Pixel and Image Characteristics. Prof. Arnon Karnieli The Remote Sensing Laboratory Jacob Blaustein Institute for Desert Research Ben-Gurion University of the Negev Sede-Boker Campus 84990, ISRAEL. Pixel (picture element).
Prof. Arnon Karnieli
The Remote Sensing Laboratory
Jacob Blaustein Institute for Desert Research
Ben-Gurion University of the Negev
Sede-Boker Campus 84990, ISRAEL
A pixel having both spatial and spectral properties. The spatial property defines the "on ground" 2 dimensions. The spectral property defines the intensity of spectral response for a cell in a particular band.
Digital number (DN) =
Gray Level (GL) =
Brightness Value (BV)
At sensor radiance
Optical system, detectors, electronics
(W m-2 sr-1m-1)
The output (DN) is proportional to the input (at sensor radiance)
A row of pixels represents a scan line collected as the sensor moves left to right or collected through the use of a linear array of photodetectors.
An image is composed of pixels geographically ordered and adjacent to one another consisting of \'n\' pixels in the x direction and ‘m\' pixels in the y direction.
When only one band of the EM spectrum is sensed, the output device (color monitor) renders the pixels in shades of gray (there is only one data set).
Multispectral sensors detect light reflectance in more than one or two bands of the EM spectrum. These bands represent different data. When combined into the red, green, blue guns of a color monitor, they form different colors.
A multispectral image is composed of \'n\' rows and \'n\' columns of pixels in each of three or more spectral bands. There are in reality more than one "data set" which makes up one image.
These different data sets are referred to as spectral bands, bands, or channels.
Resolution - The smallest observable (measurable) difference.
Instantaneous Field of View (IFOV) = Pixel
Field of View (FOV)
Ground projected Instantaneous Field of View (GIFOV)
GIFOV depends on satellite height (H)
Ninety 61 cm mirrors, 2.25 km across.
Landsat MSS - 80 m NOAA-AVHRR - 1,100 m Meteosat - 5,000 m
Scale - mathematical relationship between the size of objects as represented on maps, aerial photographs, or images. Measured as the ratio of distance on an image to the equivalent distance on the ground.
1 cm on the map represents 50,000 cm or 0.5 km on the ground
2 bit - 4 levels
3 bit - 8 levels
1 bit - 2 levels
4 bit - 16 levels
6 bit - 64 levels
8 bit - 256 levels
Part of the IKONOS (11-bit acquisition level) image is under cloud shadow. It can be recovered due to high radiometric resolution.
The features under cloud shadow are recovered by applying a simple contrast and brightness enhancement technique.
Laboratory Kaolinite spectrum convolved in various signal to noises
High temporal resolution is important for:
- infrequent observational opprtunity (e.g., when clouds often obscure the surface)
- short-lived phenomenon (floods, oil spills, dust storms, etc.)
- rapid response (fires, hurricanes)
- detection changes properties of a feature to distinguish it from otherwise similar features (phenology)
Pixel values (DNs) are scaled to byte values:
Lλ = "gain" * DN + "offset"
Lλ= Spectral radiance at the sensor’s aperture in watts/(meter2*ster*µm)
"gain" = Rescaled gain in watts/(meter2*ster*µm)
"offset"= Rescaled bias in watts/(meter2*ster*µm)
“gain” and “offset” values are provided with the image.
Which is also expressed as:
Lminλ= the spectral radiance that is scaled to DNmin in watts/(m2 * ster * µm)
Lmaxλ= the spectral radiance that is scaled to DNmax in watts/(m2 * ster * µm)
DNmin = the minimum quantized calibrated pixel value (corresponding to Lminλ) in DN = 0Dnmax = the maximum quantized calibrated pixel value (corresponding to Lmaxλ) in DN = 255
Lmin, Lmax = radiance in w m-2st-1m-1
Example for the Landsat ETM+ sensor, high gain, after July 1, 2000
p= Unitless planetary reflectance
L= Spectral radiance at the sensor\'s aperture
d = Earth-Sun distance in astronomical units from nautical handbook
ESUN = Mean solar exoatmospheric irradiances
s= Solar zenith angle in degrees