Revision history [back]

Only way to set paramaters in SIFT is create method

About octave it is defined in source code here SIFT implementation in OpenCV is based on the code from http://robwhess.github.io/opensift/

reference paper is Distinctive Image Features from Scale - Invariant Keypoints (David G. Lowe International Journal of Computer Vision November 2004, Volume 60, Issue 2, pp 91–110)

Only way to set paramaters in SIFT is create method

About octave it is defined in source code here SIFT implementation in OpenCV is based on the code from http://robwhess.github.io/opensift/

Best values depend on its context.

reference paper is Distinctive Image Features from Scale - Invariant Keypoints (David G. Lowe International Journal of Computer Vision November 2004, Volume 60, Issue 2, pp 91–110)

Only way to set paramaters in SIFT is create method

About octave it is defined in source code here SIFT implementation in OpenCV is based on the code from http://robwhess.github.io/opensift/

Best values depend on its context.

reference paper is Distinctive Image Features from Scale - Invariant Keypoints (David G. Lowe International Journal of Computer Vision November 2004, Volume 60, Issue 2, pp 91–110)

Only way to set paramaters in SIFT is create method

About octave it is defined in source code here SIFT implementation in OpenCV is based on the code from http://robwhess.github.io/opensift/

Best values depend on its context.

reference paper is Distinctive Image Features from Scale - Invariant Keypoints (David G. Lowe International Journal of Computer Vision November 2004, Volume 60, Issue 2, pp 91–110)

Update

Gaussian Pyramid is calculated here. Each layer is blurred (gaussian) of sigma[i]. i is ranged from 0 to nOctaveLayers + 3

Laplacian gaussian pyramidis pyramid is calculated here. For each layer nOctaveLayers + 2 dog are computed