A single platform, integrated by a laser-induced breakdown spectroscopy detector and a Raman spectroscopy
sensor, has been designed to remotely (5 m) and simultaneously register the elemental and molecular signatures of rocks
under Martian surface conditions. From this information, new data fusion architecture at decisions level is proposed to the
correct categorization of the rocks. The approach is based on a decisions making process from the sequential checking of
the spectral features representing the cationic and anionic counterparts of the specimen. The scrutiny of the LIBS response
by using a moving-window algorithm informs on the diversity of the elemental constituents. The output rate of emission
lines allows projecting in a loop the elements as the cationic counterpart of the rock. In parallel, the Raman response of the
unknown is compared with all the molecular counterparts of the hypothesized cation that are stored in a spectral library.
The largest similarity rate unveils the final identity of the unknown. The identification capabilities of the architecture have
been underscored through blind tests of ten natural rocks with different origins. The great majority of forecasts have
matched with the real identities of the inspected targets. The strength of this platform to simultaneously acquire the multielemental
and the molecular information from a specimen by using the same laser events greatly enhances the "onsurface"
missions for the surveillance of mineralogy. |
