Module: Classify_Points

Description:

Topologically classifies the pixels of a 3D image (especially skeletons), contained in a HxUniformScalarField3. For each pixel of the object (defined by the given threshold and comparison thresholding method), three values are determined: the two topological numbers t and t', respectively the number of connected components of the object and of the background in the pixel neighbourhood, and n the number of neighbours of the pixel¹. The following table summarises the classification, and gives the code used in the output image:

t	t'	n	Type	Output value
*	0	*	Interior	2
0	1	*	Isolated	3
1	1	1	Curve end	4
1	1	>1	Border point	5
2	1	2	Curve	6
>1	1	>2	Curve junction	7
*	>1	*	Surface or surface junction	8

NOTE: All other possibilities are impossible (such as t = 1 and n = 0).

Illustration of the classification using the Draw_Cubes module and a color code for the pixel types (with the colormap tensteps.col). Dark blue means isolated, orange means curve junctions, red are surfaces, etc.

Connections:

Image

[required]
The input image, defined in a class HxUniformScalarField3. The intensities of the pixels can only be stored as bytes.

Ports:

Threshold

Value that defines whether a pixel belongs to the object or not.

Comparison

Defines how the threshold is used to define the pixels of the object: should the pixel value be lower, lower or equal, greater, etc. than the threshold?

Connectivity

Choice of the neighbourhood.

Illustration of the different connectivities. (a): 6-, (b): 18-, and (c): 26-connectivity (in case you didn't figure it out, n-connectivity means n pixels in the neighbourhood).

Action button

Push this button to start the computation.

Commands:

Additional options can be accessed when typing in the console Classify_Points COMMAND_NAME.

`verbose`

Displays timing information after the computation. Retype to hide info.

`create`

Runs the computation. Returns the name of the output, so it can be used in a script, such as set RESULT [Classify_Points create].

References:

¹ Malandain, G., Bertrand, G., and Ayache, N. (1992). Topological segmentation of discrete surfaces. International Journal of Computer Vision 10(2): 183-197.