[
Date Prev][
Date Next][
Thread Prev][
Thread Next][
Date Index][
Thread Index]
[
List Home]
[udig-devel] Consistant GIS image files w/ geo-referencing
|
I'm doing a bit of an odd thing here, so bear with me!
I'm using uDig plus GIS information researchers have created for Venice
Italy. The entire data set is rather large:
http://www.backspaces.net/files/uDigPix/uDig001.png
The focus of our work is much smaller:
http://www.backspaces.net/files/uDigPix/uDig002.png
There are 22 markers (the black squares) which are researcher's observation
points for boat traffic within the canals. (We consider three canal types:
the larger lagoon, the Grand Canal, and the small residential canals we call
tributaries)
We then capture this data as an image, and insert it into an Agent Based
Modeling (ABM) system (NetLogo):
http://www.backspaces.net/files/uDigPix/NetLogo315.png
We do considerable processing of the data, and add two types of agents:
boats (black arrow shapes) which travel on the canals, and observers (the
red circles) who work at specific "stations". They record boat movement
such as from-to directions, boat type (water taxi, water bus, utility boats,
etc), and the time the event is recorded.
The goal is to help the city of Venice to make a "sim-venice" urban studies
software suite combining GIS, ABM, sophisticated scientific visualization,
and "what if" studies: what if canals are blocked for repair, what if we
could reduce their wake to reduce canal erosion, what if we add new boat
transportation types, and so on. Think a version of Venice "in-silico" for
general scientific urban studies.
So finally here are the problems I need to solve.
First of all, we need to export to NetLogo a 1050x600 image that is
consistantly the same, pixel for pixel. We may modify the uDig layers in a
variety of ways, but we want the same geographic extent and scale every time
we create a new image. We currently use screen capture tools like this:
http://www.backspaces.net/files/uDigPix/xScope004.png
.. which do a pretty good job of letting get an image, but pretty hard to
get consistent images when making new ones. (We can be slightly imprecise
in the output image .. +/- 10 or so pixels can be adjusted within NetLogo
easily enough. But it should be consistent so that if we make more images,
they overlay exactly the initial ones.)
So question number one is: Is there a way for me to export and image of a
GIS area in uDig and get a consistent image of it (same pixel size/area,
same uDig GIS content (i.e. same exact scale and same position, upper left
corner or some similar reference)? The screen-grab approach seems pretty
extream! :)
By now you must be thinking we're pretty weird, but wait .. there's more!
After we do a run of the ABM, we import it into Blender, a 3D modeling
system, so that we can get much more visually compelling movies. So this
means the image above must not only be consistent in terms pixel placement
and area and so on, it must be geo-registered in some way. As I understand
it, there is the notion of a "world file" in GIS which is a registration
file accompanying an image saying what the geo-coordinate of the upper left
corner is, and what the pixel size is in world coordinates (i.e. how many
meters on a side a pixel is).
I've tried making a layer of two markers that would be the desired extent,
using the uDig zoom-to-layer feature, and entering the scale as a consistent
value, but this does not appear to work. I get inconsistent zooms and
entering a scale changes whenever I enter it to a similar scale but not the
same scale. I realize its asking uDig a lot to be pixel-perfect!
Whew! So is there a way for me to achieve this sort of repeatable export of
a map area with a given pixel height and width, and with the geo-coordinate
information so that it can be imported into other programs in a standard
way?? How would you try to do this?
I bet this is the weirdest use of uDig yet!
Owen
--
View this message in context: http://www.nabble.com/Consistant-GIS-image-files-w--geo-referencing-tf4664125.html#a13323771
Sent from the udig-devel mailing list archive at Nabble.com.