Spectral Signatures are the specific combination of reflected and absorbed electromagnetic (EM) radiation at varying wavelengths which can uniquely identify an object. The spectral signature of stars indicates the spectrum according to the EM spectrum. The spectral signature of an object is a function of the incidental EM wavelength and material interaction with that section of the electromagnetic spectrum. The measurements can be made with various instruments, including a task specific spectrometer, although the most common method is separation of the Red, Green, Blue and Near Infrared portion of the EM spectrum as acquired by digital cameras. Calibration of spectral signatures under specific illumination are collected in order to apply an empirical correction to airborne or satellite imagery digital images.
When using a spectrometer, you will look through it at a tube of gas with high voltages passing through it. In the Spectrometer, you will be able to see specific rays of colour falling on graduations on the inside. Each substance will have its own unique falling of rays.
Most remote sensing applications process digital images to extract spectral signature at each pixel and use such signature to divide the image in groups of similar pixels (segmentation) using different approaches, and, as last step, assign a class to each group (classification) by comparing with known spectral signatures. Depending on pixel resolution, a pixel can represent many spectral signature "mixed" together - that is why a lot of remote sensing analysis is done to "unmix mixtures". Ultimately correct matching of spectral signature recorded by image pixel with spectral signature of existing elements leads to accurate classification in remote sensing.