protected virtual void CheckCenterLine(Bitmap image)
{
var tmpImage = image.Copy(new Rectangle(0, 0, image.Width, image.Height));
var tmpImageData = tmpImage.LockBits(new Rectangle(new System.Drawing.Point(0, 0), tmpImage.Size),
ImageLockMode.ReadWrite,
tmpImage.PixelFormat);
Drawing.Line(tmpImageData,
new IntPoint(0, 0),
new IntPoint(0, tmpImageData.Height),
Color.White);
tmpImage.UnlockBits(tmpImageData);
OCRLessonReport.Imaging.HoughLineTransformation lineTransform = new OCRLessonReport.Imaging.HoughLineTransformation();
HoughLineRequestSettings vSettings = new HoughLineRequestSettings
{
HorizontalLines = false,
VerticalLines = true
};
var hWidth = tmpImage.Width / 2;
var hHeight = tmpImage.Height / 2;
lineTransform.ProcessImage(tmpImage, vSettings);
HoughLine[] vLines = lineTransform.GetLinesByRelativeIntensity(0.9);
var vLineCoordinates = vLines.Select(line => hWidth + line.Radius).Where(x => x > 1).OrderBy(x => x).ToArray();
if (vLineCoordinates.Length > 0)
{
var x0 = vLineCoordinates.FirstOrDefault() - 1;
var x1 = vLineCoordinates.LastOrDefault() + 2;
if (x1 - x0 > 10)
{
return;
}
var imageData = image.LockBits(new Rectangle(new System.Drawing.Point(0, 0), image.Size),
ImageLockMode.ReadWrite,
image.PixelFormat);
Drawing.FillRectangle(imageData, new Rectangle(x0, 0, x1 - x0, image.Height), Color.Black);
image.UnlockBits(imageData);
}
}
protected virtual double DetectRotation(Bitmap image)
{
//Hough line transformation
OCRLessonReport.Imaging.HoughLineTransformation lineTransform = new OCRLessonReport.Imaging.HoughLineTransformation();
HoughLineRequestSettings settings = new HoughLineRequestSettings
{
HorizontalLines = false,
VerticalLines = true,
VerticalDeviation = Settings.SheetEdgeVerticalMaxAngle
};
lineTransform.ProcessImage(image, settings);
HoughLine[] lines = lineTransform.GetLinesByRelativeIntensity(Settings.SheetEdgeSensitivity);
if (lines.Length < 1)
{
throw new Exception("Can't detect left edge");
}
Func <HoughLine, double> getAngle = line =>
{
double angle = 0;
if (line.Theta > 90 && line.Radius > 0)
{
angle = 180 - line.Theta;
}
else if (line.Theta > 90 && line.Radius < 0)
{
angle = 180 - line.Theta;
}
else if (line.Theta < 90 && line.Radius > 0)
{
angle = -line.Theta;
}
else if (line.Theta < 90 && line.Radius < 0)
{
angle = -line.Theta;
}
return(angle);
};
double avgAngle = lines.Select(l => getAngle(l)).Average();
return(avgAngle);
}
protected virtual Bitmap ProcessCell(Bitmap cellImg, bool check = false)
{
//Build horizontal hough lines
OCRLessonReport.Imaging.HoughLineTransformation lineTransform = new OCRLessonReport.Imaging.HoughLineTransformation();
HoughLineRequestSettings vSettings = new HoughLineRequestSettings
{
HorizontalLines = false,
VerticalLines = true
};
HoughLineRequestSettings hSettings = new HoughLineRequestSettings
{
HorizontalLines = true,
VerticalLines = false
};
lineTransform.ProcessImage(cellImg, vSettings);
HoughLine[] vLines = lineTransform.GetLinesByRelativeIntensity(Settings.CellSensitivity);
lineTransform.ProcessImage(cellImg, hSettings);
HoughLine[] hLines = lineTransform.GetLinesByRelativeIntensity(Settings.CellSensitivity);
