/// <summary>
/// A method which creates the best possible match for a license plate
/// </summary>
/// <param name="pos1">The position of the first stripe</param>
/// <param name="pos2">The posisiton of the second stripe</param>
/// <param name="partsNum">The kenType array</param>
/// <param name="matcher">The BlobMatcher</param>
private void createBestMatch(int pos1, int pos2, kenType[] partsNum, BlobMatcher_Int16 matcher)
{
// **************************************************************
// ** Creates the best match based on: **
// ** - If there should be a number, but char found: **
// ** - Find a number in the results and change it **
// ** - If there should be a number and number found **
// ** - If the next match is a character then we can **
// ** up the confidence a lot **
// ** - If the then following match is a number we **
// ** can up it a little **
// ** **
// ** - This also aplies for chars **
// **************************************************************
// A pointer which points to the partsNum place where we are at
int partNumPlace = 0;
const double maxError = 0.20;
// Go through all the chars
for (int i = 0; i < chars.Count; i++)
{
// Check if we need to up the pointer
if (i == pos1 || i == pos2)
{
partNumPlace += 1;
}
// If there should be a number found
if (partsNum[partNumPlace] == kenType.NUMBER && chars[i].error < maxError)
{
// If it isn't a number
if (!Char.IsNumber(chars[i].character[0]))
{
// Loop through the results and find a number
foreach (PatternMatchResult patrn in chars[i].PatternResults)
{
// if this is a number
if (Char.IsNumber(matcher.PatternName(patrn.id)[0]))
{
// Change the chars value to this number
chars[i] = new LicenseCharacter(matcher.PatternName(patrn.id), -0.04, 41);
break;
}
} // End for loop
} // End number expected but not found
else if (chars[i].error > 0.11) // If there should be a number and a number was found
{
// Check if the next match is a char
if (!Char.IsNumber(matcher.PatternName(chars[i].PatternResults[1].id)[0]))
{
// If so we can up the confidence
chars[i] = new LicenseCharacter(chars[i].character, -0.02, 41);
} // If the then following match is a char
else if (!Char.IsNumber(matcher.PatternName(chars[i].PatternResults[2].id)[0]))
{
// Then we up the confidence a little
chars[i] = new LicenseCharacter(chars[i].character, -0.01, chars[i].confidence + 0.1);
}
}
} // Check if we should expect a char and the error value isn't to high
else if (partsNum[partNumPlace] == kenType.CHAR && chars[i].error < maxError)
{
// Char expected but number found
if (Char.IsNumber(chars[i].character[0]))
{
// Loop through the matches
foreach (PatternMatchResult patrn in chars[i].PatternResults)
{
if (!Char.IsNumber(matcher.PatternName(patrn.id)[0]))
{
// If a char was found change the chars[i] value to it
chars[i] = new LicenseCharacter(matcher.PatternName(patrn.id), -0.04, 41);
break;
}
} // End for loop
} // If we already have a char
else if (chars[i].error > 0.11)
{
// Check if the second match is a number
if (Char.IsNumber(matcher.PatternName(chars[i].PatternResults[1].id)[0]))
{
// If so we can up the confidence a lot
chars[i] = new LicenseCharacter(chars[i].character, -0.02, 41);
} // Check if the then following match is a number
else if (Char.IsNumber(matcher.PatternName(chars[i].PatternResults[2].id)[0]))
{
// If so then we can up the confidence a little
chars[i] = new LicenseCharacter(chars[i].character, -0.01, chars[i].confidence + 0.1);
}
} // End char expected and char found
} // End char expected
} // End loop
// Done
}
/// <summary>
/// A method which creates the best possible match for a license plate
/// </summary>
/// <param name="pos1">The position of the first stripe</param>
/// <param name="pos2">The posisiton of the second stripe</param>
/// <param name="partsNum">The kenType array</param>
/// <param name="matcher">The BlobMatcher</param>
private void createBestMatch(int pos1, int pos2, kenType[] partsNum, BlobMatcher_Int16 matcher)
{
// **************************************************************
// ** Creates the best match based on: **
// ** - If there should be a number, but char found: **
// ** - Find a number in the results and change it **
// ** - If there should be a number and number found **
// ** - If the next match is a character then we can **
// ** up the confidence a lot **
// ** - If the then following match is a number we **
