Home

Basic Contents

Computer Vision ... (Published 1992)

Simulated Human Vision..... Ian Overington

Location: Eastbourne. UK
ianoverington@simulatedvision.co.uk ............ www.simulatedvision.co.uk

12.5.2. Practical verification.
12.5.2.1. Ideal profiles.
12.5.2.2. Practical figures.
12.5.2.2.1. Noise-free square.
12.5.2.2.2. Noisy image of square.
12.5.2.2.3. Dodecagon.
12.5.2.2.4. Star.
12.5.2.2.5. Typical scene from the selection of ‘car exemplar’ images.
12.5.3. Progressive synthesis.
12.6. Conclusions.

Appendix 12A. Progressive smoothing by recursive filtering on a 3 X 3 neighbourhood.
Appendix 12B. Profile synthesis and chain coding by recursive filtering.

Chapter 13. Absolute motion and stereo disparity of corners and similar features.

13.1. INTRODUCTION.

In Chapter 5 it was shown how it is possible to measure components of optical flow orthogonal to
fragments of edges directly from illuminance gradient data in pairs of frames taken from temporal
image sequences. In Chapter 9 it was demonstrated how, starting with an array of such fragmentary
flow data, simple window analysis can then provide a variety of true flow data, both for local rigid
bodies and for the whole scene. Yet again, in Chapter 10 it has been explained how a directly
similar approach may be used on stereo pairs of images, in order to extract a wide variety of stereo
disparity data. However, whilst these methods of analysis are both simple and powerful, one aspect
of optical flow or stereo disparity which they cannot determine is the true local disparity of specific

Continued