nbul 2004
should. be
efinitions.
imagery
ata
d Data
]ded Data
Aodels for
adiometric
rroved the
[S-Ll as a
ulary, 4th
cations
bols, and
1S
nitions for
ology of
y. Names,
in Optical
1-9.
1s AD-784
adiometric
ymmended
Analysis of
52
d Practice:
red Sensor
is. Space
n. UT.
1. Ground-
R Sensors.
ques, New
n of Space
ite Sensing
al Camera
te Sensing.
iblished by
ing and the
International Archives of the Photogrammetry, Remote Sensing and Spatial Information Sciences, Vol XXXV, Part Bl. Istanbul 2004
32 Terms and Definitions
32. Standard Definitions: Terms and definitions provided
in the resources cited above are not included in this paper, but
there is overlap between some of those terms and those
discussed at the Task Force Workshop. These will have to be
reconciled in future deliberations.
3.2.2 Task Force Workshop Terms:
absolute geolocational accuracy: the closeness of the
agreement between image-defined location of a point or feature
with respect to ground coordinates and the true location of that
feature, typically expressed as a circular error or linear error.
See also systematic error, and random error, both of which
affect absolute accuracy.
absolute radiometric accuracy: the difference between the
radiance measured by the sensor and the true radiance of a
source. Radiometric accuracy is determined by comparing a
calibrated source that can be traced to a radiometric standard.
Absolute radiometric accuracy depends on stability,
polarization, stray light, linearity and other components
[pk1][pk2].
band-to-band registration: refers to how well the same scene
is recorded in different spectral bands (Kramer 2001). It is the
relative geometric registration between bands, usually specified
in terms of a fraction of a pixel.
bidirectional reflectance distribution function: the ratio of
radiance leaving a target to the irradiance incident upon the
target
calibration: the process of quantitatively defining the system
responses to known, controlled signal inputs.
central wavelength: For a given spectral band, the central
wavelength is the wavelength at the centre of the spectral
bandwidth.
circular error-90 (CE90): a metric to describe horizontal
accuracy (X and Y coordinates) in map or image products at the
90% confidence level (that is, 90% of well-defined points tested
fall within a certain radial distance). CE90 defines the radius of
a circle that encompasses 90% of the points.
combined standard uncertainty: combination of the
uncertainty components considering their interdependence or
correlation.
contrast transfer function (CTF): a 2-dimensional measure of
a system's sensitivity to detect rectangular patterns of
diminishing spatial width.
a) Edge target = difference between the dark and bright area
pixel values measured in digital numbers (DNs), divided by the
1-sigma standard Deviation (SD) noise level.
cross-talk (optical & electronic): In an imaging system,
spectral cross-talk is a measure of electromagnetic energy
leakage from one spectral band (spectral response for a given
band) to another.
dynamic range: the ratio of the maximum observable energy
and the minimum still useful energy (max signal/min signal); it
is defined as 10 log. The maximum signal is the signal at which
the system saturates while the minimum signal is usually
defined as the noise floor. All radiant energy less than the
minimum vanishes into noise, while the energy above the
maximum disappears into the saturation of the detector.
(Kramer, 2001) The dynamic range of an imaging system refers
to the range of radiance or reflectance values (and digital
numbers) that can be measured (recorded). Systems with low
dynamic range will saturate when measuring bright (highly
reflective) targets or not be able to resolve features at low
radiance levels. The dynamic range, or range of brightness
values, is also often defined by the system quantization, or the
number of bits. An 8-bit system will allow for 28 = 256
brightness levels, while a 16-bit system will allow for 216 —
65536 brightness levels.
edge spread function (ESF): the image signal corresponding to
an edge as an input signal (CCRS 2004). a measure of a
system's ability to distinguish a straight, high contrast edge.
effective instantaneous field-of-view (IFOV): the resolution
corresponding to a spatial frequency (ground resolution) for
which the system MTF is 50%.
emittance: the ratio of a target's radiance to the radiance
emitted from an ideal blackbody at the same thermodynamic
temperature of the sample
expanded uncertainty: the product of the combined standard
uncertainty and a coverage factor (k) whose value is chosen
based on a desired level of confidence (usually k = 2).
general image quality equation (GIQE): Leachtenauer et al.
(1997) provide a mathematical relationship between GSD,
RER, image enhancement, and SNR to estimate image quality
objectively. The GIQE also includes ringing and edge sharpness
effects associated with an edge response. One commonly
accepted form of the GIQE expresses this relationship in terms
of the National Imagery Interpretability Rating Scale (NIIRS).
Geo-locational accuracy (circular error-90 [(CE90]): a
standard metric to describe horizontal accuracy in map or image
products at the 90% confidence level (that is, 90% of well-
defined points tested must fall within a certain radial distance).
ground sample (ing) distance (GSD): the distance between the
center of adjacent pixels in an intrinsic sensor image (that is,
not a re-sampled image); the distance between the centres of
adjacent pixels. See also instantaneous field-of-view.
instantaneous field of view (IFOV): the (angular) aperture
within which the sensor is sensitive to electromagnetic
radiation. It may be expressed either as a small solid angle or as
a unit area (Kramer 2001). When expressed in degrees or
radians (solid angle), this is the smallest plane angle over which
an instrument is sensitive to radiation. When expressed in linear
or area units such as meters or hectares, it is an altitude
dependent measure of the ground resolution of the sensor
(CCRS, 2004). Often, GSD is equated with the spatial
resolution of a pixel, or simply with IFOV. However, this need
not be the case. If the radiometric and electronic performance of
a sensor allow, GSD can be made smaller than IFOV to achieve
better image quality because of the reduction of smear (Kramer,
2001).
