2

I need to interpolate whole 2d array in with c#.

I managed to write an interpolation class. It should work as intpl2 function of matlab.

slice: is the 2d array that going to be interpolated xaxis: interpolated x axis yaxis: interpolated y axis

It maps as v0 v1 | v3 v2

EDIT: Solve the issue, but I am open to more elegant solution :)

        public static float[,] ArrayBilinear(float[,] slice, float[] xaxis, float[] yaxis)
        {

            float[,] scaledSlice = new float[xaxis.Length, yaxis.Length];
            double xscale = ((xaxis.Length - 1) / (slice.GetLength(0) - 1));
            double yscale = ((yaxis.Length - 1) / (slice.GetLength(1) - 1));

            for (int x = 0; x < scaledSlice.GetLength(0) ; x++)
            {
                for (int y = 0; y < scaledSlice.GetLength(1) ; y++)
                {


                    
                    float gx = ((float)x) / xaxis.Length * (slice.GetLength(0) - 1);
                    float gy = ((float)y) / yaxis.Length * (slice.GetLength(1) - 1);

                    int xInOriginal = (int)gx;
                    int yInOriginal = (int)gy;

                    double v0 = slice[xInOriginal, yInOriginal];
                    double v1 = slice[xInOriginal + 1, yInOriginal];
                    double v2 = slice[xInOriginal + 1, yInOriginal + 1];
                    double v3 = slice[xInOriginal, yInOriginal + 1];

                    double newValue = (1 - GetScale(xscale, x)) * (1 -  GetScale(yscale, y)) * v0 + GetScale(xscale, x) * (1 - GetScale(yscale, y)) * v1 + GetScale(xscale, x) *  GetScale(yscale, y) * v2 + (1 - GetScale(xscale, x)) *  GetScale(yscale, y) * v3;

                    scaledSlice[x, y] = (float)newValue;
                }
            }




            return scaledSlice;
        }

       private static double GetScale(double scale, int i)
        {
            if (i == 0)
                return 0;
            if (i % scale > 0)
                return (i % scale) / scale;
            return 1.0;
        }

         public static float[] LinSpace(float start, float stop, int length)
        {
            if (length < 0)
            {
                throw new ArgumentOutOfRangeException(nameof(length));
            }

            if (length == 0) return new float[0];
            if (length == 1) return new[] { stop };

            float step = (stop - start) / (length - 1);

            var data = new float[length];
            for (int i = 0; i < data.Length; i++)
            {
                data[i] = start + i * step;
            }
            data[data.Length - 1] = stop;

            return data;
        }

Test with random 2d array

  [Test]
        public void testBilinearInterpolation()
        {
            float[,] array2D = new float[4, 4] { { 0, 0, 0, 5 }, {  1, 1, 4, 0 }, { 0, 1, 3, 1 }, { 0, 4, 0, 1 } };

         

            var xaxes = LinSpace(0, 3, 10);
            var yaxes = LinSpace(0, 3, 10);
            var intepolatedarr = ArrayBilinear(array2D, xaxes , yaxes);

        }

The interpolation result is wrong (Different from interp2 from matlab) I couldn't found why.

ephraim
  • 379
  • 1
  • 3
  • 15
Sugee
  • 79
  • 8
  • It seems you swapped `v1` and `v3`. Btw, I find this code structure very weird. Why is `ArrayBilinear` setting class members instead of passing them as parameters? This will get you into serious trouble as soon as you want to multi-thread. Making all methods static sounds like a more reasonable approach. Also, why are you scaling the coordinates? – Nico Schertler Sep 01 '19 at 17:16
  • Yeah I saw the v1, and v3 thank you. Code is just for test, I will convert it to static class. But this not solves my issue. – Sugee Sep 01 '19 at 17:54
  • Then what issue do you have? Where do the (20, 20) and (77, 77) come from? – Nico Schertler Sep 01 '19 at 18:22
  • The results are not right, I believe somethings wrong in my xinorignal and yinorginal calculation. (20,20 and (77,77) is just and example. I am trying to convert matlab interp2 function to c#. I Converted it to a static function. – Sugee Sep 01 '19 at 19:11
  • Can you please post a [mcve] with some data that shows the problem? Tell us what you get and what you expect. – Nico Schertler Sep 02 '19 at 00:32
  • I found the problem. – Sugee Sep 02 '19 at 08:04

0 Answers0