//http://www.gutgames.com/post/Matrix-Convolution-Filters-in-C.aspx
public static MatrixData ApplyConvolutionFilter(this MatrixData Input, int[,] filter)
{
MatrixData NewBitmap = new MatrixData(Input.Width(), Input.Height());
MatrixData OldData = Input;
double MeanPixel = Input.ModePixel();
int Width = filter.GetLength(1);
int Height = filter.GetLength(0);
for (int x = 0; x < Input.Width(); ++x)
{
for (int y = 0; y < Input.Height(); ++y)
{
double Value = 0;
double Weight = 0;
int XCurrent = -Width / 2;
for (int x2 = 0; x2 < Width; ++x2)
{
if (XCurrent + x < Input.Width() && XCurrent + x >= 0)
{
int YCurrent = -Height / 2;
for (int y2 = 0; y2 < Height; ++y2)
{
if (YCurrent + y < Input.Height() && YCurrent + y >= 0)
{
double Pixel;
try
{
Pixel = Input[YCurrent + y, YCurrent + x];
}
catch
{
Pixel = MeanPixel;
}
Value += filter[x2, y2] * Pixel;
Weight += filter[x2, y2];
}
++YCurrent;
}
}
++XCurrent;
}
double PixelVal = Input[y, x];
if (Weight == 0)
{
Weight = 1;
}
if (Weight > 0)
{
Value = System.Math.Abs(Value);
Value = (Value / Weight);
Value = (Value > 255)? 255 : Value;
PixelVal = Value;
}
NewBitmap[y, x] = PixelVal;
}
}
return(NewBitmap);
}
public static MatrixData DownScale(this MatrixData input, int factor, bool favorBrightPx = true)
{
MatrixData output = new MatrixData(input.Height() / factor, input.Width() / factor);
for (int r = 0; r < output.NumberOfRows; r++)
{
for (int c = 0; c < output.NumberOfColumns; c++)
{
MatrixData inpx = new MatrixData(input, r * factor, c * factor, factor, factor);
output[r, c] = (favorBrightPx)? (inpx.MeanPixel() + inpx.Percentile(25)) / 2: inpx.MeanPixel();
}
}
return(output);
}
public static MatrixData UpScale(this MatrixData input, int factor)
{
MatrixData output = new MatrixData(input.Height() * factor, input.Width() * factor);
for (int r = 0; r < input.NumberOfRows; r++)
{
for (int c = 0; c < input.NumberOfColumns; c++)
{
for (int oR = r * factor; oR < (r * factor) + factor; oR++)
{
for (int oC = c * factor; oC < (c * factor) + factor; oC++)
{
output[oR, oC] = input[r, c];
}
}
}
}
return(output);
}
public static MatrixData Convert18To6Px(this MatrixData input, bool favorBrightPx = true)
{
MatrixData output = new MatrixData(6, 6);
int outR = 0;
int outC = 0;
for (int r = 0; r < input.Height(); r += 3)
{
outC = 0;
for (int c = 0; c < input.Width(); c += 3)
{
MatrixData m = input.CopyData(r, c, 3, 3);
output[outR, outC] = (favorBrightPx)? (m.MeanPixel() + m.Percentile(25)) / 2 : m.MeanPixel();
outC++;
}
outR++;
}
return(output);
}
public static MatrixData ReduceDimensionByOne(this MatrixData input, int dimension)
{
bool reduceRow = dimension == 0;
int height = (reduceRow)? input.Height() - 1 : input.Height();
int width = (reduceRow)? input.Width() : input.Width() - 1;
MatrixData output = new MatrixData(height, width);
Random rand = new Random();
int[] rToSkip = Enumerable
.Repeat(0, width)
.Select(i => rand.Next(0, width))
.ToArray();
int[] cToSkip = Enumerable
.Repeat(0, height)
.Select(i => rand.Next(0, height))
.ToArray();
int rInIndex = 0;
int cInIndex = 0;
if (reduceRow)
{
int rowToSkip;
for (int c = 0; c < width; c++)
{
rowToSkip = rToSkip[c];
rInIndex = 0;
for (int r = 0; r < height; r++)
{
if (r == rowToSkip)
{
rInIndex++;
}
output[r, c] = input[rInIndex, cInIndex];
rInIndex++;
}
cInIndex++;
}
}
else
{
int colToSkip;
for (int r = 0; r < height; r++)
{
colToSkip = cToSkip[r];
cInIndex = 0;
for (int c = 0; c < width; c++)
{
if (c == colToSkip)
{
cInIndex++;
}
output[r, c] = input[rInIndex, cInIndex];
cInIndex++;
}
rInIndex++;
}
}
return(output);
}