Yesterday I bought a Kinect and hooked it up to my PC. After a couple of hours of compiling drivers and finally realizing I had to switch USB ports for the camera to work I could start writing code.
I used the CLNUIDeviceTest as a base since it already had the lines of code required to pull the image from Kinect. My initial problem was how to access the raw bytestream of image data. I overcame this and made a small copy operation that “makes it appear as if you are walking through a wall”. At least that was the point, but the Kinect obviously needs some calibrating and I’m copying source image wrong. Both easy to fix.
Anyhow I just wanted to share the code used for raw image access. Speed is great
|
1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 26 27 28 29 30 31 32 33 34 35 36 37 38 39 40 41 42 43 44 45 46 47 |
public class ImageProcessor { private byte[] myImageBytes; public ImageProcessor() { JpegBitmapDecoder myImage = new JpegBitmapDecoder(new Uri("brick_wall.jpg", UriKind.Relative), BitmapCreateOptions.DelayCreation, BitmapCacheOption.OnLoad); myImageBytes = new byte[myImage.Frames[0].PixelWidth * 4 * myImage.Frames[0].PixelHeight]; myImage.Frames[0].CopyPixels(myImageBytes, myImage.Frames[0].PixelWidth * 4, 0); } public void Process(NUIImage colorImage, NUIImage depthImage, NUIImage processedImage) { unsafe { byte* colorImgPtr = (byte*)colorImage.ImageData; byte* srcImgPtr = (byte*)depthImage.ImageData; byte* dstImgPtr = (byte*)processedImage.ImageData; int height = (int)depthImage.BitmapSource.Height; int width = (int)depthImage.BitmapSource.Width; int stride = width * 4; for (int x = 4; x < height; x++) { for (int y = 0; y < width - 4; y++) { int p = x * (width * 4) + (y * 4); //dstImgPtr[p + 0] = (byte)srcImgPtr[p + 0]; //dstImgPtr[p + 1] = (byte)srcImgPtr[p + 1]; //dstImgPtr[p + 2] = (byte)srcImgPtr[p + 2]; //dstImgPtr[p + 3] = (byte)srcImgPtr[p + 3]; byte b = srcImgPtr[p + 0]; dstImgPtr[p + 0] = b > (byte)128 ? colorImgPtr[p + 0] : myImageBytes[p + 0]; dstImgPtr[p + 1] = b > (byte)128 ? colorImgPtr[p + 1] : myImageBytes[p + 1]; dstImgPtr[p + 2] = b > (byte)128 ? colorImgPtr[p + 2] : myImageBytes[p + 2]; dstImgPtr[p + 3] = b > (byte)128 ? colorImgPtr[p + 3] : myImageBytes[p + 3]; //int avg = (srcImgPtr[p + 0] + srcImgPtr[p + 1] + srcImgPtr[p + 2] + srcImgPtr[p + 3]) / 4; //dstImgPtr[p + 0] = (byte)avg; //dstImgPtr[p + 1] = (byte)avg; //dstImgPtr[p + 2] = (byte)avg; //dstImgPtr[p + 3] = (byte)avg; } } } } } |
And a sample picture…
As you can see the black region on the right represents the picture while the blue region represents the height. I’m guessing I’m sitting a bit too close, maybe I should be processing the green channel instead.