518
D. Ishikawa et al.
22.1 Introduction
Spectral imaging, known also as chemical or hyperspectral imaging [1], combines
imaging and spectroscopy to obtain spatial and spectral information from a sample.
“Spectral imaging” is a general term covering a wide variety of spectroscopic and
spectrometry techniques (e.g., fluorescence, vibrational, Raman, mass spectrometry), wavelength ranges (e.g., UV, Vis, NIR, MIR) and spatial resolutions (from the
nanoscale to remote-sensing scale). Spectral images typically comprise hundreds of
wavebands for each spatial position in the image. For a given spatial pixel, combining
the intensity values at each waveband into a vector results in a spectrum, containing
information on the interaction between light and the object or objects contained
within that pixel. The resulting spectrum can be used to characterize the composition of that particular pixel. Another advantage of this technique is the ability to
measure many samples simultaneously, within the same field of view, and to combine
shape characteristics with spectral features.
NIR or short-wave infrared (SWIR) imaging typically refers to spectral images
obtained in the wavelength region from 750 to 2500 nm. Spectral images can be
arranged as three-dimensional blocks or “cubes” of data, comprising two spatial and
one wavelength dimensions, as shown schematically in Fig. 22.1. The spectral image
allows for the visualization of biochemical constituents of a sample, separated into
particular areas of the image, since regions of a sample with similar spectral properties
tend to have similar biochemical composition. Within the schematic in Fig. 22.1 is
an NIR spectral image of different packaging materials. Four different materials
are present in the image, identifiable by their shape. For instance, the squares are
high-density polyethylene (HDPE), and the rectangles are polystyrene (PS). The
reflectance spectrum of one pixel of a PS sample is plotted in the lower left-hand
side of Fig. 22.1; it exhibits three main absorption features (i.e., three troughs in the
reflectance spectrum) at 1146, 1209, 1412 nm, related to overtones of CH vibrations
within the polymer. Due to these strong features, PS can be used as a wavelength
reference material for NIR imaging. A single slice of the spectral image at 1146 nm is
also shown. In this image, the PS samples appear darker than the other ones because
they absorb more light than the other samples do at this particular wavelength.
An NIR spectral image of four different packaging materials (high-density
polyethylene (HDPE), polystyrene (PS), cardboard and polyethylene-terephthalate
(PET)) is shown. The materials were cut into different shapes to enable visual identification: squares (HDPE), rectangles (PS), circles (cardboard) and triangles (PET). The
image was obtained in diffuse reflectance using a pushbroom NIR spectral imaging
instrument, as described in [2]. A single pixel reflectance spectrum of the PS sample
is shown on the lower left-hand side. The largest absorbance peak for PS occurs at
1146 nm. On the lower right-hand side, a single wavelength image at 1146 nm is
displayed. The PS samples (rectangles in the image) appear darker than the other
materials in this image due to their relatively higher absorbance of light at 1146 nm.
As is the case for conventional NIR spectroscopy, NIR spectral images can be
obtained using different modalities, such as transmission, reflectance, transflectance
D. Ishikawa et al.
22.1 Introduction
Spectral imaging, known also as chemical or hyperspectral imaging [1], combines
imaging and spectroscopy to obtain spatial and spectral information from a sample.
“Spectral imaging” is a general term covering a wide variety of spectroscopic and
spectrometry techniques (e.g., fluorescence, vibrational, Raman, mass spectrometry), wavelength ranges (e.g., UV, Vis, NIR, MIR) and spatial resolutions (from the
nanoscale to remote-sensing scale). Spectral images typically comprise hundreds of
wavebands for each spatial position in the image. For a given spatial pixel, combining
the intensity values at each waveband into a vector results in a spectrum, containing
information on the interaction between light and the object or objects contained
within that pixel. The resulting spectrum can be used to characterize the composition of that particular pixel. Another advantage of this technique is the ability to
measure many samples simultaneously, within the same field of view, and to combine
shape characteristics with spectral features.
NIR or short-wave infrared (SWIR) imaging typically refers to spectral images
obtained in the wavelength region from 750 to 2500 nm. Spectral images can be
arranged as three-dimensional blocks or “cubes” of data, comprising two spatial and
one wavelength dimensions, as shown schematically in Fig. 22.1. The spectral image
allows for the visualization of biochemical constituents of a sample, separated into
particular areas of the image, since regions of a sample with similar spectral properties
tend to have similar biochemical composition. Within the schematic in Fig. 22.1 is
an NIR spectral image of different packaging materials. Four different materials
are present in the image, identifiable by their shape. For instance, the squares are
high-density polyethylene (HDPE), and the rectangles are polystyrene (PS). The
reflectance spectrum of one pixel of a PS sample is plotted in the lower left-hand
side of Fig. 22.1; it exhibits three main absorption features (i.e., three troughs in the
reflectance spectrum) at 1146, 1209, 1412 nm, related to overtones of CH vibrations
within the polymer. Due to these strong features, PS can be used as a wavelength
reference material for NIR imaging. A single slice of the spectral image at 1146 nm is
also shown. In this image, the PS samples appear darker than the other ones because
they absorb more light than the other samples do at this particular wavelength.
An NIR spectral image of four different packaging materials (high-density
polyethylene (HDPE), polystyrene (PS), cardboard and polyethylene-terephthalate
(PET)) is shown. The materials were cut into different shapes to enable visual identification: squares (HDPE), rectangles (PS), circles (cardboard) and triangles (PET). The
image was obtained in diffuse reflectance using a pushbroom NIR spectral imaging
instrument, as described in [2]. A single pixel reflectance spectrum of the PS sample
is shown on the lower left-hand side. The largest absorbance peak for PS occurs at
1146 nm. On the lower right-hand side, a single wavelength image at 1146 nm is
displayed. The PS samples (rectangles in the image) appear darker than the other
materials in this image due to their relatively higher absorbance of light at 1146 nm.
As is the case for conventional NIR spectroscopy, NIR spectral images can be
obtained using different modalities, such as transmission, reflectance, transflectance