode.Default);
Page idppage = ocre.Process(idpimg);
string[] idpdata = idppage.GetText().Trim().Split(' ');
int tar = 0;
string eid = "";
for (int i = 0; i < idpdata.Length; i++)
{
Regex rex = new Regex("\\d{2}-\\d{2}-\\d{4}-\\d{2}-\\d{1}");
if (rex.IsMatch(idpdata[i]))
{
tar = i;
Match match = rex.Match(idpdata[i]);
eid = match.Value;
break;
}
}
ocre.Dispose();
string date = idpdata[tar + 1];
string price = idpdata[tar + 2].Replace("*", String.Empty);
mform.UpdateLog("Recognize eid, date, price\n");
mform.UpdateText("Recognize kWh");
Pix kwhimg = PixConverter.ToPix(targets[0]);
ocre = new TesseractEngine(@"C:\Users\Allen Chou\Documents\Visual Studio 2013\Projects\Finder\packages\Tesseract.3.0.2.0\tessdata", "eng", EngineMode.Default);
Page kwhpage = ocre.Process(kwhimg);
string[] kwhdata = kwhpage.GetText().Trim().Split(' ');
string kwh = "";
for (int i = 0; i < kwhdata.Length; i++)
{
Regex rex = new Regex("\\*\\d{1,}");
if (rex.IsMatch(kwhdata[i]))
{
Match match = rex.Match(kwhdata[i]);
kwh = match.Value.Replace("*", String.Empty);
break;
}
}
ocre.Dispose();
mform.UpdateLog("Recognize kWh");
mform.UpdateLog("Result [ " + eid + " , " + date + " , " + price + " , " + kwh + " , " + address + " ]");
mform.UpdateText("Finished");
mform.UpdateLog("Finished");
}
public void Dispose()
{
_engine.Dispose();
}
protected override void DisposeObject()
{
_ocr.Dispose();
}
public void pic_orc()
{
if (!Directory.Exists(path + "/new/"))
{
Directory.CreateDirectory(path + "/new/");
}
int n = files.Count();
System.Diagnostics.Stopwatch sw1 = new System.Diagnostics.Stopwatch();
sw1.Start();
int tempn = 0;//记录n
for (int i = 0; i < n; i++)
{
System.Diagnostics.Stopwatch sw = new System.Diagnostics.Stopwatch();
string ans = files[i].Replace(path, "");
sw.Start();
//识别图片文字
var img = new Bitmap(files[i]);
//显示
if (this.pictureBox2.InvokeRequired)//不同线程为true，所以这里是true
{
BeginInvoke(new Action(() => {
if (this.pictureBox2.Image != null)
{
this.pictureBox2.Image.Dispose();
}
this.pictureBox2.Height = (int)this.pictureBox2.Width * cH / cW;
this.pictureBox2.Image = (Cut(img, cX, cY, cW, cH));
}));
}
if (this.pictureBox1.InvokeRequired)//不同线程为true，所以这里是true
{
BeginInvoke(new Action(() => {
if (this.pictureBox1.Image != null)
{
this.pictureBox1.Image.Dispose();
}
this.pictureBox1.Load(files[i]);
this.label6.Text = (currentnum + 1).ToString();
}));
}
var ocr = new TesseractEngine("./tessdata", "chi_sim", EngineMode.TesseractOnly);
ocr.SetVariable("tessedit_char_whitelist", "0123456789"); //设置识别变量，当前只能识别数字。
var page = ocr.Process(BlodBitmap(Cut(img, cX, cY, cW, cH))); //100, 584, 3687, 713
//var page = ocr.Process(Cut(img, 1875, 579, 1917, 693));
string ocr_text = page.GetText().Replace(" ", "").Replace("\n", "");
if (this.textBox1.InvokeRequired)//不同线程为true，所以这里是true
{
BeginInvoke(new Action(() => {
this.textBox1.Text = ocr_text;
}));
}
page.Dispose();
img.Dispose();
ocr.Dispose();
int l = ocr_text.Length;
int acc = 0;
if (l >= 10)
{
int start = ocr_text.IndexOf(codestart.ToString());
if (start < 0)//长度够，没有关键字
{
acc = 50;
ans += " >" + acc.ToString() + "%，编号：" + code1 + "_" + num.ToString();
