219 
The International Archives of the Photogrammetry, Remote Sensing and Spatial Information Sciences. Vol. XXXVII. Part B6b. Beijing 2008 
Change type 
description 
example 
operators 
Single 
object 
Geometric 
reshape 
extension 
Tl^ _ 
—► 
T2^ 
Modify 
shortening 
T1 ^ _____ 
—► 
Modify 
reshaping 
— 
—► 
Modify 
Semantics 
transformation 
/ 
thematic 
change 
reclassification 
Class 3-> Class 2 
Modify 
renaming 
Lawson St. -> Luoyu St. 
Modify 
widening 
20M -> 30M 
Modify 
Lane change 
4 lanes -> 61anes 
Modify 
Condition 
change 
Road condition: gravel -> pitch 
Modify 
appearance 
New building 
—► 
T2 _____ 
Add 
disappearance 
abandoning 
T1 RT— 
—► 
Delete 
Multipl 
e 
objects 
connection 
connection 
T1 
/ R1 • " R2 
—► 
Delete, 
modify 
split 
split 
T1 
/ R1 
—► 
T2 ^ 
R2 ♦ R3 " 
Delete, 
modify 
Table 1. Change types analysis 
4. OBJECT MATCH PROCESS 
During the change-only updating, detection and extraction of 
change-only information is the fundamental task, and main 
approach is based on object match. We can compare spatial 
objects in semantics, geometric, thematic and topological 
aspects to determine the difference. Badard (1999) provided a 
down-to-up method to match the objects, in which geometric 
node match is processed, then the geometric object and last the 
semantics. But in this method the candidate sets have much 
more objects which lead to take long time to detect the change 
information, so it’s not suitable for large volume spatial 
database. Since we have built national foundational geographic 
database according to certain specifications and standards, the 
pattern and format have many similarities. Here we develop the 
up-to-down method. 
At the beginning of object match, we must first identify the 
database pattern, and select the corresponding version databases 
and feature semantics to eliminate other irrespective objects to 
confirm the candidate objects in the right dataset. Version 
match depends on the metadata of geographic database which 
include spatial and temporal aspects. Spatial metadata describes 
the geographic range, coordinate reference system, sheet 
numbers of topographic maps, and son on. In china, 
topographic maps are produced and updated in framing map, so 
spatial metadata match can ensure that the datasets and objects 
have the consistent scale, sheet number, and layers. Through 
spatial metadata match we can make the candidate datasets 
having the comparability. Temporal match refers to the 
candidate datasets must have time span to conduce to the 
difference among the temporal datasets. 
The candidate objects must have the same or similar feature 
class before comparison. We call it semantics match in feature 
level. The two candidate datasets must have the same thematic 
content during the match, such as road, resident, hydrology. 
Obviously, the object in one feature can’t change to another 
feature, for example, the road become water. But road elements 
with class 2 may upgrade its class to 1 because they all belong 
to the same feature class and reclassify their grade state. These 
semantics transform can happen on both single object and 
multi-objects. Figure 2 shows the top-to-down match process. 
The common method for object match use geometric match by 
shape contrasts and topological relationships. If the candidate 
objects overlap in geometric shape, we can make the further 
comparison to get the details of change-only information. 
Geometric match can be processed independent of spatial 
database. For line object, there many approach and algorithms 
to deal with these problem, such buffer analysis, hausdorff 
distance (Abbas, 1994), probabilistic statistics and area occupant 
(Vauglin and Bel Hadj Ali, 1998; TONG Xiaohua et al., 2007) 
(Volker Walter and Dieter Fritsch,1999). The author (Ying 
Shen, et al.,2006) presented the matching method of road 
elements in navigation database based on GDF.