private static List <NOcrPoint> ReadPointsBytes(Stream stream)
{
int length = stream.ReadByte();
var list = new List <NOcrPoint>(length);
var buffer = new byte[4];
for (int i = 0; i < length; i++)
{
stream.Read(buffer, 0, buffer.Length);
var point = new NOcrPoint
{
Start = new Point(buffer[0], buffer[1]),
End = new Point(buffer[2], buffer[3])
};
list.Add(point);
}
return(list);
}
public static List<NOcrChar> LoadNOcrForTesseract(string xmlRessourceName)
{
var nocrChars = new List<NOcrChar>();
Assembly asm = Assembly.GetExecutingAssembly();
Stream strm = asm.GetManifestResourceStream(xmlRessourceName);
if (strm != null)
{
XmlDocument doc = new XmlDocument();
var rdr = new StreamReader(strm);
using (var zip = new System.IO.Compression.GZipStream(rdr.BaseStream, System.IO.Compression.CompressionMode.Decompress))
{
byte[] data = new byte[175000];
zip.Read(data, 0, 175000);
doc.LoadXml(Encoding.UTF8.GetString(data));
}
rdr.Close();
try
{
foreach (XmlNode node in doc.DocumentElement.SelectNodes("Char"))
{
var oc = new NOcrChar(node.Attributes["Text"].Value);
oc.Width = Convert.ToInt32(node.Attributes["Width"].Value, CultureInfo.InvariantCulture);
oc.Height = Convert.ToInt32(node.Attributes["Height"].Value, CultureInfo.InvariantCulture);
oc.MarginTop = Convert.ToInt32(node.Attributes["MarginTop"].Value, CultureInfo.InvariantCulture);
if (node.Attributes["Italic"] != null)
oc.Italic = Convert.ToBoolean(node.Attributes["Italic"].Value, CultureInfo.InvariantCulture);
if (node.Attributes["ExpandCount"] != null)
oc.ExpandCount = Convert.ToInt32(node.Attributes["ExpandCount"].Value, CultureInfo.InvariantCulture);
foreach (XmlNode pointNode in node.SelectNodes("Point"))
{
var op = new NOcrPoint(DecodePoint(pointNode.Attributes["Start"].Value), DecodePoint(pointNode.Attributes["End"].Value));
XmlAttribute a = pointNode.Attributes["On"];
if (a != null && Convert.ToBoolean(a.Value))
oc.LinesForeground.Add(op);
else
oc.LinesBackground.Add(op);
}
nocrChars.Add(oc);
}
}
catch (Exception exception)
{
MessageBox.Show(exception.Message);
}
}
return nocrChars;
}
public static void GenerateLineSegments(int numberOfLines, bool veryPrecise, NOcrChar nOcrChar, NikseBitmap nbmp)
{
int giveUpCount = 10000;
var r = new Random();
int count = 0;
int hits = 0;
bool tempVeryPrecise = veryPrecise;
while (hits < numberOfLines && count < giveUpCount)
{
var start = new Point(r.Next(nOcrChar.Width), r.Next(nOcrChar.Height));
var end = new Point(r.Next(nOcrChar.Width), r.Next(nOcrChar.Height));
if (hits < 5 && count < 100) // a few large lines
{
for (int k = 0; k < 500; k++)
{
if (Math.Abs(start.X - end.X) + Math.Abs(start.Y - end.Y) > nOcrChar.Height / 2)
{
break;
}
else
{
end = new Point(r.Next(nOcrChar.Width), r.Next(nOcrChar.Height));
}
}
}
else // and a lot of small lines
{
for (int k = 0; k < 500; k++)
{
if (Math.Abs(start.X - end.X) + Math.Abs(start.Y - end.Y) < 5)
{
break;
}
else
{
end = new Point(r.Next(nOcrChar.Width), r.Next(nOcrChar.Height));
}
}
}
var op = new NOcrPoint(start, end);
bool ok = true;
foreach (NOcrPoint existingOp in nOcrChar.LinesForeground)
{
if (existingOp.Start.X == op.Start.X && existingOp.Start.Y == op.Start.Y &&
existingOp.End.X == op.End.X && existingOp.End.Y == op.End.Y)
ok = false;
}
if (ok && IsMatchPointForeGround(op, !tempVeryPrecise, nbmp, nOcrChar))
{
nOcrChar.LinesForeground.Add(op);
//AddHistoryItem(nOcrChar);
hits++;
}
count++;
if (count > giveUpCount - 100 && !tempVeryPrecise)
tempVeryPrecise = true;
}
count = 0;
hits = 0;
tempVeryPrecise = veryPrecise;
while (hits < numberOfLines && count < giveUpCount)
{
var start = new Point(r.Next(nOcrChar.Width), r.Next(nOcrChar.Height));
var end = new Point(r.Next(nOcrChar.Width), r.Next(nOcrChar.Height));
if (hits < 5 && count < 100) // a few large lines
{
for (int k = 0; k < 500; k++)
{
if (Math.Abs(start.X - end.X) + Math.Abs(start.Y - end.Y) > nOcrChar.Height / 2)
{
break;
}
else
{
