Imaging spectrometers are essential instruments for hyperspectral remote sensing systems that capture pupil-plane spectral radiance from a scene and ultimately form it into a hyperspectral data cube for processing, analysis, and information extraction. An imaging spectrometer is an EO/IR system, thus it contains all of the basic imaging system components outlined in the previous chapter, exhibits similar design challenges, and is characterized by some of the same performance metrics. In addition to imaging, however, an imaging spectrometer must also capture radiation spectra for each image pixel with adequate fidelity; this places additional demands on system design and adds another dimension to performance characterization.
The primary design challenge is that a hyperspectral data cube is essentially a 3D construct, with two spatial dimensions and one spectral dimension, while FPAs are 2D. Therefore, some means are necessary to capture this additional dimension. Often this is done using the time dimension. Various designs that perform this function in different ways are explored in this chapter and subsequent two chapters. In this chapter, we focus on arguably the most common type, the dispersive imaging spectrometer.
Dispersive spectrometers use either a prism or grating to spatially spread a spectrum of incident radiation across a detector array. In a dispersive imaging spectrometer, depicted conceptually in Fig. 7.1, the spectrum is dispersed in one direction, while the image is relayed in an orthogonal direction. In the direction of dispersion, it is necessary to limit the field of view to one spatial element; this is performed using a slit. If the field of view is not thusly limited, irradiance distribution on the detector array will resemble an image with large lateral chromatic aberration. In that way, a back-end spectrometer only captures a single line of the 2D field produced by front-end optics. A scanning mechanism, such as the pushbroom or whiskbroom method discussed in Chapter 6, is required to image from frame to frame in a direction orthogonal to the slit. As this occurs, a hyperspectral data product is formed as a sequence of spatial-spectral frames, as depicted in Fig. 7.2. The composite set forms a hyperspectral data cube.