ERNEST: A Semantic Network System for Pattern Understanding
IEEE Transactions on Pattern Analysis and Machine Intelligence
Knowledge-based image understanding systems: a survey
Computer Vision and Image Understanding
Playing Domino: A Case Study for an Active Vision System
ICVS '99 Proceedings of the First International Conference on Computer Vision Systems
Multiple View Geometry in Computer Vision
Multiple View Geometry in Computer Vision
A Multiphase Dynamic Labeling Model for Variational Recognition-driven Image Segmentation
International Journal of Computer Vision
Multi-Aspect Detection of Articulated Objects
CVPR '06 Proceedings of the 2006 IEEE Computer Society Conference on Computer Vision and Pattern Recognition - Volume 2
High-Level Expectations for Low-Level Image Processing
KI '08 Proceedings of the 31st annual German conference on Advances in Artificial Intelligence
Pattern Recognition, Fourth Edition
Pattern Recognition, Fourth Edition
Metamodels Taken Seriously: The TGraph Approach
CSMR '08 Proceedings of the 2008 12th European Conference on Software Maintenance and Reengineering
Interaction of control and knowledge in a structural recognition system
KI'09 Proceedings of the 32nd annual German conference on Advances in artificial intelligence
Modularizing spatial ontologies for assisted living systems
KSEM'10 Proceedings of the 4th international conference on Knowledge science, engineering and management
Model-based recognition of domino tiles using TGraphs
Proceedings of the 32nd DAGM conference on Pattern recognition
Hi-index | 0.00 |
We report on a case study showing on recognition of objects under perspective distortion in projected 2d images. We use symbolic descriptions and yield similar results as heuristic or statistical methods. The knowledge is modeled in so-called TGraphs which are typed, attributed, and ordered directed graphs. We combine the search in the state space with a maximum weight bipartite graph-matching and in consequence we reduce the numerous amount of hypotheses. Furthermore we use hash tables to increase the runtime efficiency. As a result we reduce the runtime up to a factor of five in comparison to the system without hash tables and achieve a detection rate of 90.6% for a data set containing 968 perspective images of poker cards and domino tiles. Therefore, we show that model-based object recognition using symbolic descriptions is on a competitive basis.