Efficient Algorithms for Rotation Averaging Problems

The rotation averaging problem is a fundamental task in computer vision applications. It generally very difficult to solve due the nonconvex constraints. While sufficient optimality condition available literature, there lack of fast convergent algorithm achieve stationary points. In this paper, by exploring structure, we first propose block coordinate descent (BCD)-based with guaranteed convergence Afterwards, further an alternative applying successive upper-bound minimization (SUM) method. SUM-based can be implemented parallel and thus more suitable for addressing large-scale problems. Numerical examples verify that proposed algorithms have superior performance as compared state-of-the-art algorithm. Moreover, checking condition, find from extensive numerical experiments two globally optimal solutions.