【摘要】 现有的视觉定位系统中大多使用传统的欧几里得距离排序和门限阈值的方法或基于KD-Tree的局部特征匹配算法进行局部特征的匹配。本文使用SURF特征作为视觉定位的图像局部特征,提出一种基于RBF的局部特征匹配算法。通过对比已有的局部特征匹配算法与基于RBF的局部特征匹配算法在不同数据集上的表现,综合分析各种算法在精确度、时延上的指标,得出基于RBF的局部特征匹配算法是一种有效的局部特征匹配算法的结论。更多还原

【Abstract】 In existing visual localization systems, the conventional Euclidean distance ordering with threshold or KDTreebased local feature matching algorithm is commonly used to match local features. In this paper, the Speeded Up Robust Feature(SURF) used as the image local feature of visual localization, a local feature matching algorithm based on RBF is proposed. By comparing the performance of existing local feature matching algorithms with the proposed algorithm in different data sets, the metrics including accuracy and delay of different algorithms are comprehensively analyzed with the conclusion that the proposed local feature matching algorithm based on RBF is effective.更多还原