To improve the lineament detection, multiple
bands are used instead of single bands. Since
each band collects different spectral information
of the same ground scene, it may highlight
different parts of linear features. Experiments
are carried out to combine multiple band
information, such as Principle Component
Analysis and Band Ratioing. Figure 3 shows the
lineament detection resulted from the first eigen
channel of the Principle Component Analysis of
all TM bands except Band 6. Figure 4 displays
the lineament detection resulted from ratioing of
Band 4 over Band 3. These resultant lineament
maps can also be overlaid or further combined,
as shown in Figure 5. Visual interpretation of the
subimage was performed by experienced
geologists. ^ Visual examination of Figures 2, 3,
4 and 5 against the visual interpretation result
indicates that all three multi-band lineament
maps provide better results than the single-band
lineament map (in Figure 2). Among the three
multi-band lineament maps, the overlay of PCA-
1 and Ratio4/3, as shown in Figure 5, detects
the most lineaments.
To summarize, it has been demonstrated that it
is possible to develop algorithms to extract
lineament information from Landsat TM imagery
of the Canadian Shield. The comparison of single-
channel and multi-channel lineament detection
shows that the best multi-band detection
achieves better results than the best single-
channel detection. This demonstrated the
potential of multi-spectral edge detection
techniques and their applications in linear feature
identification.
REFERENCES
Duda, R. O. and Hart, P. E., 1972, Use of the
Hough Transform to detect lines and curves in
pictures, Communications ACM, 15, pp. 11-15.
Hough, P. V. C., 1962, Methods and Means for
Recognizing Complex Patterns, U.S. Patent
3,069,654.
Wang, J. and Howarth, P. J., 1987, Methodology
for automated road network extraction from
Landsat TM data, in Proc. Ann ASPRS/ACSM
Convention (Baltimore, MD), 1, pp. 429-438.
752
Wang, J. and Howarth, P. J., 1989, Edge
Following as Graph Searching and Hough
Transform algorithms for lineament detection,
in Proc. IGARSS'89/12th Can. Symp. on Remote
Sensing (Vancouver, B. C.), 1, pp. 93-96.
Wang, J. and Howarth, P.J., 1990, Use of the
Hough Transform in automated lineament
detection, IEEE Transactions on Geoscience and
Remote Sensing, 28(4), pp. 561-566.
Wang, J., 1993, LINDA - a system for
automated linear feature detection and analysis.
Canadian Journal of Remote Sensing, 19(1),
pp.9-21.
Williams, R. S. Ur. 1983, Geological
Applications in the Manual of Remote Sensing,
2nd Ed., R.N. Colwell, Ed., Falls Church, VA:
Amer. Soc. for Photogrammetry, pp. 1667-
1953, 1983.
International Archives of Photogrammetry and Remote Sensing. Vol. XXXI, Part B7. Vienna 1996