// ** can up it a little **
// ** **
// ** - This also aplies for chars **
// **************************************************************
// A pointer which points to the partsNum place where we are at
int partNumPlace = 0;
const double maxError = 0.20;
// Go through all the chars
for (int i = 0; i < chars.Count; i++)
{
// Check if we need to up the pointer
if (i == pos1 || i == pos2)
partNumPlace += 1;
// If there should be a number found
if (partsNum[partNumPlace] == kenType.NUMBER && chars[i].error < maxError)
{
// If it isn't a number
if (!Char.IsNumber(chars[i].character[0]))
{
// Loop through the results and find a number
foreach (PatternMatchResult patrn in chars[i].PatternResults)
{
// if this is a number
if (Char.IsNumber(matcher.PatternName(patrn.id)[0]))
{
// Change the chars value to this number
chars[i] = new LicenseCharacter(matcher.PatternName(patrn.id), -0.04, 41);
break;
}
} // End for loop
} // End number expected but not found
else if (chars[i].error > 0.11) // If there should be a number and a number was found
{
// Check if the next match is a char
if (!Char.IsNumber(matcher.PatternName(chars[i].PatternResults[1].id)[0]))
{
// If so we can up the confidence
chars[i] = new LicenseCharacter(chars[i].character, -0.02, 41);
} // If the then following match is a char
else if (!Char.IsNumber(matcher.PatternName(chars[i].PatternResults[2].id)[0]))
{
// Then we up the confidence a little
chars[i] = new LicenseCharacter(chars[i].character, -0.01, chars[i].confidence + 0.1);
}
}
} // Check if we should expect a char and the error value isn't to high
else if (partsNum[partNumPlace] == kenType.CHAR && chars[i].error < maxError)
{
// Char expected but number found
if (Char.IsNumber(chars[i].character[0]))
{
// Loop through the matches
foreach (PatternMatchResult patrn in chars[i].PatternResults)
{
if (!Char.IsNumber(matcher.PatternName(patrn.id)[0]))
{
// If a char was found change the chars[i] value to it
chars[i] = new LicenseCharacter(matcher.PatternName(patrn.id), -0.04, 41);
break;
}
} // End for loop
} // If we already have a char
else if (chars[i].error > 0.11)
{
// Check if the second match is a number
if (Char.IsNumber(matcher.PatternName(chars[i].PatternResults[1].id)[0]))
{
// If so we can up the confidence a lot
chars[i] = new LicenseCharacter(chars[i].character, -0.02, 41);
} // Check if the then following match is a number
else if (Char.IsNumber(matcher.PatternName(chars[i].PatternResults[2].id)[0]))
{
// If so then we can up the confidence a little
chars[i] = new LicenseCharacter(chars[i].character, -0.01, chars[i].confidence + 0.1);
}
} // End char expected and char found
} // End char expected
} // End loop
// Done
}
/*
Description:
Read the license plate
Input:
//Rectified license plate image containing six characters
Int16Image labeledRectifiedPlateImage
BlobMatcher_Int16 matcher //initialized blobmatcher
ClassLexicon lexicon //initialized lexicon
Output:
//Result by the blob matcher
ref LicensePlate result
//Result by the lexicon
ref LicensePlate lexiconResult
Return:
//six characters found
bool
*/
public static bool MatchPlate(Int16Image binaryCharacterImage, BlobMatcher_Int16 matcher,
ClassLexicon lexicon, ref LicensePlate result, ref LicensePlate lexiconResult, bool dilate)
{
// NIEUW
// 2de optie voor aanroep als eerste low confidence levert
if (dilate)
{
Int16Image temp = new Int16Image();
VisionLab.Dilation(binaryCharacterImage, temp, new Mask_Int32(3, 3, 1));
binaryCharacterImage = new Int16Image(temp);
temp.Dispose();
}
if (VisionLab.LabelBlobs(binaryCharacterImage, Connected.EightConnected) != 6)
return false;
//Calculate dimensions and locations of all characters/blobs.
vector_BlobAnalyse ba_vec = new vector_BlobAnalyse();
ba_vec.Add(BlobAnalyse.BA_TopLeft);
ba_vec.Add(BlobAnalyse.BA_Height);
ba_vec.Add(BlobAnalyse.BA_Width);
vector_Blob returnBlobs = new vector_Blob();
VisionLab.BlobAnalysis(binaryCharacterImage, VisionLab.VectorToSet_BlobAnalyse(ba_vec), VisionLab.MaxPixel(binaryCharacterImage), returnBlobs, SortOrder.SortDown, BlobAnalyse.BA_TopLeft, UseXOrY.UseX);
ba_vec.Dispose();
Int16Image binaryCharacter = new Int16Image();
//Create data structure for lexicon.
vector_vector_LetterMatch match = new vector_vector_LetterMatch();
// NIEUW
// Change the matcher params
matcher.ChangeParams(60, 10, 64, 0);
//Process each character/blob.
