Abstract
In this paper, we propose a real time 3D imaging system, Existing techniques for construction of 3D models require multiple cameras and complex calculations. In our proposed system, the 3D shape is captured by integrating multiple range images obtained simultaneously from different directions. It can accomplish the real time processing by generating the range images quickly and continuously. Moreover, the obtained 3D shape is treated as "Point Cloud", and therefore, the connectivity information of vertices and the calculation of connection topology become unnecessary. Giving importance to transmission efficiency is the advantage of our system. The obtained 3D shape is transmitted to the remote location, and it is reconstructed and depicted immediately. By repeating this procedure continuously, a 3D sequence can be generated in real time. The 3D shape is constructed by integration of multiple range images, causing overlap and annoying artifacts between adjacent images. We apply a cylindrical method to each frame of the 3D sequence to get rid of the artifact and improve the image quality. Our algorithm can remove the invalid points quickly and efficiently, and can be implemented in real time.
| Original language | English |
|---|---|
| Pages (from-to) | 57-64 |
| Number of pages | 8 |
| Journal | WSEAS Transactions on Electronics |
| Volume | 5 |
| Issue number | 3 |
| Publication status | Published - 2008 Mar |
Keywords
- Point cloud
- Real time systems
- Stereo vision
ASJC Scopus subject areas
- Electrical and Electronic Engineering