/// <summary>
/// Performs OCR on a bitmap and returns the recognized characters.
/// </summary>
/// <param name="source"></param>
/// <param name="onlyNumbers"></param>
/// <param name="whiteThreshold">Value from 0 to 255, affects which pixels are considered white or black.</param>
/// <param name="x"></param>
/// <param name="y"></param>
/// <param name="ocrControl">If given the read letters are added for further editing and comparing.</param>
/// <returns></returns>
public static string ReadImageOcr(Bitmap source, bool onlyNumbers, byte whiteThreshold, int x = 0, int y = 0, OCRControl ocrControl = null)
{
var ret = string.Empty;
var maxDistanceForNonSpace = Math.Max(2, source.Height / 4);
var imageArray = ImageUtils.GetBooleanArrayOfImage(source, whiteThreshold);
//var debugArray = new bool[source.Width, source.Height];
//for (int xx = 0; xx < source.Width; xx++)
//{
// for (int yy = 0; yy < source.Height; yy++)
// {
// debugArray[xx, yy] = db.GetPixel(xx, yy).R == 0;
// }
//}
//Boolean2DimArrayConverter.ToDebugLog(debugArray);
// some images are just lines, ignore these. the lines can be at any height, probably not worth checking these
//if (source.Width > 3 * source.Height)
//{
// bool isJustLine = true;
// var minNotSetPixel = (int)(source.Width * 0.1);
// for (int sx = 0; sx < source.Width; sx++)
// {
// if (db.GetPixel(sx, 0).R == byte.MaxValue)
// {
// minNotSetPixel--;
// if (minNotSetPixel < 0)
// {
// isJustLine = false;
// break;
// }
// }
// }
// if (isJustLine) return string.Empty;
//}
var charSymbols = SplitBySymbol(imageArray, onlyNumbers).ToArray();
int xPos = charSymbols.FirstOrDefault()?.Coords.X ?? 0;
foreach (var sym in charSymbols)
{
// read spaces
if (!onlyNumbers && sym.Coords.X - xPos > maxDistanceForNonSpace)
{
ret += " ";
}
xPos = sym.Coords.X + sym.Pattern.GetLength(0);
var c = ArkOcr.Ocr.ocrConfig.RecognitionPatterns.FindMatchingChar(sym, imageArray, onlyNumbers: onlyNumbers);
// if c==string.Empty: character was not recognized and skipped in the manual recognition
if (string.IsNullOrEmpty(c))
{
if (c == null)
{
return(CleanUpOcr(ret)); // recognition was cancelled
}
continue;
}
ret += c;
//ocrControl?.AddLetterToRecognized(c, sym.Pattern);
ocrControl?.AddLetterToRecognized(new RecognizedCharData(x + sym.Coords.X, y + sym.Coords.Y, sym.YOffset)
{
Text = c, Pattern = sym.Pattern
});
}
return(CleanUpOcr(ret));
}
public static string ReadImageOcr(Bitmap source, bool onlyNumbers, float brightAdj = 1f, OCRControl ocrControl = null)
{
var ret = string.Empty;
using (var db = ImageUtils.GetAdjustedDirectBitmapOfImage(source, brightAdj)) // TODO use whiteThreshold from user
{
var adjPic = db.ToBitmap();
var charSymbols = SplitBySymbol(db, onlyNumbers);
int xPos = charSymbols.FirstOrDefault()?.Coords.X ?? 0;
foreach (var sym in charSymbols)
{
// read spaces
if (!onlyNumbers && sym.Coords.X - xPos > 3)
{
ret += " ";
}
xPos = sym.Coords.X + sym.Pattern.GetLength(0);
var c = ArkOCR.OCR.ocrConfig.RecognitionPatterns.FindMatchingChar(sym, adjPic, onlyNumbers: onlyNumbers);
// if c==string.Empty: character was not recognized and skipped in the manual recognition
if (string.IsNullOrEmpty(c))
{
if (c == null)
{
return(ret);
}
continue;
}
ret += c;
ocrControl?.AddLetterToRecognized(c, sym.Pattern);
}
}
ret = CleanUpOcr(ret);
return(ret);
}
