Module: Canny Edge Detector ()

Description:

Classic edge detector. The algorithm consists of 4 steps:

1. Application of Gauss filter.
2. Application of Sobel filter (gradient filter).
3. Along the gradient direction (normal to edges), a non-maximal suppression (thinning) is performed.
4. A so-called "hysteresis thresholding": edges are connected with respect to lower and upper bounds provided by the user.

The result of the computation is a label image where all edge voxels are set to 1 and all others to 0. Press the Apply button to start the computation.

Connections:

Data [required]
A Uniform Scalar Field uniform scalar field.

Ports:

Gauss sigma

Sigma of the 2D Gaussian filter in units of the bounding box.

Gauss Kernel Size

Size of the 2D Gaussian filter kernel in voxels.

Suppression range

Number of neighboring voxels in both gradient directions which will be compared to each voxel.

Connect Edges Mask Size

Parameter for the hysteresis step. A value of 3 means that edges that are separated by one voxel are considered connected.

Connect Edges Threshold

All sobel edge values greater than Upper are regarded as true edges. If within their neighborhood as defined by connectEdgesMask there is a value greater than Lower, the voxel will be included in the true edges set.