CPU和GPU协同的多光谱影像快速波段配准方法
|
摘要
随着遥感影像数据量的飞速增长,传统的串行波段配准方法已无法满足大数据多光谱影像的实时配准需求。针对该问题,提出了一种CPU和GPU协同的多光谱影像快速波段配准方法。首先进行计算量和并行度分析,将同名点匹配和微分纠正映射至GPU执行,仿射变换系数拟合仍驻留在CPU执行。其次通过核函数任务映射和基本设置,使算法步骤在GPU上可执行,并设计了3种性能优化方法(访存优化、指令优化、传输计算堆叠),进一步提高了波段配准的执行效率。在NVIDIA Tesla M2050 GPU和Intel Xeon E5650 CPU组成的实验平台上,对遥感26号卫星多光谱影像的实验表明,使用该方法加速后的波段配准执行时间仅为3.25 s,与传统串行方法相比,加速比达到了32.32倍,可以满足大数据多光谱影像的近实时配准需求。
With the rapid increase of data size of remote sensing images,the traditional serial band re-gistration method cannot meet the demand for real-time processing of big-data multispectral images. Therefore,a CPU/GPU cooperative fast band registration method for multispectral imagery is proposed in this paper. Firstly,the computational amount and degree of parallelism are analyzed; point matching and differential rectification are ported to GPU to execute while the affine transformation parameter is still calculated on CPU. Secondly,kernel task assignment and basic settings are made to ensure the two above GPU steps executable. Moreover,three performance optimization methods,including memory access optimization,instruction optimization and transmission/computation overlap,are designed to further improve the efficiency of band registration. The experimental results based on NVIDIA Tesla M2050 GPU and Intel Xeon E5650 CPU show that the running time of YG-26 multispectral image band registration is only 3.25 s with our method,which got a speedup ratio of 32.32 compared with the traditional CPU serial method. The proposed method can provide quasi-real-time processing capability for multispectral imagery with big data size.
With the rapid increase of data size of remote sensing images,the traditional serial band re-gistration method cannot meet the demand for real-time processing of big-data multispectral images. Therefore,a CPU/GPU cooperative fast band registration method for multispectral imagery is proposed in this paper. Firstly,the computational amount and degree of parallelism are analyzed; point matching and differential rectification are ported to GPU to execute while the affine transformation parameter is still calculated on CPU. Secondly,kernel task assignment and basic settings are made to ensure the two above GPU steps executable. Moreover,three performance optimization methods,including memory access optimization,instruction optimization and transmission/computation overlap,are designed to further improve the efficiency of band registration. The experimental results based on NVIDIA Tesla M2050 GPU and Intel Xeon E5650 CPU show that the running time of YG-26 multispectral image band registration is only 3.25 s with our method,which got a speedup ratio of 32.32 compared with the traditional CPU serial method. The proposed method can provide quasi-real-time processing capability for multispectral imagery with big data size.
引文
[1]Yu Haiyang,Gan Fuping,Dang Fuxing.An Experimental Analysis of Band to Band Registration Error in High Resolution Satellite Remote Sensing Imagery[J].Remote Sensing for Land&Resources,2007,19(3):39-42(于海洋,甘甫平,党福星.高分辨率遥感影像波段配准误差试验分析[J].国土资源遥感,2007,19(3):39-42)
[2]Yang Jingyu,Zhang Yongsheng,Li Zhengguo,et al.GPU-CPU Cooperate Processing of RS Image OrthoRectification[J].Geomatics and Information Science of Wuhan University,2011,36(9):1 043-1 046(杨靖宇,张永生,李正国,等.遥感影像正射纠正的GPUCPU协同处理研究[J].武汉大学学报·信息科学版,2011,36(9):1 043-1 046)
[3]Fang L Y,Wang M,Li D R,et al.MOC-Based Parallel Preprocessing of ZY-3 Satellite Images[J].IEEE Geoscience and Remote Sensing Letter,2015,12(2):419-423
[4]Hu X Y,Li X K,Zhang Y J.Fast Filtering of LiDAR Point Cloud in Urban Areas Based on Scan Line Segmentation and GPU Acceleration[J].IEEE Geoscience and Remote Sensing Letter,2013,19(2):308-312
[5]Sui H G,Peng F F,Xu C,et al.GPU-Accelerated MRF Segmentation Algorithm for SAR Images[J].Computers&Geosciences,2012,43(2):159-166
[6]Cheng Boyan,Liu Qiang,Li Xiaowen,et al.Parallel Rasterization of Vector Polygon Based on CUDA[J].Bulletin of Surveying and Mapping,2014(11):97-101(程博艳,刘强,李小文,等.利用CUDA实现矢量地图栅格化的并行处理[J].测绘通报,2014(11):97-101)
