next up previous contents
Next: Getting started Up: Object Recognition Previous: Object Recognition   Contents

Introduction and Aim

This session will cover a slightly more advanced machine vision topic than grey level image analysis, object recognition. The TINA system will be used to match polygonal boundaries ( extracted from images by the Canny edge detector) using the technique of Pairwise Geometric Histograms, as described in the lectures. Here a theory is suggested that the use of a complete description of data with the desired levels of invariance, matched with an appropriate statistical method, forms an optimal solution to the recognition of arbitrary objects. In this practical we will attempt to evaluate the performance of the recognition system subject to changes in the shape representation.

root 2018-12-11