Superpixels Segmentation

This algorithm generates an unsupervised oversegmentation of an image, based on the SLIC superpixel algorithm.

Input

A single image or volume

Output

A 16bit label map with the same size of the input. Optionally, if Export Simplified Image is true, a second image where the intensity of each pixel corresponds to the average intensity within its superpixels.

Description

The algorithm is based on the method described in “SLIC Superpixels Compared to State-of-the-art Superpixel Methods” by Achanta et al. The standard parameters defined in this paper are Number of superpixels, Max size of a superpixel, Adaptiveness and Number of iterations.

If the flag Random indices is set, the assigned labels will be shuffled, otherwise it will be in-order (which makes the visualization of the resulting label map less clear).

If the parameter Working Size is greater than zero, the superpixel detection will be performed on a subsample image such that its total number of pixels is equal to Working Size. The superpixel segmentation will then be refined one last time at full resolution, unless Skip Refinement is set.