Show categories
Show categories
This is a transcription of my PhD thesis to HTML. It's an excuse to get all the programs and utilities I was playing with at college up and running again on my Mac. Given that back then I had a 33 MHz R3000 machine (and I thought I was lucky to have that!), and that I have a 4-core 3GHz machine under my desk, I'm hoping some of the obstacles I hit back then can be overcome just by pure brute force...


Contents

Anyway, to start at the best place... the beginning.... The first problem was image recognition, where parts of images were to be recognised as individual objects
  1. Introduction
  2. Segmentation
  3. Preprocessing
  4. Feature extraction
  5. The probabilistic approach
  6. The neural-network inference engine
  7. Results



The other part of my PhD was in optic-flow-relaxation for motion estimation and hence reliable registration of image regions from frame to frame.
  1. Introduction
  2. Segmentation
  3. Flow theory
  4. Flow-relaxation theory
  5. Description of the flow algorithm
  6. Results (Waiting for re-implementation of software on the Mac)
    • Simple test data
    • Yosemite data
    • Bridge (set 1) data
    • Bridge (set 2) data
    • Filton data


Read on, Macbeth...
Show categories