RE: [Maptitude] Rubbersheeting with many control points

Mon, 30 May 2005 07:48:02 -0700

Reinaldo
 
I think that the issue if whther you can use 12000 control points is not much related to Maptitude, but to the general process of rubersheeting itself.
 
In general, the positional accuracy should increase as you use more cpontrol points, up to a limit where you start to face a diminishing returns effect.  Beyond that threshold, adding more control points would not improve your results, and in some cases more control points may even make things worse.  Some research suggests that that threshodl is a function of the lines-to-nodes-ratio in your database.
 
Take a look at this paper I got bookmarked some time ago: Quantifying coordinate improvement of vector GIS data using rubbersheeting methods   (
http://www.findarticles.com/p/articles/mi_qa4039/is_200303/ai_n9229943/pg_4
)
 
It is an interesting experiment on a case that may resemble yours.
 
Good Luck
 
Armando

-------- Original Message --------
Subject: [Maptitude] Rubbersheeting with many control points
From: Reinaldo Paul P�rez Machado <[EMAIL PROTECTED]>
Date: Mon, May 30, 2005 8:38 am
To: Maptitude Discussion List <[email protected]>

Hi:

I have a situation here that I am solving via Rubbersheeting. There are two different line maps that cover about the same area, with a different number of records in both cases. As the restitution that gave birth to both of them was not the same (the projection was), the maps are quite similar but not identical, even the lines (street central lines) that point to the same object (i.e. Street A) are not on the same position even though the proportions and angles are correct. The mentioned behavior is not constant over the territory, in some parts the two maps are almost identical, in other parts the swift is about 10 m NE, 14 m NE or 5 m S bound, there is a lot of intermediate values and directions in between. About 94 % of the lines are present in both geographic files, some are missing in Map 1 and present in Map 2 and vice versa, due to the different creation date. I have made a grid of 3x3 km to cover the territory where the Rubbersheeting procedure would be applied, and have a rough calculation of about 8,000 to 12,000 control points, and we have created about 800 by now. As this is a very painful job, and I do not want to risk the success of the whole operation just in the end, my concerns are:
1) Is there a limit for Maptitude to create and process 12 thousand control points?
2) Is there a way to create two or more separate control points files and then join them before doing a single Rubbersheeting operation? The (*.crd) files where the control points of Maptitude are stored do not seem to be in any known editable format.
3) Even if questions 1 and 2 would have an affirmative answer. Is there any alternative solution to this situation? I mean,  splitting the map in 4 pieces and then joining together after the Rubbersheeting operation, but how to control the connectivity along the cutting edges?

Thanks in advance for your aid or suggestions. Regards,

Reinaldo



Yahoo! Groups Links


Yahoo! Groups Links

Reply via email to