public bool LoadMinstFiles()
{
//clear Image Pattern List
if (m_pImagePatterns != null)
{
m_pImagePatterns.Clear();
}
//close files if opened
if (_bImageFileOpen)
{
load_ImageFile_stream.Close();
_bImageFileOpen = false;
}
if (_bLabelFileOpen)
{
load_LabelFile_stream.Close();
_bLabelFileOpen = false;
}
//load Mnist Images files.
if (!MnistImageFileHeader())
{
MessageBox.Show("Can not open Image file");
_MnistImageFileName = null;
_bImageFileOpen = false;
_bDatabase = false;
return(false);
}
if (!MnistLabelFileHeader())
{
MessageBox.Show("Can not open label file");
_MnistLabelFileName = null;
_bLabelFileOpen = false;
_bDatabase = false;
return(false);
}
//check the value if image file and label file have been opened successfully
if (_LabelFileBegin.nItems != _ImageFileBegin.nItems)
{
MessageBox.Show("Item numbers are different");
CloseMinstFiles();
_bDatabase = false;
return(false);
}
m_pImagePatterns = new List <ImagePattern>(_ImageFileBegin.nItems);
_iRandomizedPatternSequence = new int[_ImageFileBegin.nItems];
for (int i = 0; i < _ImageFileBegin.nItems; i++)
{
byte m_nlabel;
byte[] m_pPatternArray = new byte[MyDefinations.g_cImageSize * MyDefinations.g_cImageSize];
ImagePattern m_ImagePattern = new ImagePattern();
GetNextPattern(m_pPatternArray, out m_nlabel, i, true);
m_ImagePattern.pPattern = m_pPatternArray;
m_ImagePattern.nLabel = m_nlabel;
m_pImagePatterns.Add(m_ImagePattern);
}
_bDatabase = true;
CloseMinstFiles();
return(true);
}
static void ReadTraining()
{
for (int i = 0; i < TrainingSamples; ++i)
{
ImagePattern IP = Bayesian_Classifier._MnistTrainingDatabase.m_pImagePatterns[i];
int LabeledClass = IP.nLabel;
TrainingLabeledClass[i] = LabeledClass;
++NumberOfTrainingLabeledSamples[LabeledClass];
for (int j = 0; j < NumberOfFeatures; ++j)
{
if (IP.pPattern[j] == 255)
{
TrainingFeatures[i, j] = 1;
}
else
{
TrainingFeatures[i, j] = 0;
}
}
}
}
private void btn_Show_Click(object sender, EventArgs e)
{
Bitmap Bmap = new Bitmap(28, 28);
int End = 28;
int i = 0, j = 0, k = 0;
byte PixelValue;
ImagePattern IP = null;
if (rdoBtn_TrainingSet.Checked)
{
IP = Bayesian_Classifier._MnistTrainingDatabase.m_pImagePatterns[(int)(numericUpDown1.Value - 1)];
}
else if (rdoBtn_TestingSet.Checked)
{
IP = Bayesian_Classifier._MnistTestingDatabase.m_pImagePatterns[(int)(numericUpDown1.Value - 1)];
}
while (i < 28)
{
k = 0;
for (j = i * 28; j < End; j++)
{
PixelValue = IP.pPattern[j];
if (chckBx_Threshold.Checked && PixelValue < 255)
{
PixelValue = 0;
}
Bmap.SetPixel(k, i, Color.FromArgb(PixelValue, PixelValue, PixelValue));
k++;
}
i++;
End = (i + 1) * 28;
}
label1.Text = IP.nLabel.ToString();
pictureBox1.Image = (Image)Bmap;
if (rdoBtn_TestingSet.Checked)
{
textBox1.Text = Bayesian_Classifier.TestingClassifiedClass[(int)(numericUpDown1.Value - 1)].ToString();
}
}
