Prospects for biological and chemical substance classification using a standoff laser induced fluorescence detection system

Kurzfassung

After accidental or intended release of chemical or biological hazardous materials, fast detection and classification is mandatory to be able to rapidly take steps against negative consequences on humans. Standoff laser based devices are suited to localize and classify a thread on short timescales without the need of direct human contact to these materials. In this scope we present the classification prospects of our standoff detection system that utilizes laser induced fluorescence signals excited with laser pulses of two different UV excitation wavelengths. For fast data acquisition we use a rather limited spectral resolution of 32 channels per wavelength which is sufficient to achieve a good performance. The described system is able to perform a complete classification procedure including data acquisition, data analysis and classification and to get an assignment of the measured sample after a few seconds. The identification and classification prospects are evaluated by classifying measured sample spectra using a model which is based on a previously acquired reference database. Within the set, different representatives for bacterial and chemical agents as well as background materials as diesel and pollen are investigated. The performance, namely the classification accuracy and false alarm rates, is evaluated for differently sized classes down to an identification of the investigated set of materials.