Image Processing Reference
In-Depth Information
Fig. 1.2 The schematic representation of the spectral response of the scene captured by the different
sensor systems: a the single sensor system, b the multispectral system, and c the hyperspectral
system. The rectangle window defines the region of integration of the spectral response
better than those of the single sensor image. If the bands are captured corresponding
to the wavelengths of the red, green and blue colors, then it is possible to obtain an
RGB image of the scene along with other bands. Humans perceive colors better than
the grayscale changes. As the MS data facilitates the color (RGB) visualization of the
scene, human analysts and observers can have a better understanding of the scene.
We illustrate the schematic comparison of how the data at a particular pixel would
appear in the case of the MS image in Fig. 1.2 . The rectangular block represents
the response of an ideal single sensor. The output of this sensor is obtained by the
integration of the reflectance response of the scene over the bandwidth of the sensor
element. The entire response is represented by a single value which is the average
value of the response over the sensor bandwidth. In the illustration, we represent this
value by (
). However, being a single number, it does not provide much information
about the scene contents. On the other hand, the multispectral sensor captures the
scene response at multiple wavelength bands. Figure 1.2 b represents an illustration
of the response by an ideal 4-band multispectral sensor. The scene response has been
captured by 4 different sensor elements that have a lesser individual bandwidth than
that of the single sensor as in Fig. 1.2 a. Therefore, the responses of individual sensor
elements capture the scene responses better. The multispectral bands, thus, are able
to provide a more detailed information regarding the scene in the form of multiple
data samples.
However, even the multispectral images do not meet the challenges posed by
advanced applications of the remote sensing data. The small number of bands in the
MS image is not enough to discriminate between two materials that are somewhat
Search WWH ::

Custom Search