[7]Chen Xi,Qiu Yuehong,Yi Hongwei.Parallel Programming Design of Star Image Registration Based on GPU[J].Infrared and Laser Engineering,2014,43(11):3 756-3 761(陈茜,邱跃洪,易红伟.基于GPU的星图配准算法并行程序设计[J].红外与激光工程,2014,43(11):3 756-3 761)
[8]Zhu Zhichao.Research on GPU-Based Multi-resolution Infrared and Visible Image Registration[D].Nanjing:Nanjing University of Aeronautics and Astronautics,2011(朱智超.基于GPU的多分辨率红外与可见光图像配准研究[D].南京:南京航空航天大学,2011)
[9]Zhou Haifang,Zhao Jin.Parallel Programming Design and Storage Optimization of Remote Sensing Image Registration Based on GPU[J].Journal of Computer Research and Development,2012,49(1):281-286(周海芳,赵进.基于GPU的遥感图像配准并行程序设计与存储优化[J].计算机研究与发展,2012,49(1):281-286)
[10]Xu Rulin.Study of Parallel Algorithms for Remote Sensing Image Registration Based on GPU and Implement of Application System[D].Changsha:National University of Defense Technology,2014(徐如林.基于GPU的遥感图像配准并行算法研究及应用系统实现[D].长沙:国防科技大学,2014)
[11]Qiu Deyuan.GPGPU Programming Technique:From GLSL,CUDA to Open GL[M].Beijing:China Machine Press,2011(仇德元.GPGPU编程技术:从GLSL、CUDA到Open CL[M].北京:机械工业出版社,2011)
[12]NVIDIA.CUDA C Programming Guide,V5.0[S].Santa Clara:NVIDIA Corporation,2012
[13]NVIDIA.CUDA C Best Practices Guide,V5.0[S].Santa Clara:NVIDIA Corporation,2012
[14]Kirk D,Hwu W M.Programming Massively Parallel Processors[M].2nd ed.Massachusetts:Morgan Kaufmann Publishers,2012
[15]NVIDIA.NVIDIA’s White Paper of Precision&Performance:Floating Point and IEEE 754 Comp-liance for NVIDIA GPUs[S].Santa Clara:NVIDIA Corporation,2012
[2]Yang Jingyu,Zhang Yongsheng,Li Zhengguo,et al.GPU-CPU Cooperate Processing of RS Image OrthoRectification[J].Geomatics and Information Science of Wuhan University,2011,36(9):1 043-1 046(杨靖宇,张永生,李正国,等.遥感影像正射纠正的GPUCPU协同处理研究[J].武汉大学学报·信息科学版,2011,36(9):1 043-1 046)
[3]Fang L Y,Wang M,Li D R,et al.MOC-Based Parallel Preprocessing of ZY-3 Satellite Images[J].IEEE Geoscience and Remote Sensing Letter,2015,12(2):419-423
[4]Hu X Y,Li X K,Zhang Y J.Fast Filtering of LiDAR Point Cloud in Urban Areas Based on Scan Line Segmentation and GPU Acceleration[J].IEEE Geoscience and Remote Sensing Letter,2013,19(2):308-312
[5]Sui H G,Peng F F,Xu C,et al.GPU-Accelerated MRF Segmentation Algorithm for SAR Images[J].Computers&Geosciences,2012,43(2):159-166
[6]Cheng Boyan,Liu Qiang,Li Xiaowen,et al.Parallel Rasterization of Vector Polygon Based on CUDA[J].Bulletin of Surveying and Mapping,2014(11):97-101(程博艳,刘强,李小文,等.利用CUDA实现矢量地图栅格化的并行处理[J].测绘通报,2014(11):97-101)
[7]Chen Xi,Qiu Yuehong,Yi Hongwei.Parallel Programming Design of Star Image Registration Based on GPU[J].Infrared and Laser Engineering,2014,43(11):3 756-3 761(陈茜,邱跃洪,易红伟.基于GPU的星图配准算法并行程序设计[J].红外与激光工程,2014,43(11):3 756-3 761)
[8]Zhu Zhichao.Research on GPU-Based Multi-resolution Infrared and Visible Image Registration[D].Nanjing:Nanjing University of Aeronautics and Astronautics,2011(朱智超.基于GPU的多分辨率红外与可见光图像配准研究[D].南京:南京航空航天大学,2011)
[9]Zhou Haifang,Zhao Jin.Parallel Programming Design and Storage Optimization of Remote Sensing Image Registration Based on GPU[J].Journal of Computer Research and Development,2012,49(1):281-286(周海芳,赵进.基于GPU的遥感图像配准并行程序设计与存储优化[J].计算机研究与发展,2012,49(1):281-286)
[10]Xu Rulin.Study of Parallel Algorithms for Remote Sensing Image Registration Based on GPU and Implement of Application System[D].Changsha:National University of Defense Technology,2014(徐如林.基于GPU的遥感图像配准并行算法研究及应用系统实现[D].长沙:国防科技大学,2014)
[11]Qiu Deyuan.GPGPU Programming Technique:From GLSL,CUDA to Open GL[M].Beijing:China Machine Press,2011(仇德元.GPGPU编程技术:从GLSL、CUDA到Open CL[M].北京:机械工业出版社,2011)
[12]NVIDIA.CUDA C Programming Guide,V5.0[S].Santa Clara:NVIDIA Corporation,2012
[13]NVIDIA.CUDA C Best Practices Guide,V5.0[S].Santa Clara:NVIDIA Corporation,2012
[14]Kirk D,Hwu W M.Programming Massively Parallel Processors[M].2nd ed.Massachusetts:Morgan Kaufmann Publishers,2012
[15]NVIDIA.NVIDIA’s White Paper of Precision&Performance:Floating Point and IEEE 754 Comp-liance for NVIDIA GPUs[S].Santa Clara:NVIDIA Corporation,2012