Parameter-Free Image Segmentation with Slic

Author

Fabian Boemer
Edward Ratner
Amaury Lendasse, Missouri University of Science and TechnologyFollow

Abstract

In This Paper, We Develop a Parameter-Free Image Segmentation Framework using Simple Linear Iterative Clustering (Slic) and Extreme Learning Machines (Elm). Slic Requires a Single Parameter, the Number of Centroids K. Our Framework, Called Pf-Slic (Parameter-Free Slic) Uses an Elm to Predict the Optimal K, Generating a Parameter-Free Framework. Pf-Slic and its Streaming Variant Spf-Slic (Streaming Pf-Slic) Achieve Performance Comparable to Other Models on Ultra-High-Definition (4k) Images and Streams, with Runtimes Orders of Magnitude Lower.

Recommended Citation

F. Boemer et al., "Parameter-Free Image Segmentation with Slic," Neurocomputing, vol. 277, pp. 228 - 236, Elsevier, Feb 2018.

The definitive version is available at https://doi.org/10.1016/j.neucom.2017.05.096

Department(s)

Engineering Management and Systems Engineering

Keywords and Phrases

ELM; Image segmentation; SLIC; Streaming; Superpixel

International Standard Serial Number (ISSN)

1872-8286; 0925-2312

Document Type

Article - Journal

Document Version

Citation

File Type

text

Language(s)

English

Rights

© 2024 Elsevier, All rights reserved.

Publication Date

14 Feb 2018