Hyperspectral cameras

[originalfoo]

Most of the sensors required for the heliotropic platter could be condensed in to a single hyperspectral camera - if suitable components can be found.

Sensors

There are two main options identified so far (from cameras list):

• iMEC– ~600-1000 nm - already in production and purported to be affordable (although I've yet to see a price list). Existing cameras based on the tech are available, including very small ones that would fit within the tight confines of the photonics module.
• Hamamatsu – 165 to 1100 nm - easily the best option at the moment but poses many technical challenges (particularly temperature control).

Possible future contenders (not yet in production as far as I can tell) - all of these are targeting mobile phone market; tiny, inexpensive, low power and easy-to-integrate sensors:

• VTT
• PARC
• Unispectral - note: Apartheid regime

More basic options including limiting imaging to multispectral, just IR or very low resolution (often single pixel):

• Melexis - FIR sensor which works within desirable specifications (especially temperature).
• ULIS - thermal imaging cameras, much better resolution than Melexis
• Espros - like an RGB sensor, but does 400 to 900 nm (in currently released product as far as I can tell; earlier press releases stated there would be two variants: 387 – 903 nm, and 776 – 1064 nm)
• AS7262 and AS7263
• Pyreos - very low power consumption
• GridEYE – 8x8 thermopile array

Tech issues

Many of the existing hyperspectral cameras have very constrained operational temperature specifications which would make them unsuitable for use in weather extremes.

They also consume lots of energy, in the order of hundreds of milliamps. This severely limits the number of images that can be generated per day due to energy harvesting constraints. We would likely need a supplementary external power source (larger PV array for example) to operate at full capability.

Processing raw data would likely require either a FPGA for realtime processing, or remote server for post-processing.

Transmitting hypercube files would eat up bandwidth and power, especially over cellular connections.

Ideally we'd want to merge spectra in realtime; similar to infragrams, but with custom spectra being applied to RGB image. This would greatly reduce telemetry bandwidth and provide images in a simple format.

In cases where we do want to retain hypercube we would likely need either additional RAM for buffering or limit the number of spectral layers per capture.

Real time camera mode (eg. for drone use) would likely prove troublesome.