| 摘要 |
A method for in-situ focus-fusion multi-layer spectral imaging and analysis of depth dependent particulate samples. A unique method of focus-fusion is applied to focused and defocused images acquired from multi-layer spectral imaging of a depth dependent particulate sample, in order to construct fused focused cube spectral image representations of the imaged particles, thereby generating a focused image of essentially each particle in the sample. The method of the present invention features the use of a uniquely defined and calculated focus-fusion factor parameter which combines (1) empirically determined particle physicochemical characteristics relating to (i) particle chemical composition and (ii) particle morphology, with (2) empirically determined particle spectral characteristics relating to (i) image sharpness, (ii) signal-to-noise ratio (S/N), (iii) pixel intensity, (iv) spectral distances, and (v) spectral fingerprints relating to spectral emission patterns of individual particles, and is used in critical steps of image acquisition, target detection, image analysis, and image classification. This uniquely determined parameter enables achievement of high levels of accuracy and precision in imaging and classification of the particulate samples. The method includes collecting and analyzing physicochemical and multi-layer spectral data relating to the particles in the sample, including mapping of three-dimensional positions of particles, particle sizes, and characteristics of particle emission spectra. Scene information, in the form of spectral fingerprints, derived from analysis of focus-fusion of the multi-layer spectral images is further processed using classification procedures in order to extract, on-line, in-situ physicochemical information of the particles, needed for generating a report applicable to monitoring or controlling an industrial process. |
