论文部分内容阅读
针对非共线多CCD遥感图像匹配点的分布特点,该文提出一种基于聚类的误匹配点去除方法。首先,根据匹配点的沿轨方向偏移量曲线,获取匹配点的多维特征向量。然后,对匹配点集进行聚类处理,将所有点聚为一个类簇,最后根据簇半径序列曲线的变化趋势区分正确点和误匹配点。通过天绘1号02星全色遥感图像的实验和处理,结果表明在误匹配点去除和正确匹配点保留方面所提算法与其它方法相比具有更好性能。
In view of the distribution characteristics of matching point of non-collinear multi-CCD remote sensing images, this paper proposes a clustering-based method to remove mismatch points. First, according to the curve of the offset of the matching point along the track, the multi-dimensional feature vector of the matching point is obtained. Then, the matching point set is clustered, all the points are grouped into a cluster, and finally the correct point and the mismatch point are distinguished according to the change trend of the curve of the cluster radius. The experiment and processing of the full-color remote sensing imagery of Dayu-1 No. 1 satellite show that the proposed algorithm has better performance compared with other methods in the removal of mismatched points and the preservation of correct matching points.