This is documentation for Mathematica 8, which was
based on an earlier version of the Wolfram Language.

CornerFilter

 CornerFilter[image] computes a measure for the presence of a corner for each pixel in image and returns the result as an intensity image. CornerFilterdetects corners at a pixel range r.
• CornerFilter implements a variety of corner detection methods based on gradient computations.
• CornerFilter works with arbitrary grayscale and color images.
• When applied to multichannel images, CornerFilter finds corner signatures across channels.
• CornerFilter specifies different radii in vertical and horizontal directions.
• CornerFilter takes a Method option that specifies how to compute the corner metric. The default setting is . Available methods include:
 "HarmonicMean" harmonic mean of eigenvalues method "HarrisStephens" Harris-Stephens corner detection {"HarrisStephens",k} Harris-Stephens method with sensitivity parameter k "ShiTomasi" minimum eigenvalue method
• By default, the Harris-Stephens method uses a sensitivity parameter k=0.05.
• The parameters of the Gaussian derivatives to be computed can be specified by an option setting . The default setting is .
Find small-scale corners in an image:
Large-scale corners:
Find corners in text:
Use a rectangular pixel range:
Find small-scale corners in an image:
 Out[1]=

Large-scale corners:
 Out[1]=

Find corners in text:
 Out[1]=

Use a rectangular pixel range:
 Out[1]=
 Scope   (2)
Detect a corner:
Detect corners at a given scale:
 Options   (2)