Discriminative shared transform learning for sketch to image matching
作者：
Highlights：
• Proposed Discriminative Shared Transform Learning for sketch to image matching.
• Learns a shared transform for cross-domain data under supervised constraints.
• Two models are proposed: Contractive model (C-model) and Divergent model (D-model).
• D-model models intra and inter-class variation whereas C-model models only the former.
• Experiments on seven datasets demonstrate the effectiveness of the DSTL algorithm.
摘要
•Proposed Discriminative Shared Transform Learning for sketch to image matching.•Learns a shared transform for cross-domain data under supervised constraints.•Two models are proposed: Contractive model (C-model) and Divergent model (D-model).•D-model models intra and inter-class variation whereas C-model models only the former.•Experiments on seven datasets demonstrate the effectiveness of the DSTL algorithm.
论文关键词：Face recognition,Sketch to digital image matching,sketch based image retrieval,Caricature face recognition,Transform learning
论文评审过程：Received 20 June 2020, Revised 30 December 2020, Accepted 31 December 2020, Available online 12 January 2021, Version of Record 12 February 2021.
论文官网地址：https://doi.org/10.1016/j.patcog.2021.107815
