3D Hyperspectral Processor

Hyperspectral remote sensing with area array sensors can generate data cubes consisting of two spaatial axes and one spectral axis. Over a short collection period these data cubes can constitute a tremendous data reduction bottle neck for time critical missions. The proposed 3D Hyperspectral Sensor and Processor integrates a real time processing engine with the area array sensor to greatly improve the data reduction bottle neck. In the proposed example, nearly a 1000 to 1 data compression is achieved on a hyperspectral system collecting 512 x 512 spatial pixels by 512 spectral pixel data cubes in a 2 second interval. The low power (300 milliwatts) analog processor is mounted on the focal plane, integral with the detector array. The processor computes the weighted sums of the acquired spectrum with a test template and outputs a correlation or fit factor of the spectrum to the template. The processor is capable of comparing every spectra acquired against 4000 templates. The data rate between the sensor and the processor is 7x10 (super 8) Hz. This data rate is reduced to 1x10(super 6) Hz out of the processor and out of the dewar. Circuit density is achieved by construction of an 8 chip stacked module.