foreach (Blob b in returnBlobs)
{
//Cut out character
VisionLab.ROI(binaryCharacterImage, binaryCharacter, b.TopLeft(), new HeightWidth(b.Height(), b.Width()));
//Convert ROI result to binary
VisionLab.ClipPixelValue(binaryCharacter, 0, 1);
//Calculate character's classification for all classes.
vector_PatternMatchResult returnMatches = new vector_PatternMatchResult();
float conf = matcher.AllMatches(binaryCharacter, (float)-0.5, (float)0.5, returnMatches);
float err = returnMatches[0].error;
int id = returnMatches[0].id;
string chr = matcher.PatternName(id);
// NIEUW
// If error to big decrease the confidence
if(err > 0.20f)
conf -= 0.2f;
//Fill datastructure for lexicon.
match.Add(VisionLabEx.PatternMatchResultToLetterMatch(returnMatches));
//Store best match in result
result.characters.Add(
new LicenseCharacter(
chr,
err,
conf,
// NIEUW
// Extra param: The middle of a character
// (used for matching patterns)
b.TopLeft().x + ((b.TopRight().x - b.TopLeft().x)/2) ,
// NIEUW
// All other results that we're found
// So we can switch between em
returnMatches
));
}
//Validate match with lexicon.
vector_int bestWord = new vector_int();
lexiconResult.confidence = lexicon.FindBestWord(match, bestWord, Optimize.OptimizeForMinimum);
for (int i = 0; i < bestWord.Count; i++)
{
string character = matcher.PatternName(bestWord[i]);
//Store lexicon result
lexiconResult.characters.Add(new LicenseCharacter(character));
}
// NIEUW
// Create the best match with the aid of the pattern matcher
result.CalculateBestMatch(matcher);
binaryCharacter.Dispose();
returnBlobs.Dispose();
match.Dispose();
bestWord.Dispose();
GC.Collect();
return true;
}
/*
* Description:
* Read the license plate
* Input:
* //Rectified license plate image containing six characters
* Int32Image labeledRectifiedPlateImage
* BlobMatcher_Int16 matcher //initialized blobmatcher
* ClassLexicon lexicon //initialized lexicon
* Output:
* //Result by the blob matcher
* ref LicensePlate result
* //Result by the lexicon
* ref LicensePlate lexiconResult
* Return:
* //six characters found
* bool
*/
public static bool MatchPlate(Int32Image binaryCharacterImage, BlobMatcher_Int16 matcher, ClassLexicon lexicon, ref LicensePlate result, ref LicensePlate lexiconResult)
{
//Check if 6 characters/blobs have been found and label image.
if (VisionLab.LabelBlobs(binaryCharacterImage, Connected.EightConnected) != 6)
{
return(false);
}
//Calculate dimensions and locations of all characters/blobs.
vector_BlobAnalyse ba_vec = new vector_BlobAnalyse();
ba_vec.Add(BlobAnalyse.BA_TopLeft);
ba_vec.Add(BlobAnalyse.BA_Height);
ba_vec.Add(BlobAnalyse.BA_Width);
vector_Blob returnBlobs = new vector_Blob();
VisionLab.BlobAnalysis(binaryCharacterImage, VisionLab.VectorToSet_BlobAnalyse(ba_vec), VisionLab.MaxPixel(binaryCharacterImage), returnBlobs, SortOrder.SortDown, BlobAnalyse.BA_TopLeft, UseXOrY.UseX);
ba_vec.Dispose();
Int32Image binaryCharacter = new Int32Image();
Int16Image binaryCharacter16 = new Int16Image();
//Create data structure for lexicon.
vector_vector_LetterMatch match = new vector_vector_LetterMatch();
//Process each character/blob.
foreach (Blob b in returnBlobs)
{
//Cut out character
VisionLab.ROI(binaryCharacterImage, binaryCharacter, b.TopLeft(), new HeightWidth(b.Height(), b.Width()));
//Convert ROI result to binary
VisionLab.ClipPixelValue(binaryCharacter, 0, 1);
//Calculate character's classification for all classes.
vector_PatternMatchResult returnMatches = new vector_PatternMatchResult();
VisionLab.Convert(binaryCharacter, binaryCharacter16);
float conf = matcher.AllMatches(binaryCharacter16, (float)-0.5, (float)0.5, returnMatches);
float err = returnMatches[0].error;
int id = returnMatches[0].id;
string chr = matcher.PatternName(id);
//Fill datastructure for lexicon.
match.Add(VisionLabEx.PatternMatchResultToLetterMatch(returnMatches));
//Store best match in result
result.characters.Add(new LicenseCharacter(chr, err, conf));
}
//Validate match with lexicon.
