ELCOVISION ELSP Auto is the optional full automatic rectification module of ELCOVISION 10. It generates textures from oriented pictures onto a three-dimensional surface model.
It can use an existing surface model or it can calculate a surface from cloud of points, linedrawings or digital terrain models using its built in state-of-the-art full 3D surface modeller.
The surface model will be the base for all further processing. The program will process every surface in the drawing individually by picking each of the individual surfaces out of the model, checking which picture will be the best source for a rectification plane, creates an rectification plane on this picture, cuts out the valid plane range and transfers the digital rectified texture into the drawing.
See the following two examples for a better explanation.
First Sample: 3D Measurement And Visualisation Of a Whole Mountain
Step #1: Orientate The Pictures
The first step is in common with the traditional 3D photogrammetry. You need to do the orientation process with the pictures. You can use as many pictures as your license of ELCOVISION 10 permits you to use.
In this example a skiing pist was photographed from the opposite slope. In total there were 12 pictures used for this example. After finishing the orientation process the project was loaded into the AutoCAD environment.
There was also a contour map of the slope available, we used it as a base for the surface generation
With the built in surface modeller of ELCOVISION it is very easy to convert this contour map into a surface model:
Step #2: Triangulate The Surface
This simple dialog is used to configure the built in surface modeller of ELCOVISION 10.
You select all of the drawing elements which are supposed to define the resulting surface and determine the "quality" of the points. At last define which kind of result you need: a closed surface or an open surface
After the triangulation is done you can simplify the result for your needs. You can adjust the triangulation mesh from fine to wide nearly seamlessly using a slider control: This leads to savings in CPU time and memory usage during the rectification process.
Fine meshed triangulation:
By using
the slider control on the right side of the window you
can choose the dense of the triangle mesh in real-time
and nearly seamlessly.
Wide meshed triangulation:
The adjusted mesh is transferred into the AutoCAD drawing simply by clicking [OK] button
Now everything is ready for texture generation.
Step #3: Calculate The Textures
Now select all drawing entities by using all snap and select modes of AutoCAD. Also select all of the pictures which are supposed to be valid texture sources.
After clicking [OK] the texturing process will start and after a view minutes you will get the following result:
Second Example: A 3D Orthophoto from a curved ceiling
The second example shows a typical task for protection of historical monuments. A dilapidated, monument protected hotel dining hall shall be documented and plans prepared for the restoration. In this example 116 pictures were used. Nearly all of them are close-ups to make smallest details visible and measurable. After the orientation process you can see the generated orientation points and the photos point of view using the ELCOVISON 10 project explorer:
There were enough objects points for the surface triangulation generated during the orientation process so we simply load them into AutoCAD and triangulate them using the ELCOVISION 10 surface
modeller :
After this ELCOVSION 10 can generate and map all textures onto this surface automatically:
This 3D texture now represents the basis for far detailed plans. Or you can do a projection of the entire ceiling:
In order to show you the accuracy in which ELCOVISION 10 proceeds, you can see the same detail twice below. In the first image you can see the superimposed borders of the original rectification planes:
Templates for the restoration of the painting can be produced by simply tracing the outlines which is very fast and simple
Technical Information About The Projects:
Mountain
Dining
Room
Number of images
12
116
Used cameras
Fuji S2
Fuji S2, Leica R5 Elcovision
Storage needed for the images
427 MB
3.9 GB
Accuracy at the object
Better than 30 cm
Better than 2 mm
Number of textures generated
6425
1459
Kind of textures
True colour images (24 Bit)
256 colour images (8 Bit)*
Size of textures
100 MB
752 MB
CPU time for texture generation**
20 min
62min.
* 256 colours textures were used, since thereby the storage requirement sinks around 2/3. With true colour textures approx. 1,6 GB texture data would result.
