TOLDI: An effective and robust approach for 3D local shape description

作者：

Highlights：

• A novel local reference frame (LRF) is proposed for local shape descriptors.

• A new feature representation called the triple orthogonal local depth images (TOLDI) is presented.

• Extensive experiments on the shape retrieval, object recognition and 3D matching datasets are carried out.

• Our proposal achieves the best overall performance against several existing methods.

摘要

Highlights•A novel local reference frame (LRF) is proposed for local shape descriptors.•A new feature representation called the triple orthogonal local depth images (TOLDI) is presented.•Extensive experiments on the shape retrieval, object recognition and 3D matching datasets are carried out.•Our proposal achieves the best overall performance against several existing methods.

论文关键词：Local reference frame,Local feature descriptor,Shape retrieval,Object recognition,3D registration

论文评审过程：Received 29 April 2016, Revised 8 September 2016, Accepted 22 November 2016, Available online 24 November 2016, Version of Record 4 January 2017.

论文官网地址：https://doi.org/10.1016/j.patcog.2016.11.019