IEEE Geoscience and Remote Sensing Magazine - December 2020 - 27

turn processed by the image DPS. The meteorological image
data from both the CCD and VHRR sensors are first digitally
stored in a data acquisition system. The data are then coupled
with a real-time data reception system through a computer
interface unit. In turn, the header and framing information is
added to the data. The images are then taken up for geometric
and radiometric corrections. After successful data registration,
the images are available for the generation of various software
products by applying appropriate algorithms.
EXAMPLE OF NEED FOR IMAGE REGISTRATION IN
KALPANA-1 AND INSAT-3A DATA
In this section, we illustrate the need for image registration
in the VHRR and CCD data using a real example for each.
A scenario is shown in Figure 3 using two ground control
points (GCPs). It should be noted that the images are displayed using SAC Image Display software developed by the
Data Products Software Group. Figure 3(a) and (b) points
out a need for registration in a VHRR image acquired by the
KALPANA-1 payload. The same portion of two images of the
same (IR) band and same time (0600) but acquired on different dates is shown in full-resolution windows. As shown
in Figure 3(a) and (b), a point near Muscat (i.e., scan: 461,
pixel: 652) in Figure 3(a) is shifted to scan: 457, pixel: 652 in
Figure 3(b). Similarly, Figure 3(c) and (d) depicts the need
for data registration in a CCD image acquired by the INSAT3A imager. The same portion of two images of the same band
acquired on different dates and times is shown. A same point
on Gujarat state (i.e., scan: 2641, pixel: 3031) in Figure 3(c) is
shifted to scan: 2672, pixel: 3039 in Figure 3(d). Such random
deviations over the sensed area raise the need for the image registration in meteorological imagery. Considering the
ground (spatial) resolution of both of the images (Tables 1
and 3), it is clear that the data need to be registered with

Full Resolution Window Near Muscat

K1VHR_29SEP2007_0600_IR.rad K1VHR_25OCT2007_0600_IR.rad

(a)

(b)

their respective GCPs. Hence, correction is required for the
generation of accurate data products for further analysis.
METHODOLOGY AND SOFTWARE DEVELOPMENT FOR
METEOROLOGICAL IMAGE REGISTRATION
We now discuss the methodology and development of software
for registering the meteorological imagery from KALPANA-1
and INSAT-3A. We first define the problem using a block diagram of a general registration process. We highlight the steps,
preliminaries, challenges, and major requirements for this
operational software. We then describe the operational blocks
with theoretical details. Toward the end of this section, we present a simplified flowchart of the implemented software.
The registration of images (sensed images) is to be done
with a reference image that is considered as a template. The
complete implementation is carried out in the following steps:
1) automatic identification of the GCPs in the sensed data, 2)
finding the optimal transformation model based on the match
points between the sensed and reference images, and 3) resampling the transformed imagery to the reference coordinates.
As shown in a general schematic (Figure 4), the incoming images of the KALPANA-1 and INSAT-3A are treated as
sensed images; hence, correction is required geometrically
with respect to the reference template, whereas radiometric
statistics of a registered image should be close to (ideally the
same as) the sensed image. Preliminaries and major requirements from this software are as follows:
1) A set of GCPs may be selected or given from the reference
image.
2) The algorithm should automatically be able to identify
the corresponding match points in the sensed image,
find the most appropriate transformation from that set
of match points, and perform the necessary resampling
to warp the sensed image onto the reference coordinates.

Full Resolution Window Near Gujarat

3ACCD_29OCT2006_0701_03.rad 3ACCD_11FEB2007_0901_03.rad

(c)

(d)

FIGURE 3. The need for image registration in the VHRR and CCD data acquired on two different days. (a) and (b) A shift in the VHRR image
near Muscat and (c) and (d) a shift in the CCD image near Gujarat. The images are shown using SAC Image Display software.
DECEMBER 2020

IEEE GEOSCIENCE AND REMOTE SENSING MAGAZINE

IEEE Geoscience and Remote Sensing Magazine - December 2020

Table of Contents for the Digital Edition of IEEE Geoscience and Remote Sensing Magazine - December 2020