EPSRC logo

Details of Grant 

EPSRC Reference: GR/G53255/01
Title: A STRATEGY FOR VISUAL RECOGNITION OF 3D OBJECTS FROM A LARGE MODEL BASE
Principal Investigator: Kittler, Professor J
Other Investigators:
Researcher Co-Investigators:
Project Partners:
Department: Sch of Electronics & Physical Sciences
Organisation: University of Surrey
Scheme: Standard Research (Pre-FEC)
Starts: 01 May 1992 Ends: 30 June 1995 Value (£): 80,194
EPSRC Research Topic Classifications:
Image & Vision Computing
EPSRC Industrial Sector Classifications:
Related Grants:
Panel History:  
Summary on Grant Application Form
The aim of the project is to perform 3D object recognition from 2D images when a large number of models is involved. This requires the ability to index efficiently into a large model database.Progress:Progress has been made in two main aspects of the task, (1) derivation of database index on the basis of their 3D shapes and (2) a computational solution to the extraction of shape descriptions from 2D images. A geon based approach has been adopted for shape description.On the first front, investigation has produced a theory for generic object classification based on 3D primitive shapes. This theory stipulates that object classification is the outcome of an evolution process where both object functionality and shape are taken into account. Using such a scheme, the object identity in terms of generic class is produced. In turn this classification yields database indices. The proposed generic classification-based indexing employs a reasoning system where both functional and geometric shape rules are used for object classification. This approach is highly robust in comparison with those relying purely on shape recovery from images.We have also developed a direct indexing approach for object classification. With this approach a direct link between an object identity and its shape description is established through supervised training. The performance of such a strategy depends on the training examples. The advantage of this approach is the low computational cost. We have developed a computational method for extracting shape description (in terms of geons) from real image data. The method uses convex polygon as the key perceptual group to facilitate grouping of features. The grouped features are then used to generate hypotheses of geon which are subsequently verified. The method is capable of handling occluded geons. Experiments have revealed that this method performs significantly better than existing geon extraction methods.
Key Findings
This information can now be found on Gateway to Research (GtR) http://gtr.rcuk.ac.uk
Potential use in non-academic contexts
This information can now be found on Gateway to Research (GtR) http://gtr.rcuk.ac.uk
Impacts
Description This information can now be found on Gateway to Research (GtR) http://gtr.rcuk.ac.uk
Summary
Date Materialised
Sectors submitted by the Researcher
This information can now be found on Gateway to Research (GtR) http://gtr.rcuk.ac.uk
Project URL:  
Further Information:  
Organisation Website: http://www.surrey.ac.uk