| abstract |
An optical identification system includes a light source with a broad wavelength spectrum that is directed on an object to be identified. Suitable optical components, such as, one or more collimating lenses gather light that is reflected from the object and direct this light into a spectrometer. The spectrometer disperses the collimated light using a dispersing element, such as one or more gratings, prisms or a combinations of both, onto an array of detectors. The array of detectors may be comprised of a linear diode array or a charge-coupled device (CCD) array which indicates the amount of light at each of a finely-spaced set of wavelengths covering a wide spectral range. The detectors are sensitive over a wavelength region, for example, in the case of silicon detectors from near-infrared plus the visible region, e.g., from 250 nm to 1100 nm. The set of signals from the detectors is read with an analog to digital converter, and transferred to a computer in the form of a spectrum. A set of known spectra determine the reference spectra and the unknown test spectrum is compared with the reference sets. A software program in the computer compares the test spectrum with reference spectra sets utilizing a statistical program. The program takes into account how much the known spectra vary from one another in addition to the average values. A display reads out a list of possible matches in rank order that have a probability of match greater than a predetermined threshold. An operator checks that the first listed item is correct and either accepts the first choice or indicates the correct choice. As an alternative, the system could automatically accept the first choice. |