LoTEC
Research Project LoTEC
Subject
Development of a chip set generating location and orientation tolerant edge clouds in video real time
Abstract
Edge representations offer elegant descriptions of forms. Yet, often those representations are very sensitive to small shifts. The introduction of location and orientation tolerant edge representations leads to a description of contours as clouds of oriented edge elements which enables position invariant recognition. The edge clouds of a learnt and a currently presented object allow a match even in the case of small lateral and diagonal shifts caused by fixation errors. This increases the robustness of the recognition process.
In the LoTEC project, we developed and tested a chip set which generates location and orientation invariant edge clouds of objects in video real time and which is intended to operate in a robot based image processing system.
Status
Completed 7/1997, duration 3.5 years, IMA Lab, University of Hamburg
Partners
Financing
Deutsche Forschungsgemeinschaft