ISPRS, Vol.34, Part 2W2, "Dynamic and Muiti-Dimensional GIS”, Bangkok, May 23-25, 2001 
circle we have drawn, we use SQL sentence to query the whole 
layer, so it is inefficient, particular when the map is very large 
Fig. 5 
and the linear road layer contains too many linear objects, the 
query action will take relatively long time. So we have improved 
the LAA, with the using of the buffer of the road object, we can 
save our time. We call it Buffer Linear Abstract Algorithm (BLAA). 
Here are the details of the algorithm. 
Step 1 
If the distance of the vehicle point and one crossroad is less than 
a certain value (100m, for example), then put the corrected point 
directly on the position of the crossroad. 
Step 2 
If there is no crossroad point which will satisfy the condition, then 
find the road object which is nearest to the point. The road 
should locate between two crossroad points. Then create the 
buffer of the road object, the radius of the buffer depends on the 
precision of the GPS receiver, for GN-77 receiver, 100m is OK. 
Step 3 
Determine the relationship between the vehicle point and the 
buffer. If the vehicle is in the range of the buffer, then put the 
objects of the linear road layer which intersect the buffer on 
space position into a temporary table, then do the correcting 
algorithm in this temporary table. Generally, its much smaller 
than the whole linear road layer, this is the key point of efficiency. 
Step 4 
If the vehicle point intersects no buffer, we just determine that the 
vehicle is no longer on the original road. Go to step 1. 
Step 5 
Put the corrected point on the map. Get the next point and go to 
step 1. 
BLAA is the improved algorithm of LAA, which increase the 
efficiency of querying and the ability of processing by the means 
of using the buffer of a road object. The biggest advantage of 
BLAA is that it converts the relationship between the vehicle 
point and the whole road map into the relationship between the 
vehicle point and one of the objects in the road map. What we 
need is to determine the spacial relationship between the vehicle 
point and one buffer, rather than to find a road object in the 
whole road map. Note that it takes time to create a buffer in 
Maplnfo. So if there are not too many objects in a map, BLAA 
may have no advantage against LAA; but if the map is very large 
(i.e. contains many objects), you’d better use BLAA. 
3. SIMULATION RESULTS 
Maplnfo is the most prevalent desktop GIS, and MapBasic is an 
ideal programming language based on the Maplnfo platform. The 
authors realize LAA algorithm by integrating Maplnfo into VB 5.0. 
We use the map of North District of Changsha in the experiment, 
and the GPS receiver is a GN-77 receiver, Fig. 6 shows the 
result of the experiment, the points on the roads are corrected 
points. 
Look at point A, there is a crossroad near it, the distance 
between the crossroad and the vehicle point is less than 100m 
(87.3m in deed). So according to step 1 of LAA, we should put 
the corrected point on the crossroad. Similarly, we put corrected 
points of point B and point C respectively on crossroad nearest 
to them. But at point D, road DE is the nearest to point D. 
However, we can’t decide whether the vehicle is to go straight or 
is to turn left, so according to step 3 of LAA, we ignore this point. 
Point E is a common point. The names of the road nearest to 
point D and point E are of the same, so we put the corrected 
point of E on the line segment nearest to it. The coordinates are 
calculated in step 5 of LAA. Other points like point F are the 
same as point E.