C++ Computer Vision functions available

BaySAC (and RANSAC and PROSAC and SimSAC) for Essential Matrix estimation and Homography estimation. Topdown and non-linear refinement of solution.

• Cmake will build all libraries including util, params and ransac. Link to all three and call getE and getH. Example of use at https://github.com/kayak-tom/tom-cv/blob/master/ransacTest/example.cpp (example also requires cameraGeom, this file is a bit of a mess at the moment).

Several different 5-point Essential Matrix estimation algorithms.

See workspace/cameraGeom/geom/EssentialMatGoldStandard.cpp for Levenberg-Marquardt refinement of E (also function to minimise Sampson error by LM which is a lot faster and works just as well).

Homography decomposition (fit rotation/translation/plane-normal by Levenberg refinement) -- see geom.cpp

Bag-of-Words library: Real-time performance and dynamic retraining. Supports 10k+ images. Fast pairwise correspondences.

• Cmake will build all libraries including util, params, featureDescription and bagOfWords. Link to all four. Example of use https://github.com/kayak-tom/tom-cv/blob/master/bowExample/bowExample.cpp.

K-medoids clustering

• See https://github.com/kayak-tom/tom-cv/blob/master/featureDescription/description/descriptorMetricSpace.cpp
• See method CMetricSpaceDescriptorSet::kMedoidCluster. I will extract this to be used independently at some stage.
• Currently a hard-coded limit of 450 data points under Windows--could easily be adapted.