×

You are using an outdated browser that does not fully support the intranda viewer.
As a result, some pages may not be displayed correctly.

We recommend you use one of the following browsers:

Full text

Title
Fusion of sensor data, knowledge sources and algorithms for extraction and classification of topographic objects
Author
Baltsavias, Emmanuel P.

International Archives of Photogrammetry and Remote Sensing, Vol. 32, Part 7-4-3 W6, Valladolid, Spain, 3-4 June, 1999
(c) (d)
Fig. 5. (a) 255 x 255 detail of MOMS-2P Band 3; (b) 153 x 153 TM (Bands 4-5-7 as R-G-B colour composite); (c) MOMS-2P
Band 3 spectrally enhanced by TM Bands 5 and 7; (d) high-frequency components injected into the TM Bands 5 and 7 (biased and
rescaled for displaying purpose). All the data are shown on a 18 m scale.
4. EXPERIMENTAL RESULTS
The area of San Rossore pinewood, of which MOMS-2P and
Landsat TM data were available, is located on the Tyrrhenian
coast of Tuscany, in Central Italy. The TM data were collected on
04/08/90, while the MOMS-2P image on 15/03/97. The gap oc
curred between the two acquisitions is compensated by the little
amount of scene variations, which justifies the fusion experiment.
In fact, soil use, which includes evergreen woodland, crops, and
urban area is mostly unchanged, at least in the zone analysed.
All the images (a whole stripe of MOMS and a slice of the
TM scene covering the same area) were registered on the same
cartographic base, each retaining its own scale. A first-order
polynomial model was used together with cubic resampling. The
absence of relief makes the accuracy of registration (a complete
affine transformation) be satisfactory when measured on control
points (absolute error less than one fourth of pixel) and, visually,
on the whole (less than one pixel). Obviously, registration accu
racy influences the performances of data fusion, but that is not
the concern of this experiment. The pyramid method was applied
with a ratio p/q = 5/3 and two levels of decomposition (K = 1).
Since the three available multi-spectral bands of MOMS-2P,
labeled as Band 1 to 3, overlap with Bands 2, 3 and 4 of TM, the
experiment consists of synthesizing the three missing MOMS-2P
bands (one in the blue and two in the infrared wavelengths) from
the TM data. In the former case TM Band 1 (blue) was spatially
enhanced to 18 m scale by means of high-pass features borrowed
from MOMS-2P Band 2 (green, the most similar) and displayed
as R-G-B true colour together with the available Bands 2 and 3
from MOMS. In the latter case two infrared bands at 18 m were
created from TM Bands 5 and 7 to which the high-pass frequen
cies of MOMS-2P Band 3 (near infrared) were superimposed.