收稿日期: 2016-11-27
修回日期: 2016-12-30
网络出版日期: 2017-11-30
基金资助
国家重点研发计划基金（No.2016YFB0502200）；国家自然科学基金（No.41127901）资助
Parallel Algorithm of UAV Image Match Considering Spatial Contiguity
Received date: 2016-11-27
Revised date: 2016-12-30
Online published: 2017-11-30
为解决无人机（unmaned aerial vehicle，UAV）影像匹配并行算法的负载均衡和特征数据传输问题，提出一种顾及空间邻接关系的影像匹配并行算法，旨在不降低原有匹配精度的前提下，最大限度地提高影像匹配效率.在特征提取阶段，根据影像空间邻接关系完成各对应节点的初始任务划分，随后进行稍细粒度的二次划分，以确定最终的特征提取任务.在任务调度时，根据计算节点的状态优先分配对应节点的任务.对应节点的任务全部分配完毕再分配其他节点对应的任务.在匹配阶段，首先根据特征提取任务划分匹配任务，然后根据特征提取任务的节点编号划定对应于每一节点的匹配任务单元，使用相同的任务调度方法完成整个测区的影像匹配.对一套有1 463幅UAV影像的典型数据进行实验，结果表明该算法不但能实现并行系统的负载平衡，还能减小特征数据的传输量，从而显著提高影像匹配效率.
张春森, 仇振国, 郭丙轩, 肖雄武, 朱师欢 . 顾及空间邻接关系的无人机影像匹配并行算法[J]. 应用科学学报, 2017 , 35(6) : 775 -785 . DOI: 10.3969/j.issn.0255-8297.2017.06.011
To deal with the problem of load balancing and feature data transmission in parallel algorithms of unmaned aerial vehicle (UAV) image match, aparallel algorithm for image match considering spatial contiguity is proposed. In the feature extraction phase, initial task partition is carried out according to spatial contiguity between images. The final feature extraction task is determined by performing a fine-grain second partition based on the initial partition. In task scheduling, corresponding tasks are assigned according to the computational node state, after which other tasks are assigned. In the image matching phase, matching tasks are assigned first according to the feature extraction tasks. Matching task unit of each node is confirmed by the node number of the feature extraction task. The same method is applied to image match in the whole measured area. Experiments on a typical data set including 1 463 UAV images show that the algorithm can realize load balancing of the parallel system and reduce the amount of feature data transmission,thus significantly improving efficiency of image match.
[1] 张剑清,柯涛,孙明伟. 基于集群计算机的海量航空数码影像并行处理-并行计算在航空数字摄影测量中的应用[J]. 计算机工程与应用,2008,44(13):12-15. Zhang J Q, Ke T,Sun M W. Parallel processing of mass aerial digital images based on cluster computer-the application of parallel computing in aerial digital photogrammetry[J]. Computer Engineering and Applications, 2008, 44(13):12-15. (in Chinese)
[2] 邹亚未,刘玉红. 基于多核CPU的无人机影像连接点并行匹配方法[J]. 测绘通报, 2014(3):32-36. Zou Y W, Liu Y H. A parallel matching method of tie points on UAV image based on multicore CPU[J]. Bulletin of Surveying and Mapping, 2014(3):32-36. (in Chinese)
[3] 肖雄武,郭丙轩,李德仁. 一种具有仿射不变性的倾斜影像快速匹配方法[J]. 测绘学报, 2015, 44(4):414-421.Xiao X W, Guo B X, Li D R. A quick and affine invariance matching method for oblique images[J]. Acta Geodaetica et Cartographica Sinica, 2015, 44(4):414-421. (in Chinese)
[4] Xiao X W, Guo B X, Li D R. Multi-view stereo matching based on self-adaptive patch and image grouping for multiple unmanned aerial vehicle imagery[J]. Remote Sensing, 2016, 8(2):89.
