遥感图像自动配准的串行与并行策略研究  

A Research on Serial and Parallel Strategies of the Automatic Image Registration for Remote Sensing

作　　者：周海芳[1] 刘光明[1] 郑明玲[1] 杨学军[1] 

机构地区：[1]国防科技大学计算机学院,湖南长沙410073

出　　处：《国防科技大学学报》2004年第2期56-61,共6页Journal of National University of Defense Technology

基　　金：国家杰出青年科学基金项目(69825104);军口863基金项目(2002AA714021);国家863基金项目(2002AA1Z201)

摘　　要：面对数据量呈指数增长的全球遥感图像,研究快速、有效、高精度的自动图像配准算法成为遥感领域迫切需要解决的问题。综述了遥感图像自动配准算法的研究现状和关键技术,且从新的角度分类总结了已有算法的特点及适用情况,并给出了综合的解决方案;同时研究并提出了三种新的并行策略,获得了较好的实验结论。由于算法采用粗粒度的数据并行方法,因此可扩展性和可移植性较好,无论在理论上还是在实践上都能为实际的遥感应用提供有益的指导和借鉴。最后提出了有待进一步研究的问题。The amounts of remote sensing images of global coverage will grow exponentially. To study and achieve fast and effective automatic registration of these digital data with high accuracies has become a critical problem in remote sensing. The status of the research on automatic registration of remote sensing images is presented firstly, and some relevant key techniques are discussed. And then, the existing registration algorithms are classified and analyzed from the point of a novel view, and an integrative solution is given. At the same time, three new parallel strategies are proposed with the good experiment results. Because the proposed algorithms are based on the data parallel model of coarse grain, their salability and portability are good, which can be a valuable reference for practical applications in remote sensing. Finally, the problems and challenges of the future research are pointed out.

关 键 词：遥感 自动图像配准 点匹配 全局配准 数据并行 

分 类 号：TP391.41[自动化与计算机技术—计算机应用技术] TP75[自动化与计算机技术—计算机科学与技术]