Module: 2D-Histogram Texture Segmentation ()

Description:

This module is experimental. Experimental features offer a preview of features to come in future releases of . Note that no technical support will be provided for experimental features.

Using the 2D-Histogram Texture Segmentation script module, you can make the segmentation of your images including areas containing high-frequency textured areas.

Compared to 2D-Histogram Segmentation, the initial marker image is produced a 2D scatter-plot (histogram) of Image Intensity vs Texture Map. The region expansion is performed using the marker-based watershed algorithm using the image gradient magnitude as a landscape image. The texture map is obtained as a difference between the grayscale closing and opening of the input intensity image, using the structuring element defined by the corresponding ports. This is equivalent to summing a black and white top-hat images. The texture map will therefore give a low intensity response in homogeneous areas, and a higher response in high-frequency textured areas (e.g. small checker-board or fibrous patterns)

The 6 steps of the 2D histogram segmentation:

• Step 1 - Compute gradient magnitude and texture map to begin. The shape, size and possibly orientation of the structuring element is chosen in the corresponding ports, and control the behavior of the Texture Map to be computed. Refer to the documentation of Line Closing 3D or Disc Closing 3D modules for the interpretation of the angles. Detail in Computing the gradient magnitude:
  Choose between precise computation of the gradient magnitude or a faster approximation of the gradient magnitude.
  Hit Next: Compute Gradient Magnitude to continue.
  
• Step 2 - Plot histogram see details in Interpreting the histogram:
  Hit Next: Plot Histogram to continue.
  
• Step 3 - Draw windows - see details in Use the histogram to initialize the classification:
  If you used the mouse to drag the plot window, and you wish to reset to its original size and position, you can simply drag the Gamma Correction slider on the Properties panel.
  Use gamma correction to visualize peaks on the 2D histogram. Use drawing tools at top of plot window to draw windows directly on the plot. Generally, you want to select regions that have very low gradient-magnitude (peaks close to the x axis).
  Hit Next: Compute Seed to proceed to the next step.
  
• Step 4 - Review seed labels:
  Use weight factor to update the view to show the grayscale data, the seeded label image, or a weighted fusion of the two data sets.

  Use the Hide/Show Histogram button to toggle the display of the histogram so that it is easier to view the results in the Viewer. Use slice number to adjust the slice index plane that is being displayed. If you are not satisfied with the seeds computed from your chose windows, hit Delete Last Window to delete windows one at a tie (in the reverse order from which they were added). Then you can draw new windows. After you have drawn new windows, hit Recompute Seed. You can iterate until you are satisfied with the seed.

  Hit Next: Apply Watershed to proceed (see details in Use watershed to expand)

• Step 5 - Confirm watershed results:
  If the results are not satisfactory, draw new regions, and then reapply watershed with hitting Previous: Recompute Seed.
  Hit Next to proceed to the final step, which is optional.
  
• Step 6 - [Optional] Delete a channel:
  If one of the assigned labels corresponds to void space, you may choose to delete that label from the label image.

  You should delete this module when you are done.

Connections:

Data [required]
Image data to be segmented.

Ports:

Intensity

Displays information on the input gray-level image.

Gradient Magnitude

Displays information on the computed gradiant magnitude.

Visibility

This port is only displayed at the step 3, step 4, step 5 and step 6.
Controls the visibility of the histogram.

Slice Number

This port is only displayed at the step 4, step 5 and step 6.
Adjusts the slice index plane that is being displayed.

Weight Factor

This port is only displayed at the step 4, step 5 and step 6.
Choose the weight factor to update the view to show the grayscale data, the seeded label image, or a weighted fusion of the two data sets.

Colormap

This port is only displayed at the step 4, step 5 and step 6 and chooses the colormap to use for displaying seeds and segmented image.

Next Action

Displays information concerning the next action to come and the current step.

Struturing Element

This port is only displayed at the step 1 and selects the shape of the structuring element used for computing the texture map.

Size

This port is only displayed at the step 1 and chooses the size of the structuring element in pixels. See Closing.

Theta Angle

This port is only displayed at the step 1 and chooses the theta angle for the line or disc structuring elements. See Line Closing 3D or Disc Closing 3D.

Phi Angle

This port is only displayed at the step 1 and chooses the phi angle for the line or disc structuring elements. See Line Closing 3D or Disc Closing 3D.

Gradient Method

This port is only displayed at the step 1 and chooses the method to use to compute the gradient magnitude.

Gamma Correction

This port is only displayed at the step 3, step 4, step 5 and step 6.
Chooses the gamma correction to apply to the histogram.

Delete Channel

This port is only displayed at the step 6 and enables deleting a specific channel which corresponds to void space.

Action



Proposes the different actions to apply. At each step, the list of actions is updated.

Layer Order

This port is only visible if the module is an overlay of an Ortho Views.