/// <summary>
/// pulls marked from param inputFilePath image
/// </summary>
/// <param name="fileBytes">Byte array of file data</param>
/// <returns>encodes text</returns>
public static string getText(byte[] fileBytes)
{
string text = "", rootPath = HostingEnvironment.ApplicationPhysicalPath;
BitmapToPixConverter b = new BitmapToPixConverter();
using (Stream memStream = new MemoryStream(fileBytes))
using (Bitmap image = (Bitmap)Image.FromStream(memStream))
using (TesseractEngine ocr = new TesseractEngine(rootPath, "eng", EngineMode.TesseractOnly))
{
image.SetResolution(300, 300);
ocr.SetVariable("tessedit_char_whitelist", "\\/|#");
Pix p = b.Convert(image);
p = p.ConvertRGBToGray();
Page page = ocr.Process(p, PageSegMode.Auto);
text = page.GetText();
p.Dispose();
page.Dispose();
}
return text;
}
private void OCRButton_Click(object sender, RoutedEventArgs e)
{
Bitmap commodityData = cutImage.ToBitmap();
bool processing = true;
int nonBlankRow = 0;
int blankRow = 0;
Window window;
while (processing)
{
nonBlankRow = findNonBlankRow(commodityData, blankRow + 1);
blankRow = findBlankRow(commodityData, nonBlankRow + 1);
if (nonBlankRow == -1 || blankRow == -1)
{
processing = false;
break;
}
int heightOfRow = blankRow - nonBlankRow + 10;
Bitmap newBitmap = new Bitmap(commodityData.Width, heightOfRow);
using (Graphics g = Graphics.FromImage(newBitmap))
{
g.DrawImage(commodityData, 0, -(nonBlankRow) + 5);
}
BitmapSource bmpSource = Utils.BitmapToBitmapSource(newBitmap);
DisplayImage.Source = bmpSource;
//window = new Window();
//StackPanel panel = new StackPanel();
//window.Content = panel;
//window.Width = newBitmap.Width;
//window.Height = newBitmap.Height;
//System.Windows.Controls.Image image = new System.Windows.Controls.Image();
//image.Source = bmpSource;
//panel.Children.Add(image);
//window.ShowDialog();
TesseractEngine tesseract = new TesseractEngine("tessdata", "big", EngineMode.Default);
var conv = new BitmapToPixConverter();
int counter = 0;
var p = conv.Convert(newBitmap);
Tesseract.Page page = tesseract.Process(p, PageSegMode.SingleLine);
using (var iter = page.GetIterator())
{
do
{
MessageBox.Show("Text: " + iter.GetText(PageIteratorLevel.Block) + " Confidence: " + iter.GetConfidence(PageIteratorLevel.Block));
counter++;
if(counter > 10)
{
break;
}
} while (iter.Next(PageIteratorLevel.Block));
}
page.Dispose();
newBitmap.Dispose();
tesseract.Dispose();
Thread.Sleep(1000);
}
//int nonBlankRow = findNonBlankRow(commodityData, 0);
//int blankRow = findBlankRow(commodityData, nonBlankRow+1);
//MessageBox.Show("Non-white: " + nonBlankRow + " White: " + blankRow);
//nonBlankRow = findNonBlankRow(commodityData, blankRow);
//blankRow = findBlankRow(commodityData, nonBlankRow + 1);
//MessageBox.Show("Non-white: " + nonBlankRow + " White: " + blankRow);
//using (tesseract)
//{
// Tesseract.Page ocrResult = tesseract.Process(cutImage.ToBitmap());
// StringBuilder builder = new StringBuilder();
// using (var iter = ocrResult.GetIterator())
// {
// int i = 1;
// do
// {
// var text = iter.GetText(PageIteratorLevel.TextLine);
// builder.Append("Line ").Append(i).Append(": ").Append(text).Append(" Confidence: ")
// .Append(iter.GetConfidence(PageIteratorLevel.TextLine)).AppendLine();
// i++;
// } while (iter.Next(PageIteratorLevel.TextLine));
// }
// MessageBox.Show("Text: '" + builder.ToString());
//}
}
private static string AnalyseFrameUsingTesseract(Bitmap c1, TesseractEngine engine, out float cf1)
{
var conv = new BitmapToPixConverter();
var p = conv.Convert(c1);
string t1;
using (var page = engine.Process(p))
{
t1 = page.GetText().Replace("\n\n", "").Replace(".", "").Replace(",", "");
cf1 = page.GetMeanConfidence();
}
return t1;
}