Trajectory estimation and optimization through loop closure detection, using omnidirectional imaging and global-appearance descriptors
作者：
Highlights：
• A strategy is proposed to create visual models of unknown environments incrementally.
• The framework uses omnidirectional imaging and global appearance descriptors.
• The visual information is fused with the angle provided by the robot odometry.
• A loop-closure detection algorithm is included to correct and update the model.
• The relative distance between nodes is estimated through a multi-scale analysis.
摘要
•A strategy is proposed to create visual models of unknown environments incrementally.•The framework uses omnidirectional imaging and global appearance descriptors.•The visual information is fused with the angle provided by the robot odometry.•A loop-closure detection algorithm is included to correct and update the model.•The relative distance between nodes is estimated through a multi-scale analysis.
论文关键词：Catadioptric vision sensors,Global-appearance descriptors,Fourier Signature,Histogram of Oriented Gradients,Visual odometry,Loop closure,Route estimation
论文评审过程：Received 17 November 2017, Revised 6 February 2018, Accepted 28 February 2018, Available online 1 March 2018, Version of Record 19 March 2018.
论文官网地址：https://doi.org/10.1016/j.eswa.2018.02.042
