EmguCV image depth problem with Code Laboratory Kinect Calibration

I'm using the Emgu wrapper for OpenCV in C#, and I have the following nagging problem.

I'm using the Code Laboratories Kinect API and the code to get an image out of the Kinect looks like this:

           //Forms the array to store the image data in. 
           uint[] imageSource = new uint[640 * 480];

            //EmguCV needs everything to be IntPtrs, so wrapping the array in one.
            GCHandle handle1 = GCHandle.Alloc(开发者_Python百科imageSource, GCHandleType.Pinned);
            try
            {
                imageSourcePointer = handle1.AddrOfPinnedObject();
            }
            finally
            {
                if (handle1.IsAllocated)
                {
                    handle1.Free();
                }
            }

            imagePointer = IntPtr.Zero; 
            //Getting the camera information.
            IntPtr camera = CLNUIDevice.CreateCamera(CLNUIDevice.GetDeviceSerial(0)); 
            //Pulling from the image.
            CLNUIDevice.GetCameraColorFrameRGB24(camera, imageSourcePointer, 0);
            imagePointer = CvInvoke.cvCreateImageHeader(new Size(640, 480), IPL_DEPTH.IPL_DEPTH_8S, 1);
            //TODO change the widtstep to be generic based on the camera. 
            MIplImage image = (MIplImage) Marshal.PtrToStructure(imagePointer, typeof(MIplImage)); 
            //Connecting the data array to the image definition's header.
            CvInvoke.cvSetData(imagePointer, imageSourcePointer, image.widthStep);

Now, all my IntPtr wrapping seems to work since the data they point to get updated just fine.

The problem comes when I try to use the function:

//For finding the corners of a chessboard in the given image. 
int found = CvInvoke.cvFindChessboardCorners(imagePointer, board_sz, cornerPointer, ref corner_count, Emgu.CV.CvEnum.CALIB_CB_TYPE.ADAPTIVE_THRESH | Emgu.CV.CvEnum.CALIB_CB_TYPE.FILTER_QUADS);

And I get the runtime exception which says

"OpenCV: Only 8-bit grayscale or color images are supported"

I've tried all the color pull methods from the Code Laboratory Kinect - the RGB24, RGB32 and RAW pull commands, but the exception remains. I've declared my image (imageSource) as an 8 bit signed integer image, and didn't expect the error to come on.

I tried to narrow down why this was happening, and thought -

• The image pulled from the Kinect, despite being stored in an 8-bit OpenCV image header, still has its own bit depth properties.

• The uint[] that I'm using as the data array for the Kinect remains a 32 bit array, which leads the entire image into becoming a 32-bit image. This still doesn't seem as likely, though possible.

I've been stuck on this for a while, so any suggestions would be really helpful. Thanks so much!

EDIT - I think I've figured this out - some things that I didn't initially understand about Emgu. I'll post solution code soon if anyone does end up need to look at this.