Demo program for the paper:
This demo is implemented in MATLAB, and tested on a Ubuntu Machine with MATLAB R2018a.
The authors highly appreciate any bug reports or comments on the algorithm, which can be sent to: [email protected]
Run demo.m to start the demo. Two applications are provided:
- Robust linear fitting
- Robust homography estimation with quasi-convex residuals
This paper revisits the application of M-estimators for a spectrum of robust estimation problems in computer vision, particularly with the maximum consensus criterion. Cur- rent practice makes use of smooth robust loss functions, e.g. Huber loss, which enables M-estimators to be tackled by such well-known optimization techniques as Iteratively Re-weighted Least Square (IRLS). When consensus maximization is used as loss func- tion for M-estimators, however, the optimization problem becomes non-smooth. Our paper proposes an approach to resolve this issue. Based on the Alternating Direction Method of Multiplier (ADMM) technique, we develop a deterministic algorithm that is provably convergent, which enables the maximum consensus problem to be solved in the context of M-estimator. We further show that our algorithm outperforms other differen- tiable robust loss functions that are currently used by many practitioners. Notably, the proposed method allows the sub-problems to be solved efficiently in parallel, thus entails it to be implemented in distributed settings