Joint direct estimation of 3D geometry and 3D motion using spatio temporal gradients
作者：
Highlights：
• This method does not require optical flow computation, it works on normal flow.
• Normal flow is derived only from spatial and temporal gradients.
• Using normal flow prevents from accumulating error from assumptions.
• We propose a non-convex optimization and the positive depth constraint.
• We refine through linear optimization via the 3D structure estimation.
摘要
•This method does not require optical flow computation, it works on normal flow.•Normal flow is derived only from spatial and temporal gradients.•Using normal flow prevents from accumulating error from assumptions.•We propose a non-convex optimization and the positive depth constraint.•We refine through linear optimization via the 3D structure estimation.
论文关键词：3D motion,Egomotion,Structure from motion,Normal flow
论文评审过程：Received 13 May 2018, Revised 14 November 2020, Accepted 23 November 2020, Available online 28 November 2020, Version of Record 19 February 2021.
论文官网地址：https://doi.org/10.1016/j.patcog.2020.107759
