No-reference mesh visual quality assessment via ensemble of convolutional neural networks and compact multi-linear pooling

作者：

Highlights：

• The perceived quality of 3D meshes is influenced due to distortions caused by many geometric operations.

• A no-reference mesh visual quality assessment method is proposed to automatically estimate the perceived quality.

• Deep convolutional networks and compact multi-linear pooling is adopted in our method.

• Extensive experiments and comparisons with existing methods are conducted on mesh quality datasets.

摘要

•The perceived quality of 3D meshes is influenced due to distortions caused by many geometric operations.•A no-reference mesh visual quality assessment method is proposed to automatically estimate the perceived quality.•Deep convolutional networks and compact multi-linear pooling is adopted in our method.•Extensive experiments and comparisons with existing methods are conducted on mesh quality datasets.

论文关键词：Blind mesh quality assessment,Convolutional neural network,Fine-tuning,Compact multi-linear pooling,Visual saliency

论文评审过程：Received 17 June 2019, Revised 4 October 2019, Accepted 15 December 2019, Available online 25 December 2019, Version of Record 29 December 2019.

论文官网地址：https://doi.org/10.1016/j.patcog.2019.107174