Constructive solid geometry for polyhedral objects
SIGGRAPH '86 Proceedings of the 13th annual conference on Computer graphics and interactive techniques
An introduction to splines for use in computer graphics & geometric modeling
An introduction to splines for use in computer graphics & geometric modeling
Polygonization of implicit surfaces
Computer Aided Geometric Design
Computer graphics: principles and practice (2nd ed.)
Computer graphics: principles and practice (2nd ed.)
Advanced animation and rendering techniques
Advanced animation and rendering techniques
Pattern Recognition Letters
Target tracking by matching a shape represented by a tree of sigmoid functions
Pattern Recognition Letters
Neural Networks: A Comprehensive Foundation
Neural Networks: A Comprehensive Foundation
Winged edge polyhedron representation.
Winged edge polyhedron representation.
| Hi-index | 0.00 |
In this paper, an experiment is conducted in order to prove that multilayer feed forward neural networks are capable of representing most classes of 3D objects, used in computer graphics. Furthermore, the use of neural networks for the representation of light sources is introduced. Finally, networks of lights and objects are combined with a simple illumination model achieving rendered images with shadows. One network is used per one volumetric description of a 3D object. Objects that have a simple analytical form are represented by specifying the networks' parameters manually. For objects with more complicated shapes we generate training examples that consist of points on the objects' surface and points lying on inclosed and enclosing surfaces. The algorithm for generating the training data, is a simple heuristic that uses the surface normal to determine whether a point in the vicinity of the surface belongs to the inside or the outside of the object. The network is trained on the generated examples, using the back propagation technique. The experimental results prove that this representation method is accurate and compact. Feed forward neural networks being hardware implementable offer the ability for a faster representation and rendering.