When you compress a floating point image, GeoExpress automatically calculates the optimized precision value to use for quantization. Alternatively, you can manually specify a precision value.
The pixel values of the compressed output image are accurate to within half of the precision value. For example, there is a pixel in your floating point image that has a value of 3000. If you specify a precision value of 0.5, then the same pixel after quantization will have a value between 2999.75 and 3000.25.
As a rule of thumb, you can calculate the optimized precision value used by GeoExpress with the following formula:
Precision = (Maximum value - Minimum value) / 216
Whether you manually specify a precision value or allow GeoExpress to automatically calculate a precision value, GeoExpress records the precision value used during quantization in the image metadata. The precision value for each image band is stored in the IMAGE::QUANTIZATION_SCALE metadata tag. The tag lists the precision value used for each band of the image, including the alpha band. The last value in the metadata tag is for the alpha band precision value.
TIP:Open the output image in GeoExpress to view the image metadata. For more information, see Viewing Metadata.
To set a precision value, enter a value in the Precision field of the Quantization tab. The precision value that you enter depends on the range of values in the input image.
You can specify one quantization precision value for all the bands in an image, or you can specify individual precision values for each band.