next up previous
Next: Assumptions Up: Line Labelling Previous: Line Labelling

Introduction

Line Labelling is applicable when two-dimensional lines have been extracted from a two-dimensional image.

Initially, we shall restrict ourselves to polyhedral objects in our scene, i.e. have flat faces bounded by straight edges.

Under this assumption it becomes possible to interpret the edges

Consider the line drawing shown in Fig. 41.

 

Fig. 41 Line drawing of two cubes

It can be seen that

An important observation is that

Sometimes both of these faces can be seen from a given viewpoint (as in the case of line a in Fig. 41).

If only one of the faces is visible -- the edge is called an occluding edge. On the other side of an occluding edge, the background or a non-touching face is visible.

The basis of the line labelling method is to attempt to classify each edge in the image as being one of:

Edges of each type are shown in Fig. 42.

 

Fig. 42 Types of edge in a line drawing

Where edges meet only certain possible interpretations of the meeting lines are mutually consistent.

This reduces the number of possibilities for the classification of each edge.

If a line corresponds to an occluding edge:


next up previous
Next: Assumptions Up: Line Labelling Previous: Line Labelling

dave@cs.cf.ac.uk