Eonfusion forum Home

 Search       Members   File sharing   Help   Home 
Search by username
Not logged in - Login | Register 
Eonfusion forum > Advanced topics > Data fusion > Fuse and Intersect Vector Set
-> Eonfusion web site
-> Eonfusion manual

Fuse and Intersect Vector Set
 Moderated by: forumadmin  
 New Topic   Reply   Print 
AuthorPost
Francis Chui
Internal


Joined: Mon Aug 11th, 2008
Location: Hobart, Australia
Posts: 11
Status:  Offline
Mana: 
 Posted: Wed Jul 1st, 2009 06:02 am
 Quote  Reply 
Fuse and Intersect Vector Sets

What is the difference between Intersect Vector Sets and Fuse Vector Sets?  The short answer is that the Intersect Vector Sets operator does an extra step after fusion to retain coincident data on the specified fuse space.  The fuse space in both operators is based on the specified axes; this can be in one or two dimensions.  The resulting fused dataset is the intersection of multiple datasets in these axes.  One fundamental point is that both operators will only show coincident vertices once in this fused dataset.  The fact that coincident vertices only occur once can cause problems, particularly if your data overlaps itself in this coordinate space.  This is something that occurs frequently and we will demonstrate in this article.


Figure 1 Generated track and grid of a Surface to be used for draping.


An example where one would use one of the fusion operators is when draping a track over topography – you have GPS data and wish to overlay this on top of some topology.  To fuse these two datasets you would specify spatial x and y for the axes.

Take animal tracking data as an example: An animal can cross a spatial location on multiple occasions over a period of time.  So when the fused dataset is used to create a draped track, the coincident vertices can cause ambiguity when attempting to interpolation of time.


Figure 2 Fused dataset with new vertices shown in red, track vertices in green and surface vertices in blue.


This occurs because we are attempting to interpolate time for a vertex using its spatial location as a reference back to the source track data but there are multiple vertices with this position at different times.



Figure 3 Fused dataset illustrating the time interpolation from the original track


Since there is only one vertex in the fused dataset, only the first value encountered in the interpolation process along the track is used – any subsequent time values will not be used since the interpolation process only occurs for vertices which don’t have a value for the specified attribute and the vertex already has a value.  This is shown in Figure 3 where the fused set has the vertex at spatial x/y labelled B occurring twice on the track.  Interpolation from D –> B and B –> E should be t3 –> t4 and t4 –> t5 respectively but instead has become t3 –> t1 and t1 –> t5 resulting in the strange time coloration in Figure 4.



Figure 4 Track times interpolated onto the fused dataset.  The coincident vertex being fused into one causes incorrect time interpolation as shown along the line.


Another solution is to copy the elevations from the fused set back to the track and get the correct times.  The results are not ideal as the original track will not have the extra vertices needed to depict the elevations.  So if the track's resolution is significantly lower than the surface's, it can cause visualization of the track to go below and above the surface as shown in Figure 5.



Figure 5 Visualization of an incorrectly draped track on surface.  That is, the new vertices weren’t included.


Thus the best solution is to use the Intersect vector set operator, which (in addition to creating the intersection of the source datasets in the fused dataset) inserts new vertices in each source dataset where intersections occur.  So in our example, extra vertices were inserted between every tracked spatial coordinate for any intersections with the topological surface.


Figure 6 Using the intersected source track solves the issue of coincident vertices and thus allows for correct interpolation (spatial in this case).


While it is possible to use intersect vector sets in all situations, it may not be the best choice since the intersection process is an extra step performed on top of the fusion process.  Things like the fusing of topology and bathymetry shouldn’t need intersection but animal and fish tracking in particular are bound to have overlapping vertices in the coordinate space you fuse on.


Figure 7 The dataflow used to create the examples shown above.


Hopefully you have found this article useful.  You can find the dataflow (Figure 7) used to create the examples shown above
here.  If you need more information about this topic or have suggestions for another topic, please drop us an email at support@eonfusion.com.

For more general information about fusion and intersect vector set operators please refer to the manual pages:

Last edited on Wed Jul 1st, 2009 06:13 am by Francis Chui


 Current time is 11:27 pm




Powered by WowBB 1.7 - Copyright © 2003-2006 Aycan Gulez

Terms and Conditions