Operation: hessian



Author:

The operation uses the cost computation from NeuronJ by Erik Meijering. There exists a seperated plugin to compute the Hessian in FeatureJ.

Example

eigenvalues and normalized eigenvectors of the hessian for smoothing scale = 1, 2, 4              

                                  

Description

Creates a cost image and a vector field computed from the eigenvalues and eigenvectors of the Hessian of the input image. The Hessian derivative can be used to discriminate locally between plate-like, line-like, and blob-like image structures. The Hessian matrix of a scalar function of an n-dimensional vector is the symetric nxn matrix of second partial derivatives.

Options



scale: the smoothing scale is equal to the standard deviation of the Gaussian derivative kernel used in computing the second-order derivatives

Parameter

The only parameter is the input image. The input image must be a 2D, 8bit, greyscale image.

Results

The result is a Hessian image that containes both the eigenvalue array and the eigenvector array. To convert them to a normal, displayable image use the operations "get image from hessian" and "get vector image from hessian".