[5] 柯涛,张永军. SIFT特征算子在低空遥感影像自动匹配中的应用[J]. 测绘科学,2012, 34(4):23-36. Ke T, Zhang Y J. SIFT feature operator in the application of low altitude remote sensing image automatic matching[J]. Science of Surveying and Mapping, 2012, 34(4):23-36. (in Chinese)
[6] 王恺,刘民选,艾海滨. 基于GPU的快速影像匹配[J]. 测绘科学, 2014, 39(2):129-132. Wang K, Liu M X, Ai H B. Fast image matching based on GPU[J]. Science of Surveying and Mapping, 2014, 39(2):129-132. (in Chinese)
[7] 李劲澎,龚志辉,张婷. 无人机影像SIFT特征匹配的集群并行处理方法[J]. 测绘科学技术学报,2012, 29(6):440-444. Li J P, Gong Z H, Zhang T. Cluster paraallel processing method of UAV images matching based on SIFI feature[J]. Journal of Geomatics Science and Technology, 2012, 29(6):440-444. (in Chinese)
[8] 张丽,汤晓涛,戴晨光. 多视匹配方法的计算任务分析及其GPU并行实现[J]. 测绘科学技术学报, 2013, 30(5):480-483. Zhang L, Tang X T, Dai C G. Computing task analysis of muiti-view matching method and GPU parallel implementation[J]. Journal of Geomatics Science and Technology,2013, 30(5):480-483. (in Chinese)
[9] 杨靖宇. 摄影测量数据GPU并行处理若干关键技术研究[D]. 郑州:解放军信息工程大学,2010.
[10] 史凤丽. 基于集群计算机的图像并行处理[D]. 西安:西安科技大学, 2010.
[11] 郭静,祝永志,王延玲. 基于MPI的动态负载平衡算法的研究[J]. 计算机技术与发展,2009, 19(5):150-153. Guo J, Zhu Y Z, Wang Y L. Research on dynamic load balancing algorithm based on MPI[J]. Computer Technology and Development, 2009, 19(5):150-153. (in Chinese)
[12] 卢云娥,黄宗宇,李超阳. 基于微机集群系统的MPI并行计算[J]. 电子设计工程,2011, 19(5):78-81. Lu Y E, Huang Z Y,Li C Y. The MPI parallel computing based on the microcomputer cluster[J]. International Electronic Elements, 2011, 19(5):78-81. (in Chinese)
[13] 张少辉,马骏,张先哲. 一种MPI并行程序设计中的动态负载平衡策略[J]. 电脑开发与应用, 2008, 21(12):63-65. Zhang S H, Ma J, Zhang X Z. Dynamic load-balancing strategy in MPI parallel programming[J]. China Academic Journal Publishing House, 2008, 21(12):63-65. (in Chinese)
[14] 张剑清,潘励,王树根. 摄影测量学[M]. 第2版. 武汉:武汉大学出版社,2009:152.
[15] 柯涛,田一涵,段延松. 基于GPU的SIFT特征点快速匹配[J]. 测绘与空间地理信息, 2014, 37(12):14-16. Ke T,Tian Y H,Duan Y S. The fast matching of SIFT feature points based on GPU[J]. Geomatics and Spatial Information Technology,2014,37(12):14-16. (in Chinese)
[16] 刘振英,方滨兴,胡铭曾. 一种有效的动态负载平衡方法[J]. 软件学报,2001,12(4):563-569. Liu Z Y, Fang B X, Hu M Z. An effective method of dynamic load balancing[J]. Journal of Software, 2001, 12(4):563-569. (in Chinese)
[17] Penmatsa S, Chronopoulosa T. Game-theoretic static load balancing for distributed systems[J]. Journal of Parallel and Distributed Computing, 2011, 71(4):537-555,
[18] Randlesm, Lamb D, Taleb-Bendiaba. A comparative study into distributed load balancing algorithms for cloud computing[C]//IEEE International Conference on Advanced Information Networking and Applications Workshops, Perth, Australia, 2010:551-556.
[19] Mehta M A. Agrawal S, Jinwala D C. Novel algorithms for load balancing using hybrid approach in distributed systems[C]//IEEE International Conference on Parallel Distributed and Grid Computing, Solan, India, 2012:27-32
/
| 〈 |
|
〉 |