var hWidth = cellImg.Width / 2;
var hHeight = cellImg.Height / 2;
var vLineCoordinates = vLines.Select(line => hWidth + line.Radius);
var hLineCoordinates = hLines.Select(line => hHeight - line.Radius);
var xEdges = GetCellEdges(vLineCoordinates, cellImg.Width, Settings.CellXEdgeWith, Settings.CellEdgeCutting);
var yEdges = GetCellEdges(hLineCoordinates, cellImg.Height, Settings.CellYEdgeWith, Settings.CellEdgeCutting);
cellImg = cellImg.Copy(new Rectangle(xEdges.Item1, yEdges.Item1, xEdges.Item2 - xEdges.Item1, yEdges.Item2 - yEdges.Item1));
//Check central line
if (check)
{
CheckCenterLine(cellImg);
}
return(cellImg);
}
public virtual void ProccessImage(BackgroundWorker worker)
{
this.worker = worker;
UpdateProgress(3);
//Brightness and sharpen filters
BrightnessCorrection cfilter = new BrightnessCorrection(Settings.Brightness);
GaussianSharpen filter = new GaussianSharpen(4, 11);
//Apply filters
cfilter.ApplyInPlace(sourceBitmap);
UpdateProgress(15);
filter.ApplyInPlace(sourceBitmap);
UpdateProgress(30);
//Convert to gray
var tmpImage = ConvertToGrayScale(sourceBitmap);
UpdateProgress(35);
//Cut edges
tmpImage = CutEdgesAndInvert(tmpImage);
UpdateProgress(40);
//Get angle for rotating image
var rotateAngle = DetectRotation(tmpImage);
UpdateProgress(45);
if (rotateAngle != 0)
{
RotateBilinear rotate = new RotateBilinear(rotateAngle, true);
tmpImage = rotate.Apply(tmpImage);
}
//Build horizontal hough lines
OCRLessonReport.Imaging.HoughLineTransformation lineTransform = new OCRLessonReport.Imaging.HoughLineTransformation();
HoughLineRequestSettings settings = new HoughLineRequestSettings
{
HorizontalLines = true,
VerticalLines = false,
HorizontalDeviation = 2
};
lineTransform.ProcessImage(tmpImage, settings);
//Get horizontal line
HoughLine[] lines = lineTransform.GetLinesByRelativeIntensity(Settings.HorizontalSensitivity);
//Get half width and height for future calculations
int hWidth = tmpImage.Width / 2;
int hHeight = tmpImage.Height / 2;
//Get line coordinates (Y axis only - horizontal lines)
var lineCoordinates = lines.Select(line => hHeight - line.Radius);
//Grouping coords by delta
var groupedCoordinates = ImagingHelper.GroupingCoordinates(lineCoordinates, Settings.LineGroupingDelta);
if (groupedCoordinates.Count <= Settings.HeaderStartLine)
{
throw new Exception("Invalid source. Can't be recognized");
}
int headerLineY0 = groupedCoordinates[Settings.HeaderStartLine];
int headerLineY1 = groupedCoordinates[Settings.HeaderStartLine + 1];
//Copy header to new image
var headerImage = tmpImage.Copy(new Rectangle(0, headerLineY0, tmpImage.Width, headerLineY1 - headerLineY0));
//Parse header to get header lines
HoughLineRequestSettings headerSettings = new HoughLineRequestSettings
{
HorizontalLines = false,
VerticalLines = true,
VerticalDeviation = 1
};
lineTransform.ProcessImage(headerImage, headerSettings);
Func <HoughLine, int, int> getRadius = (l, w) =>
{
if (l.Theta > 90 && l.Theta < 180)
{
return(w - l.Radius);
}
else
{
return(w + l.Radius);
}
};
HoughLine[] headerLines = lineTransform.GetLinesByRelativeIntensity(Settings.VerticalSensitivity);
