Enhanced pose normalization and matching of non-rigid objects based on support vector machine modelling

作者：

Highlights：

• We improve non-rigid surface correspondence by robust computing of the canonical pose.

• The relative translation and scale are computed by discarding shape outliers by SVMs.

• The improved surface correspondence in turn improves content-based retrieval.

• The improved pose normalization can be employed with any 3D shape descriptor.

摘要

Highlights•We improve non-rigid surface correspondence by robust computing of the canonical pose.•The relative translation and scale are computed by discarding shape outliers by SVMs.•The improved surface correspondence in turn improves content-based retrieval.•The improved pose normalization can be employed with any 3D shape descriptor.

论文关键词：Non-rigid analysis,Pose normalization,Support vector machines,3D shape matching

论文评审过程：Received 17 July 2011, Revised 21 November 2012, Accepted 23 June 2013, Available online 2 July 2013.

论文官网地址：https://doi.org/10.1016/j.patcog.2013.06.024