end = new Point(r.Next(nOcrChar.Width), r.Next(nOcrChar.Height));
}
}
}
else // and a lot of small lines
{
for (int k = 0; k < 500; k++)
{
if (Math.Abs(start.X - end.X) + Math.Abs(start.Y - end.Y) < 5)
{
break;
}
else
{
end = new Point(r.Next(nOcrChar.Width), r.Next(nOcrChar.Height));
}
}
}
var op = new NOcrPoint(start, end);
bool ok = true;
foreach (NOcrPoint existingOp in nOcrChar.LinesBackground)
{
if (existingOp.Start.X == op.Start.X && existingOp.Start.Y == op.Start.Y &&
existingOp.End.X == op.End.X && existingOp.End.Y == op.End.Y)
ok = false;
}
if (ok && IsMatchPointBackGround(op, !tempVeryPrecise, nbmp, nOcrChar))
{
nOcrChar.LinesBackground.Add(op);
//AddHistoryItem(nOcrChar);
hits++;
}
count++;
if (count > giveUpCount - 100 && !tempVeryPrecise)
tempVeryPrecise = true;
}
}
private static bool IsMatchPointBackGround(NOcrPoint op, bool loose, NikseBitmap nbmp, NOcrChar nOcrChar)
{
foreach (Point point in op.ScaledGetPoints(nOcrChar, nbmp.Width, nbmp.Height))
{
if (point.X >= 0 && point.Y >= 0 && point.X < nbmp.Width && point.Y < nbmp.Height)
{
Color c = nbmp.GetPixel(point.X, point.Y);
if (c.A > 150 && c.R + 100 + c.G + c.B > VobSubOcr.NocrMinColor)
{
return false;
}
if (nbmp.Width > 10 && point.X + 1 < nbmp.Width)
{
c = nbmp.GetPixel(point.X + 1, point.Y);
if (c.A > 150 && c.R + 100 + c.G + c.B > VobSubOcr.NocrMinColor)
{
return false;
}
}
if (loose)
{
if (nbmp.Width > 10 && point.X >= 1)
{
c = nbmp.GetPixel(point.X - 1, point.Y);
if (c.A > 150 && c.R + 100 + c.G + c.B > VobSubOcr.NocrMinColor)
{
return false;
}
}
if (nbmp.Height > 10 && point.Y + 1 < nbmp.Height)
{
c = nbmp.GetPixel(point.X, point.Y + 1);
if (c.A > 150 && c.R + 100 + c.G + c.B > VobSubOcr.NocrMinColor)
{
return false;
}
}
if (nbmp.Height > 10 && point.Y >= 1)
{
c = nbmp.GetPixel(point.X, point.Y - 1);
if (c.A > 150 && c.R + 100 + c.G + c.B > VobSubOcr.NocrMinColor)
{
return false;
}
}
}
}
}
return true;
}
public NOcrChar GetMatch(NikseBitmap nbmp)
{
const int NocrMinColor = 300;
const int topMargin = 1;
double widthPercent = nbmp.Height * 100.0 / nbmp.Width;
foreach (NOcrChar oc in OcrCharacters)
{
if (Math.Abs(widthPercent - oc.WidthPercent) < 20 && Math.Abs(oc.MarginTop - topMargin) < 5)
{ // only very accurate matches
bool ok = true;
var index = 0;
while (index < oc.LinesForeground.Count && ok)
{
NOcrPoint op = oc.LinesForeground[index];
foreach (Point point in op.ScaledGetPoints(oc, nbmp.Width, nbmp.Height))
{
if (point.X >= 0 && point.Y >= 0 && point.X < nbmp.Width && point.Y < nbmp.Height)
{
Color c = nbmp.GetPixel(point.X, point.Y);
if (c.A > 150 && c.R + c.G + c.B > NocrMinColor)
{
}
else
{
Point p = new Point(point.X - 1, point.Y);
if (p.X < 0)
{
p.X = 1;
}
c = nbmp.GetPixel(p.X, p.Y);
if (nbmp.Width > 20 && c.A > 150 && c.R + c.G + c.B > NocrMinColor)
{
}
else
{
ok = false;
break;
}
}
}
}
index++;
}
index = 0;
while (index < oc.LinesBackground.Count && ok)
{
NOcrPoint op = oc.LinesBackground[index];
foreach (Point point in op.ScaledGetPoints(oc, nbmp.Width, nbmp.Height))
{
if (point.X >= 0 && point.Y >= 0 && point.X < nbmp.Width && point.Y < nbmp.Height)
{
Color c = nbmp.GetPixel(point.X, point.Y);
if (c.A > 150 && c.R + c.G + c.B > NocrMinColor)
{
Point p = new Point(point.X, point.Y);
if (oc.Width > 19 && point.X > 0)
{
p.X = p.X - 1;
}
c = nbmp.GetPixel(p.X, p.Y);
if (c.A > 150 && c.R + c.G + c.B > NocrMinColor)
{
ok = false;
break;
}
}
}
}
index++;
}
if (ok)
{
return(oc);
}
}
}
return(null);
}
private static List<NOcrPoint> ReadPoints(Stream stream)
{
var list = new List<NOcrPoint>();
int length = stream.ReadByte() << 8 | stream.ReadByte();
var buffer = new byte[8];
for (int i = 0; i < length; i++)
{
stream.Read(buffer, 0, buffer.Length);
var point = new NOcrPoint
{
Start = new Point(buffer[0] << 8 | buffer[1], buffer[2] << 8 | buffer[3]),
End = new Point(buffer[4] << 8 | buffer[5], buffer[6] << 8 | buffer[7])
};
list.Add(point);
}
return list;
}