Generic computer vision methods

A part of the CVonline computer vision resource listing methods that are used widely across cmputer vision and image processing.

1. General image segmentation methods

   1. Clustering methods

      1. DBSCAN - Density-based spatial clustering of applications with noise

   2. Compression-based methods

   3. Histogram-based methods

   4. Region-growing methods

   5. Split-and-merge methods

   6. Partial differential equation-based methods

   7. Graph partitioning methods

   8. Multi-scale segmentation

   9. Semi-automatic segmentation

   10. Trainable segmentation

   11. Segmentation benchmarking

2. Accumulation/voting methods

• 1. Hough transform

     1. Adaptive Hough Transform

     2. Hough transform of curves

     3. Cascaded Hough Transform

     4. Generalised Hough transform

     5. Hierarchical Hough Transform

     6. Maximum margin Hough transform

     7. Probabilistic Hough Transform

     8. Randomised Hough Transform

     9. Surface finding

  2. Tensor Voting

3. Diffusion/PDE/Time based evolution methods

• 1. Heat kernel -- see also scale-space which is based on linear diffusion and the Gaussian (heat) kernel

4. Eigendecompositions

5. Genetic algorithms/Genetic programming

6. Graph Methods

• 1. Graph representations

     1. Adjacency graph

     2. Association graph

     3. Attributed Graph

     4. Dynamic Feature Graph

     5. Graph embedding

     6. Hierarchical graph/Hypergraph representations

     7. Laplacian smoothing

     8. Median graph

     9. Optimal Basis Graphs

     10. Probabilistic graphical model, Probabilistic graph theory

  2. Graph matching

• • 1. Bayesian Graph Matching

    2. Bipartite matching

    3. Graph cuts

    4. Graph kernel methods

    5. Graph edit distance

    6. Maximal cliques in Association graphs

    7. Spectral decomposition methods

    8. Subgraph isomorphism problem

• 3. Multidimensional scaling

7. Image pyramids and scale reduction

• 1. Adaptive Pyramids

  2. Gaussian pyramids

  3. Laplacian pyramids

8. Level sets

• 1. Level set trees

9. Matching methods

• 1. Hungarian algorithm

10. Minimum description length

11. Model Selection Criteria

• 1. Akaike Information Criterion

  2. Bayesian Information Criterion

12. Multiple Scales/Resolutions

• 1. Multiple-scale analysis

     1. Multi-Scale Integration

  2. Fractals

  3. Ranklets

  4. Scale space

  5. Wavelets

• • 1. Noiselets

13. Graph, networks and connectionist methods

• 1. Bayesian networks

  2. Connectionist methods

  3. Gaussian processes methods

  4. Neural networks

  5. Probabilistic graphical models

     1. Expectation propagation

     2. Belief propagation

     3. Message passing

        1. Variational message passing

        2. Tree reweighted message passing

  6. Radial basis function networks

  7. Wavelet Networks

14. Regularization

15. Relaxation

• 1. Continuous

  2. Discrete

  3. Probabilistic/Stochastic

  4. Linear programming relaxation

  5. Lagrangian relaxation

16. Spatial indexing/hashing

17. Subpixel Methods

18. Super-resolution

19. Certainty/uncertainty representations

• 1. Bayesian networks

  2. Discrete (See Relaxation)

  3. Fuzzy logic

  4. Intervals

  5. Probabilities

20. Vision paradigms

• 1. Active vision

  2. Geometric vision (See Vision Geometry and Mathematics and Geometric Representation of Model Features)

  3. Purposive Vision

  4. Qualitative Vision

21. Vision system design and characterization

• 1. Bounding box overlap assessment:

     1. IoU: Intersection over Union

     2. MABO: Mean Average Best Overlap

  2. Propagation of uncertainty

  3. Performance testing in vision

  4. Receiver operating characteristic