Home  >  Three-dimensional Object Reconstruction

Funded by:
RGC Competitive Earmarked Research Grant (CERG), Hong Kong [Project 9040540]; and
Strategic Research Grant, CityU [Project 7001431].

Contents:

The Structured-Light System

The Light Encoding method

Examples of Object Reconstruction

For efficient reconstruction of entire object/environment: "next viewpoint decision"

Download the software and data examples


Introduction of the Structured Light System

The system contains a color CCD camera and a DLP projector. For more details, please read my published papers of this issue.

 


Light Encoding Method

About 7 light encoding methods are implemented in the 3-D measurement system. Typical encoding methods are "Gray-Coded Light Approach" and "Uniquely Encoded Light Patterns", as shown in following figures.

   

For more details, please read my published papers of this issue. 


Examples of Object Reconstruction

Example one

The first example is to show 3D reconstruction of a head model.

 
The object (head model) to to reconstructed

 

  Description Current View Merged All Previous View
Step 1 Obtain the first view

- Top view of the head

(no previous knowledge about the object)
Step 2 Obtain the second view - The left side the head.

And merge the two views (all known knowledge).

 

Step 3 Obtain the third view - The front side the head.

And merge the three views.

Step 4 Obtain the forth view - The back side the head.

And merge the four views.

 

Step 5 Obtain the final view - The right side the head.

And merge all the five views to bring an entire model of the head.

 

Example two
Here is another example to show 3D reconstruction of a cup within 4 steps.

 
The object (a cup) to to reconstructed

 

  Description Current View Merged All Previous View
Step 1 Obtain the first view

- Top view

(no previous knowledge about the object)
Step 2 Obtain the second view - one side of the cup.

And merge the two views (all known knowledge).

 

Step 3 Obtain the third view - Another side of the cup.

And merge the three views.

Step 4 Obtain the final view - A side of the cup.

And merge all the four views to bring an entire model of the cup.


 


Next Viewpoint Decision

The next viewpoint decision is based on a method called "trend surface".

 For more details, please read my published papers of this issue.


Download

Software: The 3D measurement system, Active3D.zip
          Note: The current version only support frame-grabber produced by Imaging Technology. Only with such a card, you can perform 3D measurement.

Software: The 3D viewer (with animation), view3d.zip.
          Note: It was developed by OpenGL. If there are problems of launching the program, see here to fix the necessary supports.

Software: The projector controller (pattern generator), dlp.zip.
          Note: To generate many light patterns and output to the DLP/LCD projector. It also runs as a server to receive the pattern-commands from the 3D measurement system.

3D data set: Example data of the head.
          Note: The *.xyz files can only be readable by the two programs here. To get other formats of the data, please contact the author directly.

 


For more information about this research, contact ShengYong Chen. Email: msg@sychen.com

Chen, ShengYong, Stephen, 2003. All rights reserved.


Go back to ShengYong Chen's home page or go back to your previous page.


Page visits: since January 1, 2000. Last updated: August 1, 2012.  sychen.com all rights reserved.