An illustration of a hyperspectral data cube [src]. The target is discretized into pixels by the camera

The hyperspectral data cube is the result of a hyperspectral image acquisition by a spectrometer. It is a three-dimensional data structure that contains an image of the target at each of the captured wavelengths. The size of the datacube in memory is calculated as:

$$ D=n_{\text{px}x} n{\text{px}x} n{\text{px}y} n{\text{bit}} $$

Where:

• $D$ [bit]: Size of datacube
• $n_{\text{px}_x}$ [px]: Number of pixels on the sensor in the across-track direction. This is the axis on which spatial information is encoded
• $n_{\text{px}_y}$ [px]: Number of pixels on the sensor in the along-track direction. This is the axis of the sensor on which spectral information is encoded
• $n_{\text{bit}}$ [bit/px]: Bit depth of the sensor

Assumptions:

• The image acquisition method is pushbroom
• The spatial image aspect ratio is square

The equation comes out of the fact the datacube has side lengths that are $n_{\text{px}x}$ by $n{\text{px}x}$ by $n{\text{px}_y}$. We multiply these dimensions together to obtain the volume, or total number of voxels (3D pixels) in the cube, which we then multiply by the size of their representation in digital bits