spectral
ationary
nt does
dulated
ntensity .
present
esponse
red He-
halogen
dure of
esponse
ints are
maging
996).
ng high
nd the
solution
and the
nttila et
vides a
(SNR).
am it is
icy. of
ata-rate
coming
'trieved
ition, a
in the
Section
1 of the
alibrate
Section
| and in
Sagnac
International Archives of the Photogrammetry, Remote Sensing and Spatial Information Sciences, Vol XXXV, Part B-YF. Istanbul 2004
P
Figure 1. Conceptual layout of the stationary interferometer
arranged in the Sagnac (triangular or cyclic path) geometry
The light from the object is first collimated by the objective L,
and, by means of a beam-splitter BS and two folding mirrors
M1 and M2, is focused onto a CCD plane by the lens P. Like
the Michelson device, this Sagnac interferometer also uses a
beam-splitter (as source-doubling technique) but inside of a
common (triangular) path with a tilt introduced between the two
interfering image wavefronts to form a fixed (stationary) pattern
of interference fringes of “equal thickness”.
The BS is the fundamental component, which provides phase-
delay between the two coherent interfering rays so that the OPD
changes linearly with varying the angle of the entering ray with
respect to the instrument optical axis. Therefore, the device
acquires the image of an object superimposed to a fixed pattern
of across-track interference fringes. Then, introducing a relative
motion between the sensor and the object, each scene pitch
experiences the same interference pattern and an entire image-
frame is recorded for each autocorrelation phase offset.
The 3-D array of these frames has to be “realigned” in order to
extract the complete autocorrelation function (interferogram) of
each scene pixel, then this record is inverse cosine transformed
to yield a wavenumber hyperspectral data cube.
Differently from the FTHSI optical configuration, our prototype
interferometer doesn't employ the slit nor the cylindrical lens in
front of the detector array. In fact, in the developed version a
complete 2-D image of the observed scene is acquired with
superimposed the autocorrelation modulation pattern.
The main feature is that we need acquire as many frames as the
number of photosensitive elements. For example, if we utilize a
detector array with 512 pixels, it is necessary to acquire a set of
512 records to reconstruct the complete autocorrelation function
of each scene pixel. It is clear that the requested data-rate is
very high.
Figuré 2 shows a picture of the developed instrument.
Figure 2. Picture of the developed instrument
The characteristics of the CCD camera used to spatially resolve
the interferogram are listed in Table 3.
Type: CCD frame-transfer with
anti-blooming, TH7888A
1024 x 1024
Number of pixel:
Pixel size 14 um x 14 um
430 nm-1000 nm @QE>3%
Spectral range:
Responsivity: 23DN/(nJ em^) @450 nm
18% @680 nm
Dynamic range: 3200:1
Maximum frame/rate: | 60 fps
Digitalization: 12 bit
Table 3. Characteristics of the frame grabber used to spatially
resolve the interferogram record
In our case, the data cube of a typical measurement contains
more than 1 million data points for a total size of 2 MB before
cosine transform.
3. PHYSICAL BACKGORUND
In a Fourier Transform Spectrometer (FTS) the acquired.
physical information is the interferogram /(OPD), which
depends on OPD values according to the following fundamental
law (Barnes, 1985; Okamoto et al, 1984):
19 (k
I(OPD) = | ) [1+ cos(2nkOPD dk (1)
2
where [1 (k)
interferometer, OPD is the optical path difference introduced
by the BS in the considered propagation direction, k =14A is
the wavenumber and the integration takes into account for a
non-monochromatic light spectrum.
is the intensity of the ray entering the
