強化学習に基づくファジィLMedS アルゴリズムのサンプリング行動戦略

抄録

The computer vision involves many modeling problems for preventing noise caused by sensing units such as cameras. In order to improve computer vision system performance, a robust modeling technique must be developed for essential models in the system. The RANSAC and least median of squares (LMedS) algorithm have been widely applied in such issues. However, the performance deteriorates as the noise ratio increases and the modeling time for algorithms tends to increase in industrial applications. As a promising technique, we proposed a new fuzzy LMedS method based on reinforcement learning concept for robust modeling. In this study, we investigated selecting strategies as an indispensable concept of reinforcement learning through camera homography experiments. Their results found the proposed ε-roulette strategy to be promising in improving calculation time, model optimality, and robustness in modeling performance.