//list.Add(filename + "-" + code);
code = code1 + "_" + num.ToString();
num++;
}
else
{
code = ocr_text.Substring(start, 10); //更新
code1 = code;
if (code.Length == 10)
{
acc = 100;//长度够，有关键字
ans += " >" + acc.ToString() + "% 编号：" + code;
}
else
{
acc = 90;//裁剪后长度不够，有关键字！！！
ans += " >" + acc.ToString() + "% 编号：" + code;
}
num = 0;
num++;
}
}
else
{
int start = ocr_text.IndexOf(codestart.ToString());
if (start < 0)//长度不够，也没有找到关键字
{
acc = 0;
ans += " >" + acc.ToString() + "%，编号：" + code1 + "_" + num.ToString();
//list.Add(filename + "-" + code);
code = code1 + "_" + num.ToString();
num++;
}
else//长度不够，找到关键字
{
acc = 70;
ans += " >" + acc.ToString() + "%，编号：" + code1 + "_" + num.ToString();
//list.Add(filename + "-" + code);
code = code1 + "_" + num.ToString();
num++;
}
}
sw.Stop();
TimeSpan ts2 = sw.Elapsed;
string dettime = (ts2.TotalMilliseconds / 1000.0).ToString("0.00") + "秒";
ans += " 耗时：" + dettime + "\r\n";
ocr_text = null;
if (this.textBox2.InvokeRequired)//不同线程为true，所以这里是true
{
BeginInvoke(new Action(() => { this.textBox2.Text += ans; this.textBox2.SelectionStart = this.textBox2.TextLength; this.textBox2.ScrollToCaret(); }));
}
if (this.label4.InvokeRequired)//不同线程为true，所以这里是true
{
BeginInvoke(new Action(() => { this.label4.Text = (i + 1).ToString() + "/" + n.ToString(); }));
}
if (this.progressBar1.InvokeRequired)//不同线程为true，所以这里是true
{
BeginInvoke(new Action(() => { this.progressBar1.Value += this.progressBar1.Step; }));
}
dt.Rows.Add(files[i], code, dettime.ToString(), acc);
Thread thread1 = new Thread(() => updatelistview(i, code, dettime.ToString(), acc.ToString()));
thread1.Start();
if (this.label1.InvokeRequired)//不同线程为true，所以这里是true
{
BeginInvoke(new Action(() => {
this.label1.Text = code;
}));
}
/*
* if (this.label6.InvokeRequired)//不同线程为true，所以这里是true
* {
* BeginInvoke(new Action(() => {
* this.label6.Text = (currentnum + 1).ToString();
* }));
* }
*/
try
{
if (dt.Rows[i]["acc"].ToString() == "100" || dt.Rows[i]["acc"].ToString() == "0")
{
System.IO.File.Copy(dt.Rows[i]["url"].ToString(), path + "/new/" + dt.Rows[i]["code"].ToString() + ".jpg", true);
File.AppendAllText(path + "/new/log.txt", "\r\n" + "复制" + dt.Rows[i]["url"].ToString().Replace(path, "") + " 到 " + dt.Rows[i]["code"].ToString() + ".jpg <" + dt.Rows[i]["acc"].ToString() + ">");
}
else
{
File.AppendAllText(path + "/new/log.txt", "\r\n" + "复制" + dt.Rows[i]["url"].ToString().Replace(path, "") + " 到 " + dt.Rows[i]["code"].ToString() + ".jpg失败 <" + dt.Rows[i]["acc"].ToString() + ">!!!");
needstop = true;
}
}
catch (Exception exp)
{
File.AppendAllText(path + "/new/error.txt", "\r\n" + "复制" + dt.Rows[i]["url"].ToString().Replace(path, "") + " 到 " + dt.Rows[i]["code"].ToString() + ".jpg失败，" + exp.ToString());
needstop = true;
}
tempn = i + 1;
//判断是否结束
if (needstop)
{
MessageBox.Show("识别出错，将结束剩下的操作！");
needstop = false;
break;
}
currentnum++;
}
if (this.button5.InvokeRequired)//不同线程为true，所以这里是true
{
BeginInvoke(new Action(() => { this.button5.Enabled = true; }));
}
if (this.button7.InvokeRequired)//不同线程为true，所以这里是true
{
BeginInvoke(new Action(() => { this.button7.Enabled = true; }));
}
