HYPERSPECTRAL SENSING SYSTEM AND PROCESSING METHODS FOR HYPERSPECTRAL DATA
A hyperspectral sensing device may include an optical collector configured to collect light and to transfer the collected light to a sensor having spectral resolution sufficient for sensing hyperspectral data. In some examples, the sensor comprises a compact spectrometer. The device further comprises a power supply, an electronics module, and an input/output hub enabling the device to transmit acquired data (e.g., to a remote server). In some examples, a plurality of hyperspectral sensing devices are deployed as a network to acquire data over a relatively large area. Methods are disclosed for performing dark-current calibration and/or radiometric calibration on data obtained by the hyperspectral sensing device, and/or another suitable device. Data obtained by the device may be represented in a functional basis space, enabling computations that utilize all of the hyperspectral data without loss of information.
1 . A computer-implemented method for predicting ground-truth data corresponding to remotely measured data, the method comprising:
acquiring a ground-truth spectrum corresponding to light measured at a first location at a first time;
acquiring first remote spectral data corresponding to the first location at the first time;
determining first weighting coefficients of a ground-truth function representing the ground-truth spectrum in a functional basis space;
determining second weighting coefficients of a first remote function representing the first remote spectral data in the functional basis space;
determining a correlating relationship predicting the first weighting coefficients based on the second weighting coefficients;
acquiring second remote spectral data and determining third weighting coefficients of a second remote function representing the second remote spectral data in the functional basis space; and
using the correlating relationship to predict, based on the third weighting coefficients, projected ground-truth weighting coefficients corresponding to a projected ground-truth function representing a projected ground-truth spectrum in the functional basis space.
2 . The method of claim 1 , wherein acquiring the ground-truth spectrum comprises measuring a spectrum of light using a hyperspectral sensing device disposed at the first location at the first time.
3 . The method of claim 1 , wherein the first remote spectral data comprises multi-spectral data measured by a sensor carried by a satellite.
4 . The method of claim 1 , wherein the second remote spectral data corresponds to a second location different from the first location.
5 . The method of claim 1 , wherein the functional basis space is defined by a set of radial basis functions.
6 . The method of claim 1 , wherein determining the first weighting coefficients includes performing a least-squares regression on the ground-truth spectrum.
7 . The method of claim 1 , wherein determining the correlating relationship includes fitting the second weighting coefficients to the first weighting coefficients using Tikhonov regularization.