public void computeAndExtract()
{
using (detector = new SURF(30))
using (matcher = new BFMatcher(DistanceType.L2))
{
bowDE = new BOWImgDescriptorExtractor(detector, matcher);
BOWKMeansTrainer bowTrainer = new BOWKMeansTrainer(100, new MCvTermCriteria(100, 0.01), 3, Emgu.CV.CvEnum.KMeansInitType.PPCenters);
foreach(FileInfo[] folder in _folders)
foreach (FileInfo file in folder)
{
using (Image<Bgr, Byte> model = new Image<Bgr, byte>(file.FullName))
using (VectorOfKeyPoint modelKeyPoints = new VectorOfKeyPoint())
//Detect SURF key points from images
{
detector.DetectRaw(model, modelKeyPoints);
//Compute detected SURF key points & extract modelDescriptors
Mat modelDescriptors = new Mat();
detector.Compute(model, modelKeyPoints, modelDescriptors);
//Add the extracted BoW modelDescriptors into BOW trainer
bowTrainer.Add(modelDescriptors);
}
input_num++;
}
//Cluster the feature vectors
bowTrainer.Cluster(vocabulary);
//Store the vocabulary
bowDE.SetVocabulary(vocabulary);
//training descriptors
tDescriptors = new Mat();
labels = new Matrix<int>(1, input_num);
int index = 0;
//compute and store BOWDescriptors and set labels
for (int i = 1; i <= _folders.Count; i++)
{
FileInfo[] files = _folders[i-1];
for (int j = 0; j < files.Length; j++)
{
FileInfo file = files[j];
using (Image<Bgr, Byte> model = new Image<Bgr, Byte>(file.FullName))
using (VectorOfKeyPoint modelKeyPoints = new VectorOfKeyPoint())
using (Mat modelBOWDescriptor = new Mat())
{
detector.DetectRaw(model, modelKeyPoints);
bowDE.Compute(model, modelKeyPoints, modelBOWDescriptor);
tDescriptors.PushBack(modelBOWDescriptor);
labels[0, index++] = i;
}
}
}
}
}
public Prep_L2BOF(int classnum, string src)
{
SecondLayerDic = null;
directory = src;
classNum = classnum;
bowTrainer = new BOWKMeansTrainer(classNum, new MCvTermCriteria(10, 0.01), 3,
KMeansInitType.PPCenters);
}
public Prep_L1BOF(int classnum,string loc)
{
directory = loc;
topLayerDic = null;
classNum = classnum;
bowTrainer = new BOWKMeansTrainer(classNum, new MCvTermCriteria(10, 0.01), 3,
KMeansInitType.PPCenters);
}
public Load_L2BOF(int classnum, string src)
{
SecondLayerDic = null;
directory = src;
classNum = classnum;
bowTrainer = new BOWKMeansTrainer(classNum, new MCvTermCriteria(10, 0.01), 3,
KMeansInitType.PPCenters);
IFormatter formatter = new BinaryFormatter();
FileStream fs = File.OpenRead(directory + "\\obj\\dic.xml");
SecondLayerDic = (Matrix<float>)formatter.Deserialize(fs);
fs.Dispose();
bowDe.SetVocabulary(SecondLayerDic);
SecondLayerSVM = new SVM();
SecondLayerSVM.Load(directory + "\\obj\\svm.xml");
Console.WriteLine("Finished Loading L2 SVM.");loaded = true;
}
public Load_L1BOF(int classnum,string loc)
{
prefix = loc;
topLayerDic = null;
classNum = classnum;
bowTrainer = new BOWKMeansTrainer(classNum, new MCvTermCriteria(10, 0.01), 3,
KMeansInitType.PPCenters);
IFormatter formatter = new BinaryFormatter();
FileStream fs = File.OpenRead(loc + "obj\\dic.xml");
topLayerDic = (Matrix<float>)formatter.Deserialize(fs);
fs.Dispose();
bowDe.SetVocabulary(topLayerDic);
topLayerSVM = new SVM();
topLayerSVM.Load(loc + "obj\\svm.xml");
Console.WriteLine("Finished Loading L1 SVM.");
loaded = true;
}
public Prep(List<Image<Bgr, byte>> imgs,string directory)
{
dir = directory;
images = imgs;
bowTrainer = new BOWKMeansTrainer(classNum, new MCvTermCriteria(10, 0.01), 3,
KMeansInitType.PPCenters);
labels = new Matrix<float>(images.Count, 1);
foreach (Image<Bgr, byte> img in images)
{
Extract(img);
}
MakeDic();
double i = 1;
foreach (Image<Bgr, byte> img in images)
{
int label_value = Convert.ToInt32(Math.Floor(i));
MakeDescriptors(img,label_value);
i += 0.5;
}
Save();
}
//make Dictionarys
private void MakeDic()
{
l1:
try
{
Dic = bowTrainer.Cluster();
bowDe.SetVocabulary(Dic);
trainingDescriptors = new Matrix<float>(images.Count, classNum);
}
catch (Exception e)
{
int int_classnum = classNum;
int_classnum++;
trainingDescriptors = new Matrix<float>(images.Count, int_classnum);
bowTrainer = new BOWKMeansTrainer(int_classnum, new MCvTermCriteria(10, 0.01), 3,
KMeansInitType.PPCenters);
goto l1;
}
}
public void TestBOWKmeansTrainer2()
{
Image<Gray, byte> box = EmguAssert.LoadImage<Gray, byte>("box.png");
Brisk detector = new Brisk(30, 3, 1.0f);
VectorOfKeyPoint kpts = new VectorOfKeyPoint();
Mat descriptors = new Mat();
detector.DetectAndCompute(box, null, kpts, descriptors, false);
Mat descriptorsF = new Mat();
descriptors.ConvertTo(descriptorsF, CvEnum.DepthType.Cv32F);
//Matrix<float> descriptorsF = descriptors.Convert<float>();
BOWKMeansTrainer trainer = new BOWKMeansTrainer(100, new MCvTermCriteria(), 3, CvEnum.KMeansInitType.PPCenters);
trainer.Add(descriptorsF);
Mat vocabulary = new Mat();
trainer.Cluster(vocabulary);
BFMatcher matcher = new BFMatcher(DistanceType.L2);
BOWImgDescriptorExtractor extractor = new BOWImgDescriptorExtractor(detector, matcher);
Mat vocabularyByte = new Mat();
vocabulary.ConvertTo(vocabularyByte, CvEnum.DepthType.Cv8U);
extractor.SetVocabulary(vocabularyByte);
Mat descriptors2 = new Mat();
extractor.Compute(box, kpts, descriptors2);
}
public void TestBOWKmeansTrainer()
{
Image<Gray, byte> box = EmguAssert.LoadImage<Gray, byte>("box.png");
SURF detector = new SURF(500);
VectorOfKeyPoint kpts = new VectorOfKeyPoint();
Mat descriptors = new Mat();
detector.DetectAndCompute(box, null, kpts, descriptors, false);
BOWKMeansTrainer trainer = new BOWKMeansTrainer(100, new MCvTermCriteria(), 3, CvEnum.KMeansInitType.PPCenters);
trainer.Add(descriptors);
Mat vocabulary = new Mat();
trainer.Cluster(vocabulary);
BFMatcher matcher = new BFMatcher(DistanceType.L2);
BOWImgDescriptorExtractor extractor = new BOWImgDescriptorExtractor(detector, matcher);
extractor.SetVocabulary(vocabulary);
Mat descriptors2 = new Mat();
extractor.Compute(box, kpts, descriptors2);
}