if (this.button8.InvokeRequired)//不同线程为true，所以这里是true
{
BeginInvoke(new Action(() => { this.button8.Enabled = true; }));
}
if (this.button6.InvokeRequired)//不同线程为true，所以这里是true
{
BeginInvoke(new Action(() => { this.button6.Enabled = true; }));
}
if (this.button9.InvokeRequired)//不同线程为true，所以这里是true
{
BeginInvoke(new Action(() => { this.button9.Enabled = true; }));
}
//MessageBox.Show("识别完毕");
sw1.Stop();
TimeSpan ts3 = sw1.Elapsed;
if (this.textBox2.InvokeRequired)//不同线程为true，所以这里是true
{
BeginInvoke(new Action(() => { this.textBox2.Text += "共耗时：" + (ts3.TotalMilliseconds / 1000.0).ToString("0.00") + "秒";; this.textBox2.SelectionStart = this.textBox2.TextLength; this.textBox2.ScrollToCaret(); }));
}
File.AppendAllText(path + "/new/log.txt", "\r\n" + "共耗时：" + (ts3.TotalMilliseconds / 1000.0).ToString("0.00") + "秒");
MessageBox.Show("复制完毕");
needstop = false;
//回复listview
}
public void orconce(string filename)
{
//识别图片文字
var img = new Bitmap(filename);
var ocr = new TesseractEngine("./tessdata", "chi_sim", EngineMode.Default);
ocr.SetVariable("tessedit_char_whitelist", "0123456789"); //设置识别变量，当前只能识别数字。
var page = ocr.Process(BlodBitmap(Cut(img, cX, cY, cW, cH)));
string ocrtext = page.GetText().Replace(" ", "").Replace("\n", "");
//using (var page = ocr.Process(Cut(img, 1869, 561, 1865, 661)))
{
if (this.textBox1.InvokeRequired)//不同线程为true，所以这里是true
{
BeginInvoke(new Action(() => {
this.textBox1.Text = ocrtext;
}));
}
else
{
this.textBox1.Text = ocrtext;
}
page.Dispose();
ocr.Dispose();
img.Dispose();
}
int l = this.textBox1.Text.Length;
if (l >= 10)
{
int start = this.textBox1.Text.IndexOf(codestart.ToString());
if (start < 0)
{
if (this.label1.InvokeRequired)//不同线程为true，所以这里是true
{
BeginInvoke(new Action(() =>
{
this.label1.Text = "无，50%，长度够，但没有找到关键字";
}));
}
else
{
this.label1.Text = "无，50%，长度够，但没有找到关键字";
}
}
else
{
if (this.label1.InvokeRequired)//不同线程为true，所以这里是true
{
BeginInvoke(new Action(() => {
this.label1.Text = "订单编号：" + this.textBox1.Text.Substring(start, 10);
}));
}
else
{
this.label1.Text = "订单编号：" + this.textBox1.Text.Substring(start, 10);
}
}
}
else
{
int start = this.textBox1.Text.IndexOf(codestart.ToString());
if (start < 0)
{
if (this.label1.InvokeRequired)//不同线程为true，所以这里是true
{
BeginInvoke(new Action(() => {
this.label1.Text = "无 0%，长度不够，也没有找到关键字";
}));
}
else
{
this.label1.Text = "无 0%，长度不够，也没有找到关键字";
}
}
else
{
if (this.label1.InvokeRequired)//不同线程为true，所以这里是true
{
BeginInvoke(new Action(() => {
this.label1.Text = "无 70%，长度不够，但有关键字！！！！";
}));
}
else
{
this.label1.Text = "无 70%，长度不够，但有关键字！！！！";
}
}
}
/*
* //查找订单编号
* int start = this.textBox1.Text.IndexOf("订单编号");
* if (start < 0)
* {
* this.label1.Text = "无";
* }
* else
* {
* this.label1.Text = "订单编号：" + this.textBox1.Text.Substring(start + 5, 10).Replace("〇", "0");
* }
*/
}
public void Dispose()
{
tesseractEngine.Dispose();
}
public void Dispose()
{
_tessEngine?.Dispose();
}
static void Main(string[] args)
{
Queue <int> answers = new Queue <int>();
Queue <int> question = new Queue <int>();
PicConverter converter = new PicConverter();
Clicker clicker = new Clicker();
Stopwatch sw = new Stopwatch();
