Module: Grayscale Reconstruction / Reconstruction By Dilation ()

Description:

Reconstruction By Dilation is one of the configurations of the module Grayscale Reconstruction. For other configurations, see the port Type.

This module can be used to remove the light points in a dark image. It performs a numerical reconstruction from a grayscale marker image into the grayscale input image.

At each step the marker image is dilated by a structuring element of size 1. This algorithm is based on a recursive method whereby the minimum of the dilated image and the mask image is retained and used as the marker image at the next step. This operation is repeated until stability. The number of iterations depends on the input image and the marker image.

As shown in the figure 1, peaks and ridges are levelled down, which results in a much more homogeneous output image.

This algorithm is used by the H-Maxima command.

Figure 1: 1D example of the Reconstruction By Dilation

See also: Reconstruction By Erosion, Watershed, Catchment Basins, Add To Label, Fill Label Boundaries.

Connections:

Input Image [required]
The image to be reconstructed. Supported types are: grayscale/color image (Uniform Scalar Field/Uniform Color Field), binary (Uniform Label Field with 2 labels) and label image (Uniform Label Field).

Input Marker Image [required]
The image of markers for reconstruction starting point. Supported types are: grayscale/color image (Uniform Scalar Field/Uniform Color Field), binary (Uniform Label Field with 2 labels) and label image (Uniform Label Field). Compatibility of input image and marker image types is required:

• grayscale 8-bit or 16-bit input image requires grayscale 8-bit or 16-bit marker image;
• grayscale 32-bit input image requires grayscale 32-bit marker image;
• grayscale floating point input image requires grayscale floating point marker image;
• binary or label input image requires binary or label input image.

Ports:

Type

This port selects the configuration of this module. The available configurations are:

• Reconstruction By Dilation,
• Reconstruction By Erosion.

Interpretation

This port specifies whether the input will be interpreted as a 3D volume or a stack of 2D images for processing.

• "3D": the module configuration is set to 3D. The image will be processed as a whole in 3D.
• "XY planes": the module configuration is set to 2D. The image will be processed slice per slice.

Neighborhood

In 3D configuration, this port refers to the type of connectivity considered for processing adjacent voxels:

• 6: voxels with a common face are considered connected
• 18: voxels with at least one common edge are considered connected
• 26: voxels with at least one common vertex are considered connected