Tool Superpixel Segmentation
The Superpixel Segmentation tool implements the 'Simple Linear Iterative Clustering' (SLIC) algorithm, an image segmentation method described in Achanta et al. (2010).
SLIC is a simple and efficient method to decompose an image in visually homogeneous regions. It is based on a spatially localized version of k-means clustering. Similar to mean shift or quick shift, each pixel is associated to a feature vector.
This tool is follows the SLIC implementation created by Vedaldi and Fulkerson as part of the VLFeat library.
References
- SLIC at VLFeat.org
- Achanta, R., Shaji, A., Smith, K., Lucchi, A., Fua, P., & Süsstrunk, S. (2010): Slic Superpixels. EPFL Technical Report no. 149300, June 2010. epfl.ch.
- Achanta, R., Shaji, A., Smith, K., Lucchi, A., Fua, P., & Süsstrunk, S. (2012): SLIC Superpixels compared to state-of-the-art superpixel methods. IEEE transactions on pattern analysis and machine intelligence, 34(11), 2274-2282. ieee.org.
- Author: O.Conrad (c) 2019
- Menu: Imagery|Segmentation
Parameters
| Name | Type | Identifier | Description | Constraints |
Input | Features | Grid list, input | FEATURES | - | - |
Output | Segments | Shapes, output | POLYGONS | - | - |
Superpixels (*) | Grid list, output, optional | SUPERPIXELS | - | - |
Options | Grid System | Grid system | PARAMETERS_GRID_SYSTEM | - | - |
Normalize | Boolean | NORMALIZE | - | Default: 0 |
Maximum Iterations | Integer | MAX_ITERATIONS | - | Minimum: 1 Default: 100 |
Regularization | Floating point | REGULARIZATION | - | Minimum: 0.000000 Default: 1.000000 |
Region Size | Integer | SIZE | Starting 'cell size' of the superpixels given as number of cells. | Minimum: 1 Default: 10 |
Minimum Region Size | Integer | SIZE_MIN | In postprocessing join segments, which cover less cells than specified here, to a larger neighbour segment. | Minimum: 1 Default: 1 |
Create Superpixel Grids | Boolean | SUPERPIXELS_DO | - | Default: 0 |
Post-Processing | Choice | POSTPROCESSING | - | Available Choices: [0] none [1] unsupervised classification Default: 0 |
Number of Clusters | Integer | NCLUSTER | - | Minimum: 2 Default: 12 |
Split Clusters | Boolean | SPLIT_CLUSTERS | - | Default: 1 |
(*) optional |
Command-line
Usage: saga_cmd imagery_segmentation 4 [-FEATURES <str>] [-NORMALIZE <str>] [-POLYGONS <str>] [-MAX_ITERATIONS <num>] [-REGULARIZATION <double>] [-SIZE <num>] [-SIZE_MIN <num>] [-SUPERPIXELS_DO <str>] [-SUPERPIXELS <str>] [-POSTPROCESSING <str>] [-NCLUSTER <num>] [-SPLIT_CLUSTERS <str>]
-FEATURES:<str> Features
Grid list, input
-NORMALIZE:<str> Normalize
Boolean
Default: 0
-POLYGONS:<str> Segments
Shapes, output
-MAX_ITERATIONS:<num> Maximum Iterations
Integer
Minimum: 1
Default: 100
-REGULARIZATION:<double> Regularization
Floating point
Minimum: 0.000000
Default: 1.000000
-SIZE:<num> Region Size
Integer
Minimum: 1
Default: 10
-SIZE_MIN:<num> Minimum Region Size
Integer
Minimum: 1
Default: 1
-SUPERPIXELS_DO:<str> Create Superpixel Grids
Boolean
Default: 0
-SUPERPIXELS:<str> Superpixels
Grid list, output, optional
-POSTPROCESSING:<str> Post-Processing
Choice
Available Choices:
[0] none
[1] unsupervised classification
Default: 0
-NCLUSTER:<num> Number of Clusters
Integer
Minimum: 2
Default: 12
-SPLIT_CLUSTERS:<str> Split Clusters
Boolean
Default: 1