基于AKAZE特征提取与融合匹配算法的研究

Research on a Matching Strategy Based on AKAZE Feature Extraction and Fusion Matching Algorithm

图像匹配算法是很多计算机视觉应用的重要组成部分,其中算法的匹配准确度和匹配损耗时间是衡量匹配算法性能的重要指标。针对目前的图像特征匹配算法无法同时获得高匹配准确度和低匹配损耗时间的特点,提出了基于AKAZE算法特征提取与融合匹配算法相结合的匹配策略。通过AKAZE算法计算待匹配图像的特征描述子,融合匹配算法首先利用KD-Tree进行粗匹配,然后再结合PROSAC算法分两个阶段剔除误匹配特征点。通过试验测试,在获得高的匹配准确度的同时,保持低的匹配损耗时间。

Image matching algorithm is an important component of many computer vision applications,whereby the matching accuracy and match loss time of the algorithm are major indicators for the measurement of the performance of the matching algorithm.Considering that current image feature matching algorithm cannot have high matching accuracy and low matching loss time at the same time,a matching strategy based on AKAZE algorithm feature extraction and fusion matching algorithm was proposed in this paper.The AKAZE algorithm was used to calculate the feature descriptors of the image to be matched.According to the fusion matching algorithm,firstly,the KD-tree was used to complete rough matching.Then,the PROSAC algorithm was used to eliminate mismatched feature points in two stages.Experimental results showed that the algorithm proposed in this paper could not only achieve high matching accuracy,but also maintain a low matching loss time.