Appearance based background subtraction for PTZ cameras
作者：
Highlights：
• An appearance based background subtraction algorithm is presented for pan-tilt-zoom cameras.
• The proposed method extracts features for each pixel by considering its neighborhood.
• Four type of features are considered: gradient, color, texture and intensity.
• A global foreground/background model is learnt without any spatial constraint.
• It outperforms the state-of-the-art on CDnet 2014 and Hopkins 155 datasets.
摘要
Highlights•An appearance based background subtraction algorithm is presented for pan-tilt-zoom cameras.•The proposed method extracts features for each pixel by considering its neighborhood.•Four type of features are considered: gradient, color, texture and intensity.•A global foreground/background model is learnt without any spatial constraint.•It outperforms the state-of-the-art on CDnet 2014 and Hopkins 155 datasets.
论文关键词：Background subtraction,Foreground detection,Image segmentation
论文评审过程：Received 21 April 2016, Revised 28 July 2016, Accepted 28 July 2016, Available online 31 July 2016, Version of Record 18 August 2016.
论文官网地址：https://doi.org/10.1016/j.image.2016.07.008
