Hu Xiangyun 
  
experiments have been done for practical mapping from images. 
Figure 4 shows a road extraction. There are two seed points at the end of the road segment. The crosses in (b) are the 
max-correlation points. Some of them are errors. Part (c) shows the result after segmented curve fitting and optimization 
by Hopfield neural network. The result indicates the effect of the algorithm. 
Figure 5 is a semiautomatic extraction result of a road network in an aerial image of which scale is 1:25000, all 
extracted central lines of road are fitted by cubic spline. Part (b) is a snapped vector window of the result. 
  
  
  
  
  
(a) | | (b) 
Figure 5. Extraction result of a road network 
5 | CONCLUSIONS 
The practice indicates that our algorithm and human-machine cooperation has good accuracy, steady output, high speed 
and good interactivity and is much more effective than traditional manual digitizing. Our algorithm uses some 
experiential parameters or thresholds, such as the resampling range (searching range of template correlation), error 
restriction of curve fitting and weight coefficient of weight function of the Hopfield network. Face to practical 
production of digital mapping, it is good as long as it accomplishes the task. 
However, our model of road extraction is based on a "light ribbon like ' feature, there are still problems in extraction in 
downtown area and color image. The template might be more complex and the optimization should integrate more 
factors, such as texture and color information. That is the farther work. 
REFERENCES 
Fischler M A., 1981. Detection of roads and linear structures in low resolution aerial imagery using a multi-source 
knowledge integration technique. Computer Graphics and Image Processing, Vol 15, pp.201-223. 
Gruen A,Li H H., 1996. Linear feature extraction with LSB snakes from multiple images. In: International 
Archives of Photogrammetry and Remote Sensing, Vol 31,Part B3, pp. 266-272. 
Gruen A, Li H H., 1995. Road Extraction from aerial and satellite images by dynamic programming. Journal of 
photogrammetry and remote sensing, Vol 30,pp. 11-20. 
Gruen A, Li H H., 1995. Semiautomatic road extraction as a model driven optimization procedure. Digital 
Photogrammetry and Remote Sensing 95* SPIE Proceedings Series, Vol 2646, pp. 160-171. 
  
998 International Archives of Photogrammetry and Remote Sensing. Vol. XXXIII, Part B3. Amsterdam 2000.