Video saliency detection by gestalt theory

作者：

Highlights：

• Color, luminance, and texture features are extracted to compute spatial saliency maps.

• Motion features are estimated to calculate temporal saliency maps.

• A entropy weighting method is designed to fuse spatial and temporal saliency maps.

• The Gestalt principle is used to estimate spatial and temporal uncertainty.

摘要

•Color, luminance, and texture features are extracted to compute spatial saliency maps.•Motion features are estimated to calculate temporal saliency maps.•A entropy weighting method is designed to fuse spatial and temporal saliency maps.•The Gestalt principle is used to estimate spatial and temporal uncertainty.

论文关键词：Visual attention,Video saliency detection,Gestalt theory,Uncertainty weighting,Spatiotemporal saliency

论文评审过程：Received 4 September 2018, Revised 21 July 2019, Accepted 31 July 2019, Available online 2 August 2019, Version of Record 7 August 2019.

论文官网地址：https://doi.org/10.1016/j.patcog.2019.106987