基于流水线的增强型植被指数快速提取算法  

Fast extraction algorithm of enhanced vegetation index based on pipeline

作　　者：沈夏炯[1,2] 侯柏成 韩道军[1,2] 马瑞 Shen Xiajiong;Hou Bocheng;Han Daojun;Ma Rui(School of Computer&Information Engineering,Henan University,Kaifeng Henan 475004,China;Institute of Data&Knowledge Engineering,Henan University,Kaifeng Henan 475004,China;School of Information&Science,Zhengzhou Normal College,Zhengzhou 450044,China)

机构地区：[1]河南大学计算机与信息工程学院,河南开封475004 [2]河南大学数据与知识工程研究所,河南开封475004 [3]郑州师范学院信息科学与技术学院,郑州450044

出　　处：《计算机应用研究》2018年第9期2827-2830,共4页Application Research of Computers

基　　金：国家自然科学基金资助项目(61272545);河南省科技攻关计划基金资助项目(142102210390);河南省教育厅科技攻关计划基金资助项目(14A520026);河南省博士后科研项目(2015036)

摘　　要：针对传统的并行方法难以对增强型植被指数(enhanced vegetation index,EVI)提取链进行加速的问题,结合EVI提取算法的特点,依靠流水线在遥感影像多级连续处理方面的优势,提出一种基于流水线模型的EVI快速提取算法。为了解决相邻处理阶段吞吐率的差异问题,在流水线相邻节点间加入双缓冲队列来存储节点处理的中间结果,进一步提高EVI提取算法各步骤并行度。实验结果表明,提出的EVI快速提取算法的提取效率要高于传统的EVI多线程提取算法。According to the question that difficult to accelerate the EVI extraction chain for traditional parallelism methods,combined with the characteristics of EVI extraction algorithm,depended on the advantages of pipeline in multi-level continuous processing of remote sensing images,this paper proposed an EVI fast extraction algorithm based on pipeline model.In order to solve the problem of the difference in throughput between adjacent processing stages,it added a double buffer queue between adjacent nodes of the pipeline to store the intermediate result of the node processing,and further improved the parallelism of the steps of EVI extraction algorithm.Experimental results show,the EVI fast extraction algorithm is more efficient than the traditional EVI multi-thread extraction algorithm.

关 键 词：流水线 遥感影像 增强型植被指数 双缓冲队列 

分 类 号：TP751[自动化与计算机技术—检测技术与自动化装置] TP301.6[自动化与计算机技术—控制科学与工程]