Efficient Globally Optimal Consensus Maximisation with Tree Search

摘要: Maximum consensus is one of the most popular criteria for robust estimation in computer vision. Despite its widespread use, optimising criterion still customarily done by randomised sample-and-test techniques, which do not guarantee optimality result. Several globally optimal algorithms exist, but they are too slow to challenge dominance methods. We aim change this state affairs proposing a very efficient algorithm global maximisation consensus. Under framework LP-type methods, we show how wide variety vision tasks can be posed as tree search problem. This insight leads novel based on A* search. propose heuristic and support set updating routines that enable rapidly find results. On common problems, our several orders magnitude faster than previous exact Our work identifies promising solution maximisation.