Open topic with navigation

Using Quantization for Floating Point

When you compress a floating point image, GeoExpress uses quantization to compress the image. You can create 32-bit floating point images in the MrSID Generation 4 format.

Quantization is a lossy compression method that reduces the number of pixel values in the image. Quantization is necessary for floating point images for two reasons:

• Performing arithmetic on floating point values results in rounding discrepancies.
• Compressing floating point values results in unnecessarily large images. This is because some of the digits in each pixel value contain data that is not significant and cannot be compressed.

As a result, GeoExpress uses a lossy compression method for floating point images, but guarantees the accuracy of the image data to an optimized precision value.

To compress a floating point image with the quantization compression method, complete the following steps:

1. Click the Add Image icon to add a floating point image to the Job List.
2. Select the image in the Job List.
3. In the Properties tab, click More Options.
4. Select MrSID Generation 4 as the output format.

GeoExpress selects the Quantization compression method automatically for floating point images.

5. Optionally, click the Set Quantization button to configure quantization precision and other quantization options.

For more information on quantization options, see Quantization Options.