A Self-Organizing Multiple-View Representation of Three-Dimensional Objects
Abstract
We explore representation of 3D objects in which several distinct 2D views are stored for each object. We demonstrate the ability of a two-layer network of threshold summation units to support such representations. Using unsupervised Hebbian relaxation, we trained the network to recognize ten objects from different viewpoints. The training process led to the emergence of compact representations of the specific input views. When tested on novel views of the same objects, the network exhibited a substantial generalization capability. In simulated psychophysical experiments, the network's behaviour was qualitatively similar to that of human subjects.
Document Details
- Document Type
- Technical Report
- Publication Date
- Aug 01, 1989
- Accession Number
- ADA216711
Entities
People
- Daphna Weinshall
- Shimon Edelman
Organizations
- Massachusetts Institute of Technology