Feature matching using sequential evaluation on sample consensus

Consensus evaluation is one of the key issues in vision based feature matching. To improve both efficiency and accuracy of local feature matching, we proposed a Sequential Evaluation on Sample Consensus (SESAC). The addressed approach first sorts the matching pairs in terms of the similarity of the corresponding features, then it sequentially selects the samples, and uses the least squares method to fit the model, by which to reject the outlier points and update the optimal solution. The experimental results demonstrate that compared with the classic PROSAC and RANSAC algorithm, SESAC algorithm can achieve similar accuracy, whereas reduce the running time greatly.