Local ternary pattern based multi-directional guided mixed mask (MDGMM-LTP) for texture and material classification

作者：

Highlights：

• MDGMM-LTP: A new mixed LTP-and-LDP-like feature extraction method is proposed.

• A new mixed-directional kernel based on various edge extraction kernels is proposed.

• The novel descriptor is compared to several handcrafted and deep-learning techniques.

• The Wilcoxon signed-rank test confirmed the accuracy improvement of the new method.

摘要

•MDGMM-LTP: A new mixed LTP-and-LDP-like feature extraction method is proposed.•A new mixed-directional kernel based on various edge extraction kernels is proposed.•The novel descriptor is compared to several handcrafted and deep-learning techniques.•The Wilcoxon signed-rank test confirmed the accuracy improvement of the new method.

论文关键词：MDGMM-LTP,Wilcoxon signed rank test,K-Nearest Neighbors (KNN),Kirsch,Robinson,Prewitt and Frei–Chen kernels,LDP and LTP

论文评审过程：Received 29 May 2021, Revised 29 April 2022, Accepted 27 May 2022, Available online 3 June 2022, Version of Record 11 June 2022.

论文官网地址：https://doi.org/10.1016/j.eswa.2022.117646