Data-driven hair segmentation with isomorphic manifold inference

作者:

Highlights:

• Similar coarse hair probability maps should correspond to similar segmentations.

• Data-driven Isomorphic Manifold Inference is proposed to exploit the shape priors.

• We integrate the inferred shape, color and texture into a unified framework.

• Besides hair segmentation, we also validate our IMI on horse class segmentation.

• Experiments on hair and horse databases show impressive performances.

摘要

•Similar coarse hair probability maps should correspond to similar segmentations.•Data-driven Isomorphic Manifold Inference is proposed to exploit the shape priors.•We integrate the inferred shape, color and texture into a unified framework.•Besides hair segmentation, we also validate our IMI on horse class segmentation.•Experiments on hair and horse databases show impressive performances.

论文关键词:Hair segmentation,Data driven,Shape model,Isomorphic manifold inference

论文评审过程:Received 8 June 2013, Revised 28 November 2013, Accepted 25 February 2014, Available online 11 March 2014.

论文官网地址:https://doi.org/10.1016/j.imavis.2014.02.011