Median local ternary patterns optimized with rotation-invariant uniform-three mapping for noisy texture classification

作者：

Highlights：

• A noise-robust median sampling for median local ternary pattern (MLTP).

• Three MLTP descriptors: MLTP_C, MLTP_R, and MLTP_M, are defined.

• Rotation-invariant uniform three (riu3) mapping to optimize MLTP from 3P to 3P.

• Multi-scale joint feature classification framework of MLTP_CRM is developed.

摘要

•A noise-robust median sampling for median local ternary pattern (MLTP).•Three MLTP descriptors: MLTP_C, MLTP_R, and MLTP_M, are defined.•Rotation-invariant uniform three (riu3) mapping to optimize MLTP from 3P to 3P.•Multi-scale joint feature classification framework of MLTP_CRM is developed.

论文关键词：Noisy texture classification,Median local ternary pattern,Rotation-invariant uniform-three mapping,Multi-scale joint distribution

论文评审过程：Received 24 January 2017, Revised 10 August 2017, Accepted 11 February 2018, Available online 21 February 2018, Version of Record 28 February 2018.

论文官网地址：https://doi.org/10.1016/j.patcog.2018.02.009