32-bit grayscale BufferedImage?

Question

I've got a byte array corresponding to pixel values of data type unsigned short (written by a C++ application). I need to convert these values into a BufferedImage and display or save it.

I'm using the following loop:

    int xsize=1024;
    int ysize=1024;

        for (int i = 0; i < bandCount; i++) 
        {
            ints[i]=new int[pixels];
            for(int j=0;j<pixels*2;j+=2)
            {

                short temp=bands[i].getShort(index);

                ints[i][index]=(temp&0xffff);
                index++;
            }
        }
        imgBuffer = new DataBufferInt(ints, pixels);
        buffer_type = DataBuffer.TYPE_INT;
        sampleModel = new BandedSampleModel(buffer_type, xsize, ysize,
                xsize, banks, offsets);
        data_type = BufferedImage.TYPE_USHORT_GRAY;

    }
    WritableRaster raster = Raster.createWritableRaster(sampleModel,
            imgBuffer, null);
    BufferedImage img = new BufferedImage(xsize, ysize, data_type);
    img.setData(raster);

Where pixels is simply the width*height of the image. bands[] is the ByteBuffer containing the pixel values. I read in the next short, AND it into an int value, and after the loop, create a DataBuffer of these int values. Then I try to create a BandedSampleModel from the converted pixel values. The DataBuffer type is INT (I think) and the BufferedImage is set to be USHORT_GRAY.

The image looks horrible- it's skipping every other pixel, I think. I've tried changing the BufferedImage type (data_type above), but there's no 32-bit grayscale option. I've tried using TYPE_INT_ARGB and gotten an exception stating that setData on the BufferedImage is using an out-of-bounds index(1024). The image is much larger than 1024x1024; I locked xsize and ysize to those values to reduce the amount of time spent producing an error.

I'm a bit out of my depth here; how can I create a BufferedImage using an array of unsigned short pixel values? Am I reading the pixel values incorrectly in the loop, or am I simply setting the DataBuffer or BufferedImage types incorrectly?

Thanks!

Answer 1

I'd use ImageJ > Import > Raw… , which handles unsigned short correctly. It has a convenient macro facility, too.

Addendum: One convenience of using ImageJ is that it maps the range of unsigned short values present in the input to an evenly-spaced palette of 256 RGB gray values. You can see the effect by importing a raw image and selecting Analyze > Histogram . Of course, you can apply any desired LUT using Image > Lookup Tables .