### Abstract

We propose a new graph-based approach for performing a multilabel, interactive image segmentation using the principle of random walks. Using the random walk principle, given a set of user-defined (or prelabeled) pixels as labels, one can analytically calculate the probability of walking from each unlabeled pixel to each labeled pixel, thereby defining a vector of probabilities for each unlabeled pixel. By efficiently combining this vector of probabilities obtained for each unlabeled pixel, they can be assigned to one of the labels using the watershed algorithm to obtain an image segmentation. We present quantitative and qualitative results, comparing our new algorithm with the original random walker image segmentation algorithm.

Original language | English (US) |
---|---|

Title of host publication | 2013 IEEE International Conference on Acoustics, Speech, and Signal Processing, ICASSP 2013 - Proceedings |

Pages | 1473-1477 |

Number of pages | 5 |

DOIs | |

State | Published - Oct 18 2013 |

Event | 2013 38th IEEE International Conference on Acoustics, Speech, and Signal Processing, ICASSP 2013 - Vancouver, BC, Canada Duration: May 26 2013 → May 31 2013 |

### Publication series

Name | ICASSP, IEEE International Conference on Acoustics, Speech and Signal Processing - Proceedings |
---|---|

ISSN (Print) | 1520-6149 |

### Other

Other | 2013 38th IEEE International Conference on Acoustics, Speech, and Signal Processing, ICASSP 2013 |
---|---|

Country | Canada |

City | Vancouver, BC |

Period | 5/26/13 → 5/31/13 |

### Fingerprint

### Keywords

- Image segmentation
- combinatorial Dirichlet problem
- graph theory
- random walks
- watersheds

### ASJC Scopus subject areas

- Software
- Signal Processing
- Electrical and Electronic Engineering

### Cite this

*2013 IEEE International Conference on Acoustics, Speech, and Signal Processing, ICASSP 2013 - Proceedings*(pp. 1473-1477). [6637896] (ICASSP, IEEE International Conference on Acoustics, Speech and Signal Processing - Proceedings). https://doi.org/10.1109/ICASSP.2013.6637896