| Term |
Definition/Explanation |
| Data Cube |
A data cube is the three dimensional (one spectral and two spatial dimensions) dataset commonly used in hyperspectral imaging. Of course, the data "cube" is an actual cube in very rare cases. The hyperspectral dataset should be dubbed "cuboid" to be correct (anyway, the term "cube" is commonly used within the scientific community). |
| Spectral Jitter |
The variation of the spectral calibration accuracy along the pixels of an image. Spectral jitter can be a problem in airborne NIR/UVVIS spectrometers.
|
| Spectral Misregistration |
This occurs if a spectrometer that requires a substantial time for spectral measurement is moved during the measurement. Thus the spectrum of a single pixel is not taken from a contant spot but from a line along the movement of the spectrometer. |
| Spectral Resolution |
Spectral resolution results from the convolution of the spectral sample spacing with the spectral bandpass inherent to a particular spectrometer. In many applications the sample spacing is significantly smaller than the spectral bandwidth of the instrument. With some spectroscopic techniques (e.g. low resolution mass spectrometry) the sample spacing is equal to spectral "bandwidth" (i.e. peak intensities are mapped to integer masses).
|
| Spectral Smile |
An artifact which occurs if the center wavelength of a band varies across the spatial dimension. Plotting an image of constant wavelength results in an arc (thus "spectral smile").
|
Appendix 
