Hyperspectral Imaging

Objects that are hidden from view or very hard to pick out against a confusing background can be clearly revealed using 'hyperspectral imaging'. This uses special cameras that are able to analyse a scene into its component spectrum wavelengths. Custom software then searches the scene for objects with a predifined spectral 'fingerprint' . If a match is found, then the object is exposed.

Conventional broad band sensors (eg thermal imagers) have great difficultly in detecting targets when they are in low contrast with the background. This can occur under a number of circumstances such as dusk and dawn or where the target is camouflaged or against a cluttered background. Conventional sensors also have problems discriminating between targets and decoys.

The picture (above) illustrates a scene with a real figure and a decoy. After processing the image data with a real time hyperspectral algorithm, the real figure is detected.

Hyperspectral data can also be used to recognise an object by its spectral signature, such as a particular military uniform. This technology has potential applications such as distinguishing friend from foe, aiding search and rescue operations, forensic analysis, face recognition and forgery detection.