//After standarddeviation and mean value are calculted for each bin then
public void computeGaussianNormalizedWeight(double minValue)
{
while (currentImageId < 100)
{
Dictionary <int, double> imageHistogram = allImageHistogramRfMethod.ElementAt(currentImageId).getHistogramBinRfMethod();
string imageID = allImageHistogramRfMethod.ElementAt(currentImageId).imageID;
//new IMageDetails for each image
RFMethodImageDetails imageDetails = new RFMethodImageDetails();
//Every image has a unique id - fileapth of image
imageDetails.imageID = imageID;
//Every image has a histogram
imageDetails.ImageHistogram = imageHistogram;
foreach (KeyValuePair <int, double> pair in imageHistogram)
{
double gaussianNormalizedWeight = 0.0;
double meanValue = 0.0;
double standardDeviation = 0.0;
int binNumber = pair.Key;
RFMethodFeatureDetails featureValues = allFeature.ElementAt(binNumber);
meanValue = featureValues.meanValue;
standardDeviation = featureValues.standardDeviation;
if (standardDeviation == 0)
{
//check mean value
if (meanValue == 0)
{
gaussianNormalizedWeight = 0.0;//set gaussiannormalized weight to 0
imageDetails.GaussianNormalizedWeight.Add(gaussianNormalizedWeight);
}
else
{
standardDeviation = 0.5 * minValue;
gaussianNormalizedWeight = (double)((pair.Value - meanValue) / standardDeviation);
imageDetails.GaussianNormalizedWeight.Add(gaussianNormalizedWeight);
}
}
else
{
gaussianNormalizedWeight = (double)((pair.Value - meanValue) / standardDeviation);
//Each image has 89 Gaussiannormalized weight(computed only once) for each bin, adding one item in a list at a time
imageDetails.GaussianNormalizedWeight.Add(gaussianNormalizedWeight);
}
}
allImageDetails.Add(imageDetails);
currentImageId++;
}
}
//In each iteration for each feature calculate mean value,standard deviation, weight and updated weight
public void addFeatureValues()
{
//ADD average value and standard deviation of each bin
while (currentFeature < 89)
{
RFMethodFeatureDetails feature = new RFMethodFeatureDetails();
//1) compute mean value
double average = calculateMeanValue();
feature.meanValue = average;
//2)compute standard deviation
double standardDeviation = computeStandardDeviation(average);
feature.standardDeviation = standardDeviation;
//feature.updatedWeight = computeUpdatedWeight(standardDeviation,average);
//ID assigned to each bin from 1-89
feature.featureID = currentFeature;
//Add each feature details in one place i.e. List of RFMethodFeatureDetails type
allFeature.Add(feature);
currentFeature++;
}
}
public List <ResultCBIR> computeMinkowskiDistance()
{
//Clear result list before adding value
this.DistancebetweenImage.Clear();
List <double> selectedImage = new List <double>();
foreach (RFMethodImageDetails eachImage in allImageDetails)
{
string image = eachImage.imageID;
//if(image==queryImage)
if (new Uri(image) == new Uri(queryImage))
{
//compute distance between two images ,
selectedImage = eachImage.GaussianNormalizedWeight;
break;
}
}
foreach (RFMethodImageDetails histogramImage in allImageDetails)
{
ResultCBIR dis = new ResultCBIR();
int binNumber = 0;
double totalDistance = 0.0;
while (binNumber < 89)
{
double queryImage = selectedImage.ElementAt(binNumber);
double databaseImage = histogramImage.GaussianNormalizedWeight.ElementAt(binNumber);
RFMethodFeatureDetails featureSet = allFeature.ElementAt(binNumber);
double weight = featureSet.normalizedWeight;
double distance = weight * (Math.Abs(queryImage - databaseImage));
totalDistance = distance + totalDistance;
binNumber++;
}
dis.imageid = histogramImage.imageID;
dis.totaldistance = totalDistance;
DistancebetweenImage.Add(dis);
}
//Orderby
DistancebetweenImage = DistancebetweenImage.OrderBy(i => i.totaldistance).ToList();
return(DistancebetweenImage);
}
