New technology maintains the data in multispectral images

Computer scientists at the University of East Anglia (Norwich, UK) have developed novel techniques for visualizing hyperspectral and multispectral images in color.

A number of key applications capture images with hyperspectral and multispectral imaging systems. The data are often viewed by analysts to perform tasks such as target detection and classification. The key challenge is to communicate the spectral image data in a way that maximizes the information content available to the final decision maker.

A standard approach to this visualization problem is to choose three channels from those available and to arbitrarily assign them as red, green, and blue (RGB) for display on a color monitor. This downside of this approach is that all the information contained in the remaining channels is lost.

The University of East Anglia color group's technology measures the local contrast in the multichannel image and generates an RGB output image with matching local contrast. A result of this is that important details, lost by standard methods, "pop out," making them visible to decision maker or analyst.

For moore information on the technique, see www.uea.ac.uk/business/ideas/spectraledge.

-- By Dave Wilson, Senior Editor, Vision Systems Design