| abstract |
The point based 3D models represent a viable alternative to the mesh based 3D models. As in the case of other media types, the huge size and high redundancy in the data make specific data compression a must. A framework for progressive compression of point-based models is described in [Waschbsch et al. 2004]. It is based on the multiresolution decomposition of the point set, allowing for progressive decoding and removing the need for additional resampling. Specific coding operations can be defined for each attribute of the model, allowing further extensions and making the method general. Coding operations are defined for position, normal and color. The fact that the points lay on a surface is exploited in the position coding. We will investigate possible modifications of decomposition and attribute coding methods that will allow a better prediction error, thus a higher compression rate.
|
