Abstract
Standard geometric model fifitting methods take as an input a fifixed set of feature pairs greedily matched based only on their appearances. Inadvertently, many valid matches are discarded due to repetitive texture or large baseline between view points. To address this problem, matching should consider both feature appearances and geometric fifitting errors. We jointly solve feature matching and multi-model fifitting problems by optimizing one energy. The formulation is based on our generalization of the assignment problem and its effificient mincost-max-flflow solver. Our approach signifificantly increases the number of correctly matched features, improves the accuracy of fifitted models, and is robust to larger baselines.