An iterative approach for obtaining multi-scale superpixels based on stochastic graph contraction operations

作者：

Highlights：

• Presents a stochastic multi-scale superpixels approach to over-segment images.

• Relies on graph contraction operations and a stochastic strategy.

• Uses a tree structure to represent the multi-scale superpixels.

• Experiments show that the IHSGC outperforms others recent superpixels methods.

摘要

•Presents a stochastic multi-scale superpixels approach to over-segment images.•Relies on graph contraction operations and a stochastic strategy.•Uses a tree structure to represent the multi-scale superpixels.•Experiments show that the IHSGC outperforms others recent superpixels methods.

论文关键词：Superpixels,Stochastic segmentation,Graph contraction,Multi-scale analysis

论文评审过程：Received 3 July 2017, Revised 31 January 2018, Accepted 17 February 2018, Available online 19 February 2018, Version of Record 19 March 2018.

论文官网地址：https://doi.org/10.1016/j.eswa.2018.02.027