A continuous linear optimal transport approach for pattern analysis in image datasets
作者：
Highlights：
• A continuous version of the LOT framework is described that bypasses many of the difficulties associated with the discrete formulation.
• Using continuous transport maps, a forward and inverse transform operation is defined for images.
• An improved reference/average image estimation algorithm is proposed.
• The range within which points in LOT space are invertible according to the continuous formulation is calculated.
• We show that our method significantly speeds up the computation of the LOT embedding.
摘要
Highlights•A continuous version of the LOT framework is described that bypasses many of the difficulties associated with the discrete formulation.•Using continuous transport maps, a forward and inverse transform operation is defined for images.•An improved reference/average image estimation algorithm is proposed.•The range within which points in LOT space are invertible according to the continuous formulation is calculated.•We show that our method significantly speeds up the computation of the LOT embedding.
论文关键词：Optimal transport,Linear embedding,Generative image modeling,Pattern visualization
论文评审过程：Received 3 March 2015, Revised 27 August 2015, Accepted 17 September 2015, Available online 30 September 2015, Version of Record 27 November 2015.
论文官网地址：https://doi.org/10.1016/j.patcog.2015.09.019
