Feature Selection

Algorithms that improve segmentation based on topological criteria.

KillBorder2d: Removes the objects connected to the borders of a segmented 2D image.
KillBorder3d: Removes the objects connected to the borders of a segmented image.
RemoveSmallSpots: Removes the smallest objects of a binary image.
RemoveIslands: Remove areas of connected voxels (islands) less than or equal to a specified size.
ObjectBoundaries: Extracts the boundary points of objects from a segmented image.
BinarySmoothing2d: Transforms a binary two-dimensional image into another binary image with respect to the real local mean of each pixel neighborhood.
BinarySmoothing3d: Transforms a binary three-dimensional image into another binary image with respect to the real local mean of each voxel neighborhood.
LabelInterfaces: Generates a binary image containing points having a minimum number of different neighbors in a label input image.

The Feature Selection group contains algorithms helpful to improve a segmentation by retaining some pixels in respect with their topological properties like:

Their connectivity to image borders
Size of the objects they are belonging to
Local configuration around their neighborhood