Image Statistics

Some of the docstrings for this module have been extracted from the scikit-image library and are covered by their respective licenses.

class node_imagestatistics.ImageStatistics

Extracts tabular data from an image using one of a selection of algorithms. The implemented filters are to a large extent based on scikit-image, but some filters are not from this package.

Algorithms:

• blob, DoG

  Detects blobs in a grayscale image using the Difference of Gaussians (DoG) method. Returns table with X,Y positions

  and the standard deviation of the gaussian kernel that detected each blob. Operates only on first channel of image.

  The radius of each blob is approximately sqrt(2) sigma. Operates only on first channel of image

  min sigma:

  Minimum standard deviation for Gaussian kernels. Keep low to detect smaller blobs (default 1)

  max sigma:

  Maximum standard deviation for Gaussian kernels. Keep high to detect larger blobs (default 50)

  sigma_ratio:

  The ratio between the standard deviation of Gaussian Kernels used for computing the Difference of Gaussians (default 1.6)

  threshold:

  The absolute lower bound for scale space maxima. Local maxima smaller than thresh are ignored. Reduce this to detect blobs with less intensities. (default 2.0)

  overlap:

  A value between 0 and 1. If the area of two blobs overlaps by a fraction greater than threshold, the smaller blob is eliminated. (default 0.5)

• blob, DoH

  Detects blobs in a grayscale image using the Determinant of Hessian (DoH) method. Returns table with X,Y positions

  and the standard deviation of the gaussian kernel used for the hessian matrix which detected each blob.

  The radius of each blob is approximately sigma. Operates only on first channel of image

  min sigma:

  Minimum standard deviation for Gaussian kernels. Keep low to detect smaller blobs (default 1)

  max sigma:

  Maximum standard deviation for Gaussian kernels. Keep high to detect larger blobs (default 50)

  num sigma:

  The number of intermediate values for Sigma to consider (default 10)

  threshold:

  The absolute lower bound for scale space maxima. Local maxima smaller than thresh are ignored. Reduce this to detect blobs with less intensities. (default 0.01)

  overlap:

  A value between 0 and 1. If the area of two blobs overlaps by a fraction greater than threshold, the smaller blob is eliminated. (default 0.5)

  log scale:

  If true then interpolation of intermediate values for Sigma are interpolated in a logrithmic scale, otherwise linear interpolation is used

• blob, LoG

  Detects blobs in a grayscale image using the Laplacian of Gaussian method. Returns table with X,Y positions

  and the standard deviation of the gaussian kernel used for detecting each blob.

  The radius of each blob is approximately sqrt(2) sigma. Operates only on first channel of image

  min sigma:

  Minimum standard deviation for Gaussian kernels. Keep low to detect smaller blobs (default 1)

  max sigma:

  Maximum standard deviation for Gaussian kernels. Keep high to detect larger blobs (default 50)

  num sigma:

  The number of intermediate values for Sigma to consider (default 10)

  threshold:

  The absolute lower bound for scale space maxima. Local maxima smaller than thresh are ignored. Reduce this to detect blobs with less intensities. (default 0.2)

  overlap:

  A value between 0 and 1. If the area of two blobs overlaps by a fraction greater than threshold, the smaller blob is eliminated. (default 0.5)

  log scale:

  If true then interpolation of intermediate values for Sigma are interpolated in a logrithmic scale, otherwise linear interpolation is used

• cdf

  Computes the cumulative distribution function (cdf) over all pixels.

  Returns cdf value and bin-center for a number of bins. Processes each channel separately and returns separate columns

  bins:

  Number of bins for processing

• features, CENSURE+DAISY

  Combines the CENSUR keypoint detector with a DAISY feature extractor at the detected keypoints – picking the closest DAISY feature if non are at the exact keypoint locations

  step:

  DAISY: Distance between descriptor samples (default 4)

  radius:

  DAISY: Radius of the outermost ring (default 15)

  rings:

  DAISY: Number of rings (default 3)

  histograms:

  DAISY: Number of histograms per ring (default 8)

  orientations:

  DAISY: Orientations per histogram (default 8)

  daisy normalization:

