How imagej convert 16bit to 8bit?

emartini · April 8, 2021, 12:45pm

Dear forum
Does someone know which math function is used in ImageJ to convert 16bit grayscale to 8bit grayscale?
The one used by Menu->Image->Type->8bit

thanks
Emanuele

Bio7 · April 8, 2021, 1:02pm

Here is the Java source:

github.com

imagej/imagej1/blob/32aeb17f323ea102afb5fe1c66332070fd1395fa/ij/process/TypeConverter.java#L49


			case FLOAT:
				return convertFloatToByte();
			case RGB:
				return convertRGBToByte();
			default:
				return null;
		}
	}

	/** Converts a ShortProcessor to a ByteProcessor. */
	ByteProcessor convertShortToByte() {
		int size = width*height;
		short[] pixels16 = (short[])ip.getPixels();
		byte[] pixels8 = new byte[size];
		if (doScaling) {
			int value, min=(int)ip.getMin(), max=(int)ip.getMax();
			double scale = 256.0/(max-min+1);
			for (int i=0; i<size; i++) {
				value = (pixels16[i]&0xffff)-min;
				if (value<0) value = 0;
				value = (int)(value*scale+0.5);

emartini · April 8, 2021, 2:56pm

thanks some ignorance here…
what (pixels16[i]&0xffff) will do?

The bottom line of this question is to replicate the 16bit to 8bit conversion in Matlab.

Bio7 · April 8, 2021, 3:09pm

See:

http://imagej.1557.x6.nabble.com/Handling-signed-and-unsigned-16-bit-images-td3693185.html