var ocr1 = new TesseractEngine(@"./tessdata", "eng", EngineMode.TesseractAndCube);
var ocr2 = new TesseractEngine(@"./tessdata", "eng", EngineMode.TesseractAndCube);
int correctAnswer = 0;
Console.WriteLine("Введите что-нибудь для старта");
Console.ReadLine();
sw.Start();
try
{
//берём картинку и делаем чёрно-белой
// метод CutImgFromScreen для скрина экрана
Bitmap image = new Bitmap(@"C:\Users\gvozd\source\repos\Ocr1\Ocr1\NewFolder1\image.png");
Bitmap imageBW = converter.BitmapToBlackWhite2(image, 0.9);
//Bitmap image = new Bitmap(b1, new Size(450, 150)); //если нужно ресайзим
imageBW.Save(@"D:\image.png");
//вырезаем вопрос и ответы
Bitmap questionImg = converter.CutImgFromImg(20, 8, 120, 38, imageBW, "question.png");
Bitmap stackAns = converter.StackImg(imageBW);
var task1 = Task.Factory.StartNew(() =>
{
Console.WriteLine("таск 1 начался");
var text1 = ocr1.Process(questionImg);
Console.WriteLine(text1.GetText());
PageToNumbers(text1, ref question);
for (int i = 0; i < 2; i++)
{
correctAnswer += question.Dequeue();
}
ocr1.Dispose();
Console.WriteLine("таск 1 закончился");
});
var task2 = Task.Factory.StartNew(() =>
{
Console.WriteLine("таск 2 начался");
var text2 = ocr2.Process(stackAns);
Console.WriteLine(text2.GetText());
PageToNumbers(text2, ref answers);
ocr2.Dispose();
Console.WriteLine("таск 2 закончился");
});
Task.WaitAll(task1, task2);
sw.Stop();
Console.WriteLine((sw.ElapsedMilliseconds / 1000.0).ToString() + " секунд");
clicker.DoCorrectClick(answers, correctAnswer);
}
catch (Exception exception)
{
Console.WriteLine(exception.Message);
}
Console.ReadKey();
}
public void Uninit()
{
_tesseract.Dispose();
}
public void Dispose()
{
_engine?.Dispose();
_engine = null;
}
private Slide[] BuildThumbnailsWin(ConversionConfiguration config, string slidesFileName)
{
var thumbOutDir = Path.Combine(config.OutputDirectory, "thumbs");
Directory.CreateDirectory(thumbOutDir);
var ocrEngine = new TesseractEngine(Path.Combine("resources", "tessdata"), "eng", EngineMode.Default);
List <Slide> result = new List <Slide>();
try
{
dynamic projectJson = JsonConvert.DeserializeObject(File.ReadAllText(config.MetadataPath));
int currentId = 0;
var keyframes = new List <TimeSpan>();
keyframes.Add(TimeSpan.Zero);
foreach (string timestamp in projectJson["slides"])
{
keyframes.Add(TimeSpan.Parse(timestamp));
}
foreach (var keyframe in keyframes)
{
TimeSpan?nextKeyframe = null;
if (keyframes.IndexOf(keyframe) != keyframes.Count - 1)
{
nextKeyframe = keyframes[keyframes.IndexOf(keyframe) + 1];
}
else
{
nextKeyframe = FFmpegHelper.GetMediaLength(config.SlideVideoPath);
}
string thumbName = FFmpegHelper.ExportThumbnail((float)nextKeyframe.GetValueOrDefault().TotalSeconds - 2.0f, config.SlideVideoPath, thumbOutDir,
(currentId++).ToString());
var slide = new Slide
{
StartPosition = (float)keyframe.TotalSeconds + 0.2f,
Thumbnail = "thumbs/" + thumbName,
Ocr = PerformOcr(Path.Combine(thumbOutDir, thumbName), ocrEngine)
};
if (keyframe.Equals(TimeSpan.Zero))
{
slide.StartPosition = 0.0f;
}
result.Add(slide);
}
}
catch (Exception ex)
{
_logger.LogError(ex.Message);
}
finally
{
ocrEngine.Dispose();
}
return(result.ToArray());
}