  DAISY: Method for normalizing descriptors

  min scale:

  CENSURE: Minimum scale to extract keypoints from.

  max scale:

  CENSURE: Maximum scale to extract keypoints from. The keypoints will be extracted from all the scales except the first and the last i.e. from the scales in the range [min_scale + 1, max_scale - 1]. The filter sizes for different scales is such that the two adjacent scales comprise of an octave.

  non max threshold:

  CENSURE: Threshold value for maximas and minimas with a weak magnitude response obtained after Non-Maximal Suppression.

  line threshold:

  CENSURE: Threshold for interest points which have ratio of principal curvatures greater than this value.

• features, daisy

  Extracts all (densely) DAISY features into a table.

  DAISY is a feature descriptor similar to SIFT formulated in a way that allows for fast dense extraction.

  Typically, this is practical for bag-of-features image representations. Operates only on first channel of image

  step:

  Distance between descriptor sampling points (default 4)

  radius:

  Radius in pixels of the outermost ring (default 15)

  rings:

  Number of rings (default 3)

  histograms:

  Number of histograms sampled per ring (default 8)

  orientations:

  Number of orientations per histogram (default 8)

  daisy normalization:

  Method for normalizing descriptors

• features, orb

  Oriented FAST and rotated BRIEF feature detector and binary descriptor extractor.

  N:

  Number of keypoints to be returned. The best N keypoints will be returned, or all found keypoints if fewer.

  fast threshold:

  The threshold parameter in feature.corner_fast. Decrease the threshold when more corners are desired (0.08)

  fast n:

  The n parameter in skimage.feature.corner_fast (9)

  harris k:

  The k parameter in skimage.feature.corner_harris. Sensitivity factor to separate corners from edges. (0.04)

  downscale:

  Downscale factor for the image pyramid (1.2)

• features, probabilistic hough lines

  Return lines from a progressive probabilistic line Hough transform.

  threshold:

  Threshold for hough accumulators

  line length:

  Minimum accepted length of detected lines. Increase parameter to extract only longer lines

  line gap:

  Maximum gap between pixels to still form a line. Increase the parameter to merge broken lines more aggresively.

  min theta:

  Minium angle in radians for lines, default -pi/2

  max theta:

  Maximum angle in radians for lines, default +pi/2

  num theta:

  Number of theta samples in the given range

• histogram

  Computes the histogram over all pixels.

  Returns histogram count and bin-center for a number of bins. Processes each channel separately and returns separate columns

  bins:

  Number of bins for processing

  min value:

  Lower end of range for bins

  max value:

  Upper end of range for bins

• keypoints, CENSURE

  CENSURE keypoint detector.

  min scale:

  Minimum scale to extract keypoints from.

  max scale:

  Maximum scale to extract keypoints from. The keypoints will be extracted from all the scales except the first and the last i.e. from the scales in the range [min_scale + 1, max_scale - 1]. The filter sizes for different scales is such that the two adjacent scales comprise of an octave.

  non max threshold:

  Threshold value used to suppress maximas and minimas with a weak magnitude response obtained after Non-Maximal Suppression.

  line threshold:

  Threshold for rejecting interest points which have ratio of principal curvatures greater than this value.

• meta

  Returns meta information about image: width, height, channels, min_value, max_value, sum

  prefix:

  Prefix added before column names

• peaks, corners

  Gives a table with coordinates of corners in a corner measure response image.

  Suppresses multiple connected peaks with same accumulator value. Operates only on first channel of image

  threshold:

  Threshold for minimum intenstity of peaks expressed as fraction of image maximum (default 0.1)

  min distance:

  Minimum distance between two corners

• peaks, local max

  Gives a table with coordinates of the peaks in an image

  Peaks are local maxima in a region defined by “min distance”. Operates only on first channel of image

  threshold:

  Threshold for minimum intenstity of peaks expressed as absolute values (default 0)

  min distance:

  Minimum distance between two corners

  threshold relative:

  Threshold for minimum intenstity of peaks expressed as fraction of image maximum (default 0)

Inputs:

source : image

Image to extract statistics from

Outputs:

result : table

Table with results