Garth Wales
  • Introduction
  • META
  • To be added elsewhere
  • University
    • 400 Project Research
    • Aims
    • Presentation
    • Report
    • Work Diary
  • Deep Declarative Networks
    • Potential avenues
    • Work Diary
    • Deep Declarative Networks
    • Image Segmentation
  • shortcuts / syntax / setup
    • Jupyter
    • SSH
    • Python
    • CUDA
    • Screen
    • Markdown Syntax
    • Git
    • Dendron
    • Visual Studio Code
    • LaTeX
    • Windows
    • Bash
    • Vim
  • Machine Learning
    • Markov decision process
    • Reinforcement learning
    • Deep Q Network
    • Modern hopfield network
    • Object Detection
    • Face detection
    • Activation Functions
    • Gym
    • RL frameworks
    • Hyperparameters
    • Interpretability
  • Reinforcement learning
    • Memory
    • Exploration techniques
    • Agents
  • garth.wales
    • Website redesign
  • Computer Graphics and Vision
    • Image Mosaicing
    • Stereo Vision
    • Rendering
    • Exam papers
  • Neural Networks
    • Introduction
    • Learning
    • Learning theory
    • Stochastic Neural Networks
    • Interpretability
    • Word Respresentation
    • Language modelling with recurrent networks
    • Sequence-to-sequence networks
    • Exam
Powered by GitBook
On this page
  • Botteril et al (Real-Time Aerial Image Mosaicing)
  • Mills (4 and 7 point algorithms)
  • Broxton et al (Immersive Light Field Video)

Was this helpful?

  1. Computer Graphics and Vision

Exam papers

The exam was based partly on questions relating to these papers

Botteril et al (Real-Time Aerial Image Mosaicing)

Fast instead of sift blobs for SPEED

  • Using corners means no invariance to most things, but the drone flies in a straight line continuously so might be less important.

  • Using corners doesn't have as good

Mills (4 and 7 point algorithms)

Biggest disadvantage

  • SLAM sorta thing, it doesn't do well if no rotation or side to side motion leading to high reprojection error (for the 4 point algorithm)

Broxton et al (Immersive Light Field Video)

Difference between lightfields and multi-plane images

  • Multi-plane changed to a sphere representation allowing you to look around more. IMMERSIVE! More DoF for viewing.

PreviousRenderingNextIntroduction

Last updated 3 years ago

Was this helpful?