An Automatic Global Registration Algorithm Based on Wavelet and Its Parallel Implementation
-
摘要: 随着遥感技术的发展,遥感图像处理领域对自动图像配准技术的需求越来越迫切. 首先提出了点匹配和全局配准相结合的自动图像配准算法;从实现快速配准的途径出发,提 出了利用多分辨率小波缩小搜索空间及进行全局配准的自动算法;详细设计了逐级求精的搜 索策略,并比较总结了算法的特点;在此基础上,提出了两种可行的数据并行方案;最后在一 个小规模的机群系统上实现了上述串、并行算法,给出了客观的性能评价.实验结果表明文 中提出的算法达到了预期的目标,即针对多传感器、大数据量的遥感图像,在保证精度的前 提下,进行快速高效的自动配准.Abstract: Image registration is widely used in image processing. With the development of remote sensing technology, the demand of fast and automatic image registration becomes more urgent. Based on a thorough investigation on automatic registration methods for remote sensing image, a new automatic image registration algorithm is proposed by combining the point-to-point matching with the global registration. The focus is on fast registration, whereby a wavelet-based automatic global registration algorithm is designed, which relies on iteratively refining strategy to reduce the search space. The features of the algorithm are concluded after comparison with some other methods. In addition, two parallel computation schemes are proposed and implemented on a cluster system with 16 processors. The experiment results show that the proposed algorithms could achieve high computational efficiency with satisfactory registration accuracies for large remote sensing images.
计量
- 文章访问数: 3010
- HTML全文浏览量: 55
- PDF下载量: 1424
- 被引次数: 0