摘要
为解决图像拼接技术中特征点冗余的问题,该论文提出一种基于SIFT(尺度不变特征变换)的改进算法,该算法在标准的SIFT算法基础上加上筛选规则,通过结合特征点与欧氏距离的关系自动优选特征点,根据欧氏距离设立距离等值区间,统计区间特征点数量,运用双阈值(特征点阈值、距离筛选阈值)选择较好的匹配点后再用RANSAC算法消除误匹配点。利用客观图像质量评估指标分析,实验表明,改进后的算法减少了特征点的数量,提高拼接图像质量。
In order to solve the problem about redundant feature point in image splicing technology, this paper proposes an im-proved method based on Scale Invariant Feature Transform SIFT （SIFT） algorithm. The algorithm added filtering rules, by com-bining the relationship between feature points and Euclidean distance automatically selecting the feature points. According to Euclidean distance set up the equivalent distance interval, then count the number of interval statistical feature points, using dual- threshold （feature point threshold, range filter threshold） choose matching points, last eliminate false matching points with RAN-SAC algorithm t. Using the objective image quality evaluation index analysis, the experimental results show that the improved algorithm reduces the number of feature points and improves the quality o f mosaic images.
出处
《信息通信》
2017年第10期100-102,共3页
Information & Communications
