How to calculate the average rgb color values of a bitmap

Question

In my C# (3.5) application I need to get the average color values for the red, green and blue channels of a bitmap. Preferably without using an external library. Can this be done? If so, how? Thanks in advance.

Trying to make things a little more precise: Each pixel in the bitmap has a certain RGB color value. I'd like to get the average RGB values for all pixels in the image.

Answer 1

This kind of thing will work but it may not be fast enough to be that useful. from here http://www.dreamincode.net/code/snippet3879.htm

public static Color getDominantColor(Bitmap bmp)
{

       //Used for tally
       int r = 0;
       int g = 0;
       int b = 0;

     int total = 0;

     for (int x = 0; x < bmp.Width; x++)
     {
          for (int y = 0; y < bmp.Height; y++)
          {
               Color clr = bmp.GetPixel(x, y);

               r += clr.R;
               g += clr.G;
               b += clr.B;

               total++;
          }
     }

     //Calculate average
     r /= total;
     g /= total;
     b /= total;

     return Color.FromArgb(r, g, b);
}

Answer 2

The fastest way is by using unsafe code:

BitmapData srcData = bm.LockBits(
            new Rectangle(0, 0, bm.Width, bm.Height), 
            ImageLockMode.ReadOnly, 
            PixelFormat.Format32bppArgb);

int stride = srcData.Stride;

IntPtr Scan0 = dstData.Scan0;

long[] totals = new long[] {0,0,0};

int width = bm.Width;
int height = bm.Height;

unsafe
{
  byte* p = (byte*) (void*) Scan0;

  for (int y = 0; y < height; y++)
  {
    for (int x = 0; x < width; x++)
    {
      for (int color = 0; color < 3; color++)
      {
        int idx = (y*stride) + x*4 + color;

        totals[color] += p[idx];
      }
    }
  }
}

int avgR = totals[0] / (width*height);
int avgG = totals[1] / (width*height);
int avgB = totals[2] / (width*height);

Beware: I didn't test this code... (I may have cut some corners)

This code also asssumes a 32 bit image. For 24-bit images. Change the x*4 to x*3

Answer 3

Here's a much simpler way:

Bitmap bmp = new Bitmap(1, 1);
Bitmap orig = (Bitmap)Bitmap.FromFile("path");
using (Graphics g = Graphics.FromImage(bmp))
{
    // updated: the Interpolation mode needs to be set to 
    // HighQualityBilinear or HighQualityBicubic or this method
    // doesn't work at all.  With either setting, the results are
    // slightly different from the averaging method.
    g.InterpolationMode = InterpolationMode.HighQualityBicubic;
    g.DrawImage(orig, new Rectangle(0, 0, 1, 1));
}
Color pixel = bmp.GetPixel(0, 0);
// pixel will contain average values for entire orig Bitmap
byte avgR = pixel.R; // etc.

Basically, you use DrawImage to copy the original Bitmap into a 1-pixel Bitmap. The RGB values of that 1 pixel will then represent the averages for the entire original. GetPixel is relatively slow, but only when you use it on a large Bitmap, pixel-by-pixel. Calling it once here is no biggie.

Using LockBits is indeed fast, but some Windows users have security policies that prevent the execution of "unsafe" code. I mention this because this fact just bit me on the behind recently.

Update: with InterpolationMode set to HighQualityBicubic, this method takes about twice as long as averaging with LockBits; with HighQualityBilinear, it takes only slightly longer than LockBits. So unless your users have a security policy that prohibits unsafe code, definitely don't use my method.

Update 2: with the passage of time, I now realize why this approach doesn't work at all. Even the highest-quality interpolation algorithms incorporate only a few neighboring pixels, so there's a limit to how much an image can be squashed down without losing information. And squashing a image down to one pixel is well beyond this limit, no matter what algorithm you use.

The only way to do this would be to shrink the image in steps (maybe shrinking it by half each time) until you get it down to the size of one pixel. I can't express in mere words what an utter waste of time writing something like this would be, so I'm glad I stopped myself when I thought of it. :)

Please, nobody vote for this answer any more - it might be my stupidest idea ever.