private void frmSplash_Activated(object sender, EventArgs e)
{
this.Refresh();
proBarLoad.Value = 60;
lblStatus.Text = "Features Extractor Module . . .";
this.Refresh();
FeatureExtractorclass FEC = new FeatureExtractorclass();
proBarLoad.Value = 90;
lblStatus.Text = "Image Processing Tools . . .";
this.Refresh();
ImageManipulationclass IMC = new ImageManipulationclass();
proBarLoad.Value = 100;
frmMain mainScreen = new frmMain();
mainScreen.Show();
this.Hide();
}
/// <summary>
/// Creates new instance of the featureExtractor Class
/// </summary>
/// <param name="ImagePath">The path of the image to extract features from</param>
public featureExtractor(string ImagePath)
{
this.imagePath = new MWCharArray(ImagePath);
FEC = new FeatureExtractorclass();
}
/// <summary>
/// Finds an image that matches the example image
/// </summary>
/// <param name="ExampleImagePath">The path of the example image</param>
/// <param name="DBC">The controller through which the search will be done</param>
public List <string[]> SearchByExample(string ExampleImagePath, databaseController DBC, double Max, double Min)
{
try
{
List <string[]> res = new List <string[]>();
//Get the features of the example image
featureExtractor fe = new featureExtractor(ExampleImagePath);
List <Feature> fv = fe.AllFeatures();
double area = fv[fv.Count - 1].FeatureValue * fv[fv.Count - 2].FeatureValue;
double MaxArea = area * Max;
double MinArea = area * Min;
fv.RemoveAt(fv.Count - 1);
fv.RemoveAt(fv.Count - 1);
fv.RemoveAt(fv.Count - 1);
fv.RemoveAt(fv.Count - 1);
//Example color features
double[] exCF = new double[4];
for (int i = 0; i < 4; i++)
{
exCF[i] = fv[i].FeatureValue;
}
//Example shape features
double[] exSF = new double[7];
for (int i = 4; i < 11; i++)
{
exSF[i - 4] = fv[i].FeatureValue;
}
//Example texture features
double[] exTF = new double[24];
for (int i = 11; i < 35; i++)
{
exTF[i - 11] = fv[i].FeatureValue;
}
//Example edge features
double[] exEF = new double[37];
for (int i = 35; i < 72; i++)
{
exEF[i - 35] = fv[i].FeatureValue;
}
//Calculate similarities between example image and images in the database
FeatureExtractorclass fec = new FeatureExtractorclass();
MWArray mwXCF = new MWNumericArray(exCF.Length, 1, exCF);
MWArray mwXSF = new MWNumericArray(exSF.Length, 1, exSF);
MWArray mwXTF = new MWNumericArray(exTF.Length, 1, exTF);
MWArray mwXEF = new MWNumericArray(exEF.Length, 1, exEF);
DBC.CommandText = "SELECT * FROM FeaturesVector where (CropH*CropW >=" + MinArea.ToString() + ") AND (CropH*CropW <=" + MaxArea.ToString() + ");";
DataSet ds = DBC.ExecuteQuery();
if (ds.Tables[0].Rows.Count < 1)
{
throw new Exception("The database is empty or there are no matches!");
}
//Calculate similiraties for each group of feature . . .
List <Result> ColorSimilarties = new List <Result>();
List <Result> ShapeSimilarties = new List <Result>();
List <Result> TextureSimilarties = new List <Result>();
List <Result> EdgeSimilarties = new List <Result>();
double[] Cfrow;
double[] Sfrow;
double[] Tfrow;
double[] Efrow;
MWArray mwRes = new MWNumericArray(1);
MWArray mwCC; //Color features for current image
MWArray mwCS; //Shape features for current image
MWArray mwCT; //Texture features for current image
MWArray mwCE; //Edge features for current image
for (int j = 0; j < ds.Tables[0].Rows.Count; j++)
{
int ID = int.Parse(ds.Tables[0].Rows[j].ItemArray[0].ToString());
//Current image color features
Cfrow = new double[4];
for (int i = 1; i < 5; i++)
{
Cfrow[i - 1] = double.Parse(ds.Tables[0].Rows[j].ItemArray[i].ToString());
}
mwCC = new MWNumericArray(Cfrow.Length, 1, Cfrow);
mwRes = fec.minkowski(mwXCF, mwCC, 4);