//Get header vertical lines
var headerLineCoordinates = headerLines.Select(line => getRadius(line, hWidth));
//Grouped lines
var groupedheaderLineCoordinates = ImagingHelper.GroupingCoordinates(headerLineCoordinates, Settings.LineGroupingDelta);
//Build cell map
List <TableCell> cellMap = new List <TableCell>();
UpdateProgress(50);
//Use tess engine for ocr
using (TesseractEngine engine = new TesseractEngine(Settings.TessdataPath, Settings.TessdataLanguage))
{
//Parse top header
var x0 = groupedheaderLineCoordinates.FirstOrDefault();
var x1 = groupedheaderLineCoordinates.LastOrDefault();
var y0 = groupedCoordinates[0];
var y1 = groupedCoordinates[1];
int fullProgress = (groupedheaderLineCoordinates.Count - 1) * (groupedCoordinates.Count - Settings.BottomStartLine - 1 - Settings.HeaderStartLine);
int curProgress = 0;
var hImage = tmpImage.Copy(new Rectangle(x0, y0, x1 - x0, y1 - y0));
hImage = ProcessCell(hImage);
using (var page = engine.Process(hImage, PageSegMode.SingleBlock))
{
cellMap.Add(new TableCell(0, 0, TableCellType.MainHeader, hImage, page.GetText(), false));
}
//Parse table
for (int i = 0; i < groupedheaderLineCoordinates.Count - 1; i++)
{
int subjectArea = (i < Settings.ColumnSubjectStart - 1) ? 0 : 1;
for (int j = Settings.HeaderStartLine; j < groupedCoordinates.Count - Settings.BottomStartLine - 1; j++)
{
int headerArea = (j == Settings.HeaderStartLine) ? 2 : 0;
TableCellType cellType = (TableCellType)(subjectArea + headerArea);
var cellImg = tmpImage.Copy(new Rectangle(groupedheaderLineCoordinates[i], groupedCoordinates[j],
groupedheaderLineCoordinates[i + 1] - groupedheaderLineCoordinates[i],
groupedCoordinates[j + 1] - groupedCoordinates[j]));
if (cellType == TableCellType.Text || cellType == TableCellType.Header || cellType == TableCellType.HeaderRotated)
{
cellImg = ProcessCell(cellImg, i == Settings.NameStartLine);
string text = String.Empty;
if (cellType == TableCellType.HeaderRotated)
{
cellImg.RotateFlip(RotateFlipType.Rotate90FlipNone);
}
using (var page = engine.Process(cellImg, PageSegMode.SingleBlock))
{
text = page.GetText();
}
cellMap.Add(new TableCell(i, j, cellType, cellImg, text, false));
}
else
{
cellImg = ProcessCell(cellImg);
BilateralSmoothing bfilter = new BilateralSmoothing();
bfilter.KernelSize = 7;
bfilter.SpatialFactor = 10;
bfilter.ColorFactor = 60;
bfilter.ColorPower = 0.5;
bfilter.ApplyInPlace(cellImg);
cellImg = FilterColors(cellImg, Settings.FilteringColor, ByteColor.Black, ByteColor.White);
BlobCounter bcounter = new BlobCounter();
bcounter.ProcessImage(cellImg);
var blobs = bcounter.GetObjects(cellImg, false);
if (blobs.Length < 1)
{
continue;
}
var biggestBlob = blobs.OrderBy(b => b.Area).LastOrDefault();
var biggestBlobsImage = biggestBlob.Image.ToManagedImage();
cellMap.Add(new TableCell(i, j, cellType, biggestBlobsImage, GetMask(biggestBlobsImage).ToString(), GetMask(biggestBlobsImage)));
}
curProgress++;
double reportProgress = (double)curProgress / (double)fullProgress * 50 + 50;
UpdateProgress((int)reportProgress);
}
}
}
this.Cells = cellMap;
UpdateProgress(100);
}