vector_int bestWord = new vector_int();
lexiconResult.confidence = lexicon.FindBestWord(match, bestWord, Optimize.OptimizeForMinimum);
for (int i = 0; i < bestWord.Count; i++)
{
string character = matcher.PatternName(bestWord[i]);
//Store lexicon result
lexiconResult.characters.Add(new LicenseCharacter(character));
}
binaryCharacter.Dispose();
binaryCharacter16.Dispose();
returnBlobs.Dispose();
match.Dispose();
bestWord.Dispose();
//GC.Collect();
return(true);
}
/*
* Description:
* Read the license plate
* Input:
* //Rectified license plate image containing six characters
* Int16Image labeledRectifiedPlateImage
* BlobMatcher_Int16 matcher //initialized blobmatcher
* ClassLexicon lexicon //initialized lexicon
* Output:
* //Result by the blob matcher
* ref LicensePlate result
* //Result by the lexicon
* ref LicensePlate lexiconResult
* Return:
* //six characters found
* bool
*/
public static bool MatchPlate(Int16Image binaryCharacterImage, BlobMatcher_Int16 matcher,
ClassLexicon lexicon, ref LicensePlate result, ref LicensePlate lexiconResult, bool dilate)
{
// NIEUW
// 2de optie voor aanroep als eerste low confidence levert
if (dilate)
{
Int16Image temp = new Int16Image();
VisionLab.Dilation(binaryCharacterImage, temp, new Mask_Int32(3, 3, 1));
binaryCharacterImage = new Int16Image(temp);
temp.Dispose();
}
if (VisionLab.LabelBlobs(binaryCharacterImage, Connected.EightConnected) != 6)
{
return(false);
}
//Calculate dimensions and locations of all characters/blobs.
vector_BlobAnalyse ba_vec = new vector_BlobAnalyse();
ba_vec.Add(BlobAnalyse.BA_TopLeft);
ba_vec.Add(BlobAnalyse.BA_Height);
ba_vec.Add(BlobAnalyse.BA_Width);
vector_Blob returnBlobs = new vector_Blob();
VisionLab.BlobAnalysis(binaryCharacterImage, VisionLab.VectorToSet_BlobAnalyse(ba_vec), VisionLab.MaxPixel(binaryCharacterImage), returnBlobs, SortOrder.SortDown, BlobAnalyse.BA_TopLeft, UseXOrY.UseX);
ba_vec.Dispose();
Int16Image binaryCharacter = new Int16Image();
//Create data structure for lexicon.
vector_vector_LetterMatch match = new vector_vector_LetterMatch();
// NIEUW
// Change the matcher params
matcher.ChangeParams(60, 10, 64, 0);
//Process each character/blob.
foreach (Blob b in returnBlobs)
{
//Cut out character
VisionLab.ROI(binaryCharacterImage, binaryCharacter, b.TopLeft(), new HeightWidth(b.Height(), b.Width()));
//Convert ROI result to binary
VisionLab.ClipPixelValue(binaryCharacter, 0, 1);
//Calculate character's classification for all classes.
vector_PatternMatchResult returnMatches = new vector_PatternMatchResult();
float conf = matcher.AllMatches(binaryCharacter, (float)-0.5, (float)0.5, returnMatches);
float err = returnMatches[0].error;
int id = returnMatches[0].id;
string chr = matcher.PatternName(id);
// NIEUW
// If error to big decrease the confidence
if (err > 0.20f)
{
conf -= 0.2f;
}
//Fill datastructure for lexicon.
match.Add(VisionLabEx.PatternMatchResultToLetterMatch(returnMatches));
//Store best match in result
result.characters.Add(
new LicenseCharacter(
chr,
err,
conf,
// NIEUW
// Extra param: The middle of a character
// (used for matching patterns)
b.TopLeft().x + ((b.TopRight().x - b.TopLeft().x) / 2),
// NIEUW
// All other results that we're found
// So we can switch between em
returnMatches
));
}
//Validate match with lexicon.
vector_int bestWord = new vector_int();
lexiconResult.confidence = lexicon.FindBestWord(match, bestWord, Optimize.OptimizeForMinimum);
for (int i = 0; i < bestWord.Count; i++)
{
string character = matcher.PatternName(bestWord[i]);
//Store lexicon result
lexiconResult.characters.Add(new LicenseCharacter(character));
}
// NIEUW
// Create the best match with the aid of the pattern matcher
result.CalculateBestMatch(matcher);
binaryCharacter.Dispose();
returnBlobs.Dispose();
match.Dispose();
bestWord.Dispose();
GC.Collect();
return(true);
}