bob.ip.SIFT¶

class bob.ip.SIFT((object)self, (int)height, (int)width, (int)n_octaves, (int)n_scales, (int)octave_min[, (float)sigma_n=0.5[, (float)sigma0=1.6[, (float)contrast_thres=0.03[, (float)edge_thres=10.0[, (float)norm_thres=0.2[, (float)kernel_radius_factor=4.0[, (BorderType)border_type=bob.sp._sp.BorderType.Mirror]]]]]]]) → None :¶

Bases: Boost.Python.instance

This class allows after configuration the extraction of SIFT descriptors.

Reference: ‘Distinctive Image Features from Scale-Invariant Keypoints’, D. Lowe, International Journal of Computer Vision, 2004

Creates an object that allows the extraction of SIFT descriptors.

__init__( (object)self, (SIFT)other) -> None

__init__((object)self, (int)height, (int)width, (int)n_octaves, (int)n_scales, (int)octave_min[, (float)sigma_n=0.5[, (float)sigma0=1.6[, (float)contrast_thres=0.03[, (float)edge_thres=10.0[, (float)norm_thres=0.2[, (float)kernel_radius_factor=4.0[, (BorderType)border_type=bob.sp._sp.BorderType.Mirror]]]]]]]) → None :¶

Creates an object that allows the extraction of SIFT descriptors.

__init__( (object)self, (SIFT)other) -> None

Methods

`__init__`((object)self, (int)height, …)	Creates an object that allows the extraction of SIFT descriptors.
`compute_descriptor`((SIFT)self, (object)src, …)	Computes SIFT descriptor for a 2D/grayscale image, at the given keypoints.
`get_descriptor_shape`((SIFT)self)	Returns the shape of a descriptor for a given keypoint
`set_sigma0_no_init_smoothing`((SIFT)self)	Sets sigma0 such that there is not smoothing at the first scale of octave_min.

Attributes

`contrast_threshold`	The contrast threshold used during keypoint detection
`conv_border`	The way the extractor deals with convolution at the boundary of the image when computing the Gaussian scale space.
`edge_threshold`	The edge threshold used during keypoint detection
`gaussian_window_size`	The Gaussian window size for the descriptor
`height`	The height of the images to process
`kernel_radius_factor`	Factor used to determine the kernel radii (size=2*radius+1).
`magnif`	The magnification factor for the descriptor
`n_bins`	The number of bins for the descriptor
`n_blocks`	The number of blocks for the descriptor
`n_intervals`	The number of intervals of the pyramid.
`n_octaves`	The number of octaves of the pyramid
`norm_epsilon`	The epsilon value added during the descriptor normalization
`norm_threshold`	The norm threshold used during descriptor normalization
`octave_max`	The index of the maximum octave (read-only).
`octave_min`	The index of the minimum octave
`sigma0`	The value sigma0 of the standard deviation for the input image
`sigma_n`	The value sigma_n of the standard deviation for the nominal/initial octave/scale
`width`	The width of the images to process

compute_descriptor((SIFT)self, (object)src, (object)keypoints) → object :¶: Computes SIFT descriptor for a 2D/grayscale image, at the given keypoints. The dst array will be allocated and returned.

contrast_threshold¶: The contrast threshold used during keypoint detection

conv_border¶: The way the extractor deals with convolution at the boundary of the image when computing the Gaussian scale space.

edge_threshold¶: The edge threshold used during keypoint detection

gaussian_window_size¶: The Gaussian window size for the descriptor

get_descriptor_shape((SIFT)self) → tuple :¶: Returns the shape of a descriptor for a given keypoint

height¶: The height of the images to process

kernel_radius_factor¶: Factor used to determine the kernel radii (size=2*radius+1). For each Gaussian kernel, the radius is equal to ceil(kernel_radius_factor*sigma_{octave,scale}).

magnif¶: The magnification factor for the descriptor

n_bins¶: The number of bins for the descriptor

n_blocks¶: The number of blocks for the descriptor

n_intervals¶: The number of intervals of the pyramid. Three additional scales will be computed in practice, as this is required for extracting SIFT features.

n_octaves¶: The number of octaves of the pyramid

norm_epsilon¶: The epsilon value added during the descriptor normalization

norm_threshold¶: The norm threshold used during descriptor normalization

octave_max¶: The index of the maximum octave (read-only). This is equal to octave_min+n_octaves-1.

octave_min¶: The index of the minimum octave

set_sigma0_no_init_smoothing((SIFT)self) → None :¶: Sets sigma0 such that there is not smoothing at the first scale of octave_min.

sigma0¶: The value sigma0 of the standard deviation for the input image

sigma_n¶: The value sigma_n of the standard deviation for the nominal/initial octave/scale

width¶: The width of the images to process