[F,D] = IPLDFEAT(I) uses Dorko's IPLD SIFT code to compute frames F and descriptors D.
Options:
- Detector ['DoG']
Detector type (DoG, Harris, Hessian, LoG).
- Angle [0]
Do oriented keypoints.
- Adapt [0] (incomplete and untested!!)
Do affine adaptation (implies 'Angle')
- Frames [[]]
Compute descriptors of these frames. Due to a limitation of IPLD interface, no angle can be specified.
- Magnif [[3]]
SIFT descriptor magnification factor. This parameter is compatible with VLFeat SIFT. For a frame of scale 1, Magnif=3 implies that the size of a spatial bin is 3 pixels. Since there are four of those along each dimension, this corresponds to a patch sioze of 12.
Verbosity [0]
Verbosity level.