/Computer-Vision

Exercises for the Computer Vision class at ETH

Primary LanguageMATLAB

Computer-Vision

Exercises for 263-5902-00L Computer Vision at ETH HS2019

1 Camera Calibration

  • Data normalization
  • Direct Linear Transform
  • Gold Standard Algorithm

2 Feature Extraction

  • Image gradients
  • Local auto-correlation matrix
  • Harris response function
  • SSD one-way nearest neighbors matching
  • Mutual nearest neighbors and ratio test

3 Particle Filter

  • Monte Carlo Localization
  • Markovian Localization problem using probability distribution
  • Resampling of particles according their a posteriori proability weight after having a observation

4 Model Fitting

  • RANSAC algorithm for line fitting
  • Fundamental matrix estimation
  • Eight-Point algorithm
  • Feature extraction and matching
  • adaptive version of RANSAC

5 Image Segmentation

  • Lab color space
  • Mean-shift Segmentation
  • EM Segmentation

6 Stereo Matching

  • Rectification
  • Disparity computation using winner-takes-all stereo and SSD or SAD
  • Graph cut
  • Generation and visualization of textured 3D model using MeshLab

7 Structure from Motion

  • Combination of all learned methods so far
  • Feature extraction
  • Fundamental matrix estimation using RANSAC and 8-point algorithm
  • Triangulation
  • Projection matrix estimation using RANSAC and DLT
  • Visualization of camera positions and matched points

8 Shape Context

  • Shape context descriptors
  • Cost matrix computation
  • Hungarian algorithm to find best matches
  • Minimum distance sampling
  • Thin plate spine