`skimage.feature.brief:BRIEF`

source: /dev/scikit-image/src/skimage/feature/brief.py :19

Signature

   class     BRIEF (    descriptor_size    =  256   ,    patch_size    =  49   ,    mode    =  normal   ,    sigma    =  1   ,    rng    =  1     )

Members

Summary

BRIEF binary descriptor extractor.

Extended Summary

BRIEF (Binary Robust Independent Elementary Features) is an efficient feature point descriptor. It is highly discriminative even when using relatively few bits and is computed using simple intensity difference tests.

For each keypoint, intensity comparisons are carried out for a specifically distributed number N of pixel-pairs resulting in a binary descriptor of length N. For binary descriptors the Hamming distance can be used for feature matching, which leads to lower computational cost in comparison to the L2 norm.

Parameters

descriptor_size : int, optional

Size of BRIEF descriptor for each keypoint. Sizes 128, 256 and 512 recommended by the authors. Default is 256.

patch_size : int, optional

Length of the two dimensional square patch sampling region around the keypoints. Default is 49.

mode : {'normal', 'uniform'}, optional

Probability distribution for sampling location of decision pixel-pairs around keypoints.

rng : {`numpy.random.Generator`, int}, optional

Pseudo-random number generator (RNG). By default, a PCG64 generator is used (see numpy.random.default_rng). If rng is an int, it is used to seed the generator.

The PRNG is used for the random sampling of the decision pixel-pairs. From a square window with length patch_size, pixel pairs are sampled using the mode parameter to build the descriptors using intensity comparison.

For matching across images, the same rng should be used to construct descriptors. To facilitate this:

rng defaults to 1
Subsequent calls of the extract method will use the same rng/seed.

sigma : float, optional

Standard deviation of the Gaussian low-pass filter applied to the image to alleviate noise sensitivity, which is strongly recommended to obtain discriminative and good descriptors.

Attributes

descriptors : (Q, `descriptor_size`) array of dtype bool: 2D ndarray of binary descriptors of size descriptor_size for Q keypoints after filtering out border keypoints with value at an index (i, j) either being True or False representing the outcome of the intensity comparison for i-th keypoint on j-th decision pixel-pair. It is Q == np.sum(mask).
mask : array of shape (N,) and dtype bool: Mask indicating whether a keypoint has been filtered out (False) or is described in the descriptors array (True).

Examples

from skimage.feature import (corner_harris, corner_peaks, BRIEF,
                             match_descriptors)
import numpy as np
square1 = np.zeros((8, 8), dtype=np.int32)
square1[2:6, 2:6] = 1
square1
square2 = np.zeros((9, 9), dtype=np.int32)
square2[2:7, 2:7] = 1
square2
keypoints1 = corner_peaks(corner_harris(square1), min_distance=1)
keypoints2 = corner_peaks(corner_harris(square2), min_distance=1)
extractor = BRIEF(patch_size=5)
extractor.extract(square1, keypoints1)
descriptors1 = extractor.descriptors
extractor.extract(square2, keypoints2)
descriptors2 = extractor.descriptors
matches = match_descriptors(descriptors1, descriptors2)
matches
keypoints1[matches[:, 0]]
keypoints2[matches[:, 1]]

✓

Aliases

skimage.feature.BRIEF