云环境下时序遥感影像的快速缓存切片方法

自然灾害与地表环境实时动态监测对短周期内接收影像的快速缓存切片提出了很高的要求。利用云计算并行编程模式Map-Reduce聚集计算资源,提出了一种在时序影像持续抵达情况下的快速缓存切片方法。围绕Map-Reduce本地化计算特性,该方法在任务分配中利用数据动态划分机制及基于空间相邻的上载机制对缓存切片算法进行优化加速,以满足环境动态监测对瓦片式缓存的及时性需求。实验证明,该方法在大数据量情况下较同类方法具有更好的扩展性能和加速性能。

Rapid Imagery Tile Generation for Remotely Sensed Time-seriesData in the Cloud Environment

Imagery tile caching is becoming an important technique for online Earth observation information sharing .Dynamic disaster and environment monitoring places great pressure on cached tiles for the generation of imagery acquired in real time or recent time. A rapid imagery tile generation approach for time-series imagery is proposed to update a Web map service based on the computing power provided by Map-Reduce，a popular parallel computing paradigm on the cloud. Based on the data locality of Map-Reduce，a strategy for dynamic data partitioning is used to reduce redundant work while spatial distribution and physical storage consistentcy is maintained to improve the data locality. These optimizations aim to provide in a timely way cached tiles for dynamic environment monitoring.Performance testing indicates that this approach is more efficient and scalable than existing similar methods.