The International Archives of the Photogrammetry, Remote Sensing and Spatial Information Sciences. Vol. XXXVII. Part B7. Beijing 2008
1206
Combing GVI and curvature feature with certain conditions
applied into sub-areas after JSEG segmentation, we can obtain
forest areas (CHM-GVI-JSEG). The comparison between
Gabor wavelet texture features between CHM-GVI-JSEG forest
area and sub-areas obtained from JSEG lead to successful forest
boundary delineation. Fig. 8 shows a schematic work flow of the
overall forest delineation process.
3. EXPERIMENTAL RESULT AND CONCLUSION
JSEG has one parameter which controls the region merge
process during the segmentation. In our experiments, we set it
as 0 so that we can obtain over-segmented sub-areas which
have more homogeneous texture pattern. Examples of the forest
delineation result are shown as Fig.9a (high-lighted with white
lines) while Fig. 9b shows the manual forest boundary
delineation result according to NFI forest definition, (high
lighted with red lines)
Fig. 9a Final forest Fig. 9 Manual delineation
Boundary derived from JSEG
Experimental results indicate that forest area can be
distinguished by using remote sensing data, e.g. aerial images
and LIDAR data. The color features extracted from aerial
images remove successfully non green areas while the curvature
feature extracted from LIDAR remove building areas. The
comparison of texture features extracted from Gabor wavelet
between CHM-GVI-JSEG forest area and sub-areas after JSEG
segmentation gives the chance for detecting sub-areas with low
quality CHM and lead to forest boundary delineation in more
semantic way. The approach presented in this paper offers an
automatic process for forest/non forest detection in Swiss NFI.
It’s a challenge to describe NFI forest /non-forest definition
with automatic computer-based method. For example, some
temporary unstocked areas are still delineated manually as
forest because their forest land-use functions. It will be difficult
to interpret such kind of subjective definition with low level
features extracted from remote sensing data. However, current
study provides an encouraging basic for further development
and testing.
REFERENCE:
Brassel P. (Ed.),(2001). Swiss national forest inventory:
methods and models of the second assessment. Birmensdorf:
Swiss Federal Research Institute WSL
Brassel P. and Brandli (Eds.),(1999). Schweizerisches
Landesforstinventar Ergebnisse der Zweitaufnahme 1993-1995.
Bern: Haupt
Deng Y. and B. S. Manjunath,2001. Unsupervised segmentation
of color-texture regions in images and video. IEEE
Transactions on Pattern Analysis and Machine Intelligence
23(8): pp. 800-810.
Desclee B., P. Bogaert and P. Defoumy,2006. Forest change
detection by statistical object-based method. Remote Sensing of
Environment, 702(1-2): pp. 1-11.
Donoghue D. N. M., P. J. Watt, N. J. Cox and J. Wilson,2007.
Remote sensing of species mixtures in conifer plantations using
LiDAR height and intensity data. Remote Sensing of
Environment, 770(4): pp. 509-522.
Haara A. and M. Haarala,2002. Tree species classification using
semi-automatic delineation of trees on aerial images.
Scandinavian Journal of Forest Research, 77(6): pp. 556-565.
Husch B., C. I. Miller and T. W. Beers, 1982. Forest
mensuration 3rd ed. New York: John Wiley & Sons,
J. Hyyppa H. H., et al,2000. Accuracy comparison of various
remote sensing data source in the retrieval of forest stand
attributes. Forest Ecology and Management, 128: pp. 109-120.
Lu D. S., P. Mausel, E. Brondizio and E. Moran,2004.
Relationships between forest stand parameters and Landsat TM
spectral responses in the Brazilian Amazon Basin. Forest
Ecology and Management, 795(1-3): pp. 149-167.
M. A. Lefsky W. B. C., and T. A. Spies,2001. An evaluation of
alternate remote sensing products for forest inventory,
monitoring, and mapping in Douglas-fir forests of western
Oregon Canadian Journal of Forest research, 31: pp. 78-87.
Maltamo M., K. Eerikainen, J. Pitkanen, J. Hyyppa and M.
Vehmas,2004. Estimation of timber volume and stem density
based on scanning laser altimetry and expected tree size
distribution functions. Remote Sensing of Environment, 90(3):
pp. 319-330.
McRoberts R. E., E. O. Tomppo, A. O. Finley and J. Heikkinen,
2007. Estimating areal means and variances of forest attributes
using the k-Nearest Neighbors technique and satellite imagery.
Remote Sensing of Environment, 777(4): pp. 466-480.
Morsdorf F., Meier, E., Kotz, B., Itten, K.I., Dobbertin, M. and
Allgower, B.,2004. LIDAR-based geometric reconstruction of