get the trimmed tree as Tab.(25). Tab.(26) is the result of 
matching, after executing distance function. 
6. Conclusion 
Form above theory analysis and examples, we can see 
that the description based on relational data structure 
makes the relational description very feasibility and 
precision. The trimming approach that the paper 
presented by using relational subset, one to one 
correspondence and one to one correspondence between 
relational subsets and unit-label table can efficiently 
simplify the procedure of consistency labeling. Finally 
the distance function make it possible to realize precision 
match. 
It should be noticed that, as to more complex recognition 
for more complex object, if selecting even more unit 
attributions and relational constraints, we also can get 
well simplified results. On the other hand, the idea of the 
paper is popular, and it also can be used for the 
recognition task using curved surfaces as units, if the 
446 
International Archives of Photogrammetry and Remote Sensing. Vol. XXXI, Part B3. Vienna 1996 
relational data structure is structed by means of the 
proper structure constraints. The trimming algorithm 
and distance function in the paper are also available. 
Reference 
l. Tsai and Fu. "Error-correcting Isomorphisms of 
Attributed Relational Graphs for Pattern Analysis", 
IEEE. Trans. on SMC. Vol. SMC-9. No. 12. p757-768. 
1979. 
2. Shapiro. L. G., Moriorty. R.M. et. "Matching Three- 
Dimension Objects Using a Relational Paradigm". 
Pattern Recognition . Vol.17, No.4.,p385-405, 1984. 
3. Haralick. R.M. ,Shapiro.L.G. "The Consistent 
Labeling Problem". IEEE. Trans., PAMI., Vol. PAMI-2, 
No.3,p04- 519, 1980. 
4. P.JFlynn and A.K.Jain. "CAD-based Computer 
Vision: from model to relation graphs". IEEE 
Trans.PAMI, Vol.13, No.2,PP.114-132,1991. 
    
    
      
    
   
    
   
   
   
   
    
   
   
    
   
   
    
    
Tab.1 
  
Tab. 
  
  
  
[ Uni 
  
  
  
SERRE AIMER SRE