一种分块的立体像对相对定向方法研究
On a Method for Relative Orientation based on Block
-
摘要: 针对立体像对相对定向时错误匹配点难以剔除的问题，提出了一种分块的立体像对相对定向方法。该方法先对影像上的特征点进行分块，基于各个数据块间的抽样一致性原则选择分布合理的特征点以计算基础矩阵的初始值。然后，以点到极线的距离最小化为判断标准对特征点集合进行筛选，从而获得误差较小的“内点”集。最后，采用选权迭代的方法计算获得基础矩阵的各个要素值，并将其转换为相对定向元素，从而完成立体像对的相对定向。近景影像和航空影像的对比实验表明，该方法具有更高的定向精度和鲁棒性。
-
关键词:
- 摄影测量、立体像对、相对定向、基础矩阵
Abstract: In order tobeagainst the abnormal data in the process of calculating the relative orientation prob-lem,a new method based on the block has beenproposed.Firstly,the feature points are set into blocks. And then the reasonable feature points are selected to calculate the initial value as the consistency between each block of data.After that,the point to the baseline is made as optimization criteria to obtain a more accurate set of matching feature points.And finally,Interior points are used to obtain the result value.Ex-perimental results of close-range image and aerial images show that the proposed method is feasible and have a greater suppression effect on the errors caused by the mismatching feature points. -
-
计量
- 文章访问数: 480
- HTML全文浏览数: 328
- PDF下载数: 0
- 施引文献: 0
下载: