International
line intersections mostly correspond for road intersection, and
these also coincide with block corners.
Energy minimization starts from making initial curve and
proceeding with the refinement. Among all road intersection
points, each four points become seed points for one initial
curve. Based on these seed points, we insert nodes for which
spacing is proportional to the distance from one seed point to
another. Figure 10a is a detailed example of one initial curve
with different point classification. A yellow dot is an
intersection point, all blue points are inserted nodes, the large
blue points are classified as corner area, and all points comprise
the initial V, in equation (9). We apply different internal
weighting coefficients based on point location. We apply «a and
B value for corner area as 1 and 3, meanwhile give a 300 and p
0 for side area. With matrix 4 and the initial V, we iteratively
solve for V, for each time evolution step until the change of
total energy is less than threshold. A result, which represents
the city blocks, is in Figure 10b. To enhance the sensitivity to
image forces, we use a spatial diffusion technique to spread the
influence of the edges. Figure 11 is the result of applying
adaptive snakes to the study area. We eliminate the non-
converged curve during minimization.
Luke ere
a indue ani
Figure 10. Adaptive snakes for urban road grids. (a) Local
coefficients. (b) Detected city blocks.
Figure 11. Detected city blocks in the study area
5. Conclusions
In this paper, we propose a new approach for urban road
extraction. A basic assumption is that urban road network has a
nominal grid pattern and a priori knowledge required is the
approximate minimum size of the city block and approximate
road width.
In region segmentation, we subdivide the region until all
regions have only two dominant directions. By doing so, we
Archives of the Photogrammetry, Remote Sensing and Spatial Information Sciences, Vol XXXV, Part B3. Istanbul 2004
can obtain each region's dominant road directions even if the
road network has multiple grid patterns. The determined road
directions were used for road detection. Searching the region
with its road directions, we find the candidacy of lines on the
road by using the *acupuncture method'. These detected lines
were used to construct initial approximations for the subsequent
snake refinement. Line intersection were considered as road
intersections and used for seed points for the snakes. Applying
local weighting coefficients to the traditional snakes, we have
developed an adaptive procedure well fitted for urban city block
delineation. Our future work will focus on achieving better
stability in the numerical estimation, improving the adaptive
algorithm, and comparing different methods of computing the
steps in the line fitting.
References
Agouris, P., A. Stefanidis, and S. Gyftakis, 2001. Differential
snakes for change detection in road segments, Photogrammetric
Engineering & Remote Sensing, 67(12):1391-1399.
Baumgartner, A., C. Steger, H. Mayer, W. Eckstein, and H.
Ebner, 1999. Automatic road extraction based on multi-scale,
grouping, and context, Photogrammetric Engineering and
Remote Sensing, 65(7):711-785.
Davis, D.N., K. Natarajan, and E. Claridge, 1995. Multiple
energy function active contours applied to CT and MRI images
of the brain, Proceedings of the 5th IEE Conference on Image
Processing and its Applications, Edinburgh, UK, pp. 114-1 18.
Gonzalez, R., and R. Woods, 1992. Digital Image Processing.
Addison Wesley Publishing Company, Reading, Massachusetts,
458 p.
Gruen, A., and H. Li, 1997. Semi-automatic Linear feature
extraction by dynamic programming and LSB-snakes.
Photogrammtric Engineering & Remote Sensing. 63(8): 985-
995.
Heipke, C., C. Steger, and R. Multhammer, 1995...
Hierarrchical Approach to Automatic Road Extraction From
Aerial Images, SPIE — Aerosense 1995 Symposium, 17-21 April,
Orlando, USA, Vol. 2486, pp.222-231.
Hinz, S., A. Baumgartner, H. Mayer, C. Widemann, and H.
Ebner, 2003. Road Extraction Focusing on Urban Area,
Automatic Extraction of Man-Made Objects from Aerial and
Space Images(III), (E. Baltsavias, A. Grün, and L. Van Gool,
editors), A.A. Balkema Publishers, Lisse, The Netherlands, pp.
255-265.
Kass, M., A. Witkin, and D. Terzopoulos, 1988. Snakes: Active
contour models, /nternational Journal of Computer Vision,
1(4):321-331.
Price, K., 1999. Road Grid Extraction and Verification,
International Archives of Photogrammetry and Remote Sensing,
32(Part 3/2w5):101-106.
Sohn, G., and I.) Dowman, 2001. Extraction of buildings from
high-resolution satellite data, Automatic Extraction of Man-
Made Objects From Aerial and Space Images(ll1) (E.
Baltsavias, A. Grün, and L. Van Gool, editors), A.A. Balkema
Publishers, Lisse, The Netherlands, pp. 345-355.
Vosselman, G., and J. de Knecht, 1995. Road tracing by profile
matching and Kalman filtering, Automatic Extraction of Man-
Made Objects from Aerial and Space Images (A. Gruen, O.
Kuebler, and P. Agouris, editors), Birkhaeuser Verlag, Berin,
Germany, pp. 265 — 274.
KEY V
ABST]
The ai
Photog
develo;
and his
Armed
militar
which
vector
GML2
The se
softwar
tested
The pa
area fe
imager
images
interac
efforts
point,
Baltsay
(auton
stage ©
2001).
be the
& Mil
The ai
softwa
Photog
Univer
*
Cor
