|
|
|
|
|
Because compounds have unique spectral patterns, spectral pattern recognition is used to identify unknown compounds. The spectrum of standard compounds, therefore, must be identified. Search of large databases takes much time, and clear differentiation among similar spectra is difficult. Given this situation, the Institute has moved to reduce the processing time and differentiate among very similar spectra by applying neural network technology, and by having a neural network learn spectral patterns. Ongoing research in this area is focused on automating the process of identifying compounds from their spectral patterns so that the technique may be applied to areas such as the identification analysis of automotive paint, on-the-spot analysis of mineral components, and non-destructive analysis of food components.
Stage wise neural network |