Table of Content

Harris

Gradient-based corner detector. bmvc88

Shi-Tomasi

Gradient-based corner detector. cvpr94 paper

SIFT

(scale invariant feature transform) Gradient-based corner detector and descriptor. iccv99 ijcv04 Main steps:

  1. Scale-space extrema detection: The image is convolved with Gaussian filters at different scales, and then the difference of successive Gaussian-blurred images are taken. Keypoints are then taken as maxima/minima of the Difference of Gaussians (DoG) that occur at multiple scales.
  2. Keypoint localization: 2.1 Interpolation of nearby data for accurate position 2.2 Discarding low-contrast keypoints 2.3 Eliminating edge responses
  3. Orientation assignment Each keypoint is assigned one or more orientations based on local image gradient directions, and the gradient magnitude and orientation are precomputed using pixel differences.
  4. 128-d Keypoint descriptor Create a set of orientation histograms on 4×4 pixel neighborhoods with 8 bins each

SURF

(speeded-up robust features) Gradient-based eccv06

FAST

(features from accelerated segment test) Intensity-based corner detector. eccv06 procedure Remarks:

  1. Check 1,5,9,13 for fast detection;
  2. Non-maximum suppression.

BRIEF

(binary robust independent elementary features) eccv10

BRISK

(binary robust invariant scalable keypoints) iccv11

ORB

(oriented FAST and rotated BRIEF) iccv11

FREAK

(fast retina keypoint) cvpr12

Back to Top