double Distance = double.Parse(mwRes.ToString());
Result sim = new Result(ID, Distance);
ColorSimilarties.Add(sim);
//Current image shape features
Sfrow = new double[7];
for (int i = 5; i < 12; i++)
{
Sfrow[i - 5] = double.Parse(ds.Tables[0].Rows[j].ItemArray[i].ToString());
}
mwCS = new MWNumericArray(Sfrow.Length, 1, Sfrow);
mwRes = fec.minkowski(mwXSF, mwCS, 4);
Distance = double.Parse(mwRes.ToString());
sim = new Result(ID, Distance);
ShapeSimilarties.Add(sim);
//Current image texture features
Tfrow = new double[24];
for (int i = 12; i < 36; i++)
{
Tfrow[i - 12] = double.Parse(ds.Tables[0].Rows[j].ItemArray[i].ToString());
}
mwCT = new MWNumericArray(Tfrow.Length, 1, Tfrow);
mwRes = fec.minkowski(mwXTF, mwCT, 4);
Distance = double.Parse(mwRes.ToString());
sim = new Result(ID, Distance);
TextureSimilarties.Add(sim);
//Current image edge features
Efrow = new double[37];
for (int i = 36; i < 73; i++)
{
Efrow[i - 36] = double.Parse(ds.Tables[0].Rows[j].ItemArray[i].ToString());
}
mwCE = new MWNumericArray(Efrow.Length, 1, Efrow);
mwRes = fec.minkowski(mwXEF, mwCE, 4);
Distance = double.Parse(mwRes.ToString());
sim = new Result(ID, Distance);
EdgeSimilarties.Add(sim);
}
ShapeSimilarties = sortList(ShapeSimilarties);
double resAvg = ListAverag(ShapeSimilarties);
while (ShapeSimilarties[ShapeSimilarties.Count - 1].Distance > resAvg)
{
Result R = ShapeSimilarties[ShapeSimilarties.Count - 1];
ShapeSimilarties.RemoveAt(ShapeSimilarties.Count - 1);
for (int r = 0; r < EdgeSimilarties.Count; r++)
{
if (EdgeSimilarties[r].ID == R.ID)
{
EdgeSimilarties.RemoveAt(r);
TextureSimilarties.RemoveAt(r);
ColorSimilarties.RemoveAt(r);
break;
}
}
}
EdgeSimilarties = sortList(EdgeSimilarties);
resAvg = ListAverag(EdgeSimilarties);
while (EdgeSimilarties[EdgeSimilarties.Count - 1].Distance > resAvg)
{
Result R = EdgeSimilarties[EdgeSimilarties.Count - 1];
EdgeSimilarties.RemoveAt(EdgeSimilarties.Count - 1);
for (int r = 0; r < TextureSimilarties.Count; r++)
{
if (TextureSimilarties[r].ID == R.ID)
{
TextureSimilarties.RemoveAt(r);
ColorSimilarties.RemoveAt(r);
break;
}
}
}
TextureSimilarties = sortList(TextureSimilarties);
resAvg = ListAverag(TextureSimilarties);
while (TextureSimilarties[TextureSimilarties.Count - 1].Distance > resAvg)
{
Result R = TextureSimilarties[TextureSimilarties.Count - 1];
TextureSimilarties.RemoveAt(TextureSimilarties.Count - 1);
for (int r = 0; r < ColorSimilarties.Count; r++)
{
if (ColorSimilarties[r].ID == R.ID)
{
ColorSimilarties.RemoveAt(r);
break;
}
}
}
ColorSimilarties = sortList(ColorSimilarties);
resAvg = ListAverag(ColorSimilarties);
while (ColorSimilarties[ColorSimilarties.Count - 1].Distance > resAvg)
{
ColorSimilarties.RemoveAt(ColorSimilarties.Count - 1);
}
List <Result> similarties = ColorSimilarties;
//Ranking :
double farestImg = (double)similarties[similarties.Count - 1].Distance; //The worst match.
double rank = 0;
string[] resStr;
for (int i = 0; i < similarties.Count; i++)
{
resStr = new string[4];
int x = similarties[i].ID;
DBC.CommandText = "SELECT I.FileName, S.FolderPath ";
DBC.CommandText += "FROM Image AS I INNER JOIN Series AS S ON FLOOR(I.ImageID / 100) = S.SID ";
DBC.CommandText += "WHERE (I.ImageID = " + x + ")";
ds = DBC.ExecuteQuery();
rank = (1 - ((double)similarties[i].Distance / farestImg)) * 100;
rank = Math.Round(rank, 5);
resStr[0] = ds.Tables[0].Rows[0].ItemArray[0].ToString();
resStr[1] = ds.Tables[0].Rows[0].ItemArray[1].ToString();
resStr[2] = " - " + rank.ToString() + "%";
resStr[3] = x.ToString();
res.Add(resStr);
resStr = null;
}
return(res);
}
catch (Exception ex)
{
return(null);
}
}
