GORFLM: Globally Optimal Robust Fitting for Linear Model
作者：
Highlights：
• The robust linear model fitting method uses BnB to guarantee the global optimality.
• A negative Gaussian function over the residuals is used to formulate the problem.
• We derive a convex quadratic function to assess the lower bound of objective function.
• The use of soft loss function better addresses the cases with different noise levels.
• The efficiency of GORFLM is stable with the increasing number of points.
摘要
•The robust linear model fitting method uses BnB to guarantee the global optimality.•A negative Gaussian function over the residuals is used to formulate the problem.•We derive a convex quadratic function to assess the lower bound of objective function.•The use of soft loss function better addresses the cases with different noise levels.•The efficiency of GORFLM is stable with the increasing number of points.
论文关键词：Robust linear model fitting,Globally optimization,Branch and bound,Gaussian function
论文评审过程：Received 12 September 2019, Revised 16 December 2019, Accepted 15 March 2020, Available online 20 March 2020, Version of Record 26 March 2020.
论文官网地址：https://doi.org/10.1016/j.image.2020.115834
