Award details

Image analysis for robot control

Reference	BBS/E/S/00000086
Principal Investigator / Supervisor	Dr Robin Tillett
Co-Investigators / Co-Supervisors
Institution	Silsoe Research Institute
Department	Silsoe Research Institute Department
Funding type	Research
Value (£)	145,869
Status	Completed
Type	Institute Project
Start date	01/04/1997
End date	31/03/1998
Duration	12 months

Abstract

Automatic image analysis can be a powerful non-contact sensor for machine guidance or control and for inspection and grading. However, when dealing with natural objects, techniques are required which can handle their inherent variability and make appropriate decisions within an unstructured or only partially structured environment. In this project generic techniques have been sought through various application areas. Techniques for detecting structure or shape features in images of plants and combining them to interpret images have been developed, to include tracking and linking branch structures, resolving conflicting interpretations of structure and using grey level shading to determine surface shape. Techniques for the intelligent detection and classification of blemishes in produce have been developed, using neural networks. Segmentation techniques were developed to recognise typical patterns. Ways of interpreting images using multiple sources of uncertain information are being sought and algorithm performance and interactions are being assessed. Methods of combining information will be developed initially using Bayesian networks. In the current year work continues on the use of image analysis for guiding robotics for plant-scale husbandry.

Summary

unavailable

Committee	Closed Committee - Agri-food (AF)
Research Topics	X – not assigned to a current Research Topic
Research Priority	X – Research Priority information not available
Research Initiative	X - not in an Initiative
Funding Scheme	X – not Funded via a specific Funding Scheme

I accept the terms and conditions of use (opens in new window)

export PDF file

back to list new search