16 bit grayscale BufferedImage

Posted by Patrick Pirrotte on
URL: http://imagej.273.s1.nabble.com/16-bit-grayscale-BufferedImage-tp3703089.html

Hi all,

I've run into trouble trying to convert a 16 bit grayscale BufferedImage to
an ImagePlus entity. Digging through the mailing list, I found out that
several other people had the same trouble:

using FileOpener on 16-bit grayscale in java app ? (06/04/10)
BufferedImage (05/09/15)
loading 16 bit images from other java code, ImageJ 1.28 (05/08/12)
...

It seems that the mechanism underlying setImage(BufferedImage image) does
not interprete the ColorModel correctly. I'm not quite sure about this,
maybe my code is buggy:

DataBuffer dataBuffer = new DataBufferUShort ((short[])data,1);
ImageTypeSpecifier it = ImageTypeSpecifier.createGrayscale(16,
DataBuffer.TYPE_USHORT, false);
ColorModel cm = it.getColorModel();
SampleModel sm = it.getSampleModel(256,256);
WritableRaster wr = Raster.createWritableRaster(sm,dataBuffer,new
Point(0,0));
BufferedImage image = new BufferedImage(cm,wr,true,new Hashtable());

When I create an ImagePlus from this BufferedImage I get an 24 bit RGB...
(ImagePlus im = new ImagePlus("my image",image))

I'd prefer to read my pixel data directly from ImageJ through the getPixel
method but unfortunately I have to rely on a BufferedImage. I've quickly
browsed through Jarek's ImagePlusCreator (IJ-ImageIO adapter plugin), and as
far as I understand it works around this problem by getting the Raster, the
ColorModel,... from the BufferedImage resulting from ImageIO. Using those it
recreates a valid 16 bit ImagePlus.

Maybe I'm not seeing the big picture here, but my question is: wouldn't it
be nicer to adapt ImagePlus.setImage and it's submethods to understand a
signed and unsigned short ColorModel from BufferedImage? Or am absolutely
out of line and my code is just plain wrong (which happens far too often to
be pleasant).

Thanks,

Patrick