首页 > 论文 > 激光与光电子学进展 > 57卷 > 20期(pp:201101--1)

GGCN:基于GPU的高光谱图像分类算法

GGCN: GPU-Based Hyperspectral Image Classification Algorithm

  • 摘要
  • 论文信息
  • 参考文献
  • 被引情况
  • PDF全文
分享:

摘要

高光谱图像分类是遥感领域的研究热点之一,是对地观测的重要手段,在地物的精细识别等领域具有重要的应用。使用卷积神经网络(CNN)可以有效地从原始图像中提取高级特征,具有较高的分类精度。但CNN计算量巨大,对硬件要求较高。为了提高模型计算效率,可以在图形处理器(GPU)上进行CNN模型的训练。现有的并行算法,比如GCN(GPU based Cube-CNN),无法充分利用GPU的并行能力,算法加速效果并不理想。为了进一步提升算法效率,提出基于通用矩阵乘法(GEMM)算法的GGCN(GPU based Cube-CNN improved by GEMM)并行加速算法,通过G-PNPE(GEMM based Parallel Neighbor Pixels Extraction)对输入数据和卷积核进行重新组织排列,实现卷积的并行计算,有效地提高了GPU的利用率并进一步提升了算法的训练效率。通过分析在三个数据集上的实验结果发现,改进算法的分类精度与原算法保持一致,而且模型的训练时间缩短了30%左右,表明算法的有效性和优越性。

Abstract

Hyperspectral image classification is one of the research hotspots in the field of remote sensing. It is an important means of earth observation and has important applications in areas such as fine identification of ground objects. The use of convolutional neural networks (CNN) can effectively extract advanced features from the original image with high classification accuracy. However, CNN has a huge amount of calculations and requires high-performance hardware. In order to improve the computational efficiency of the model, the CNN model can be trained on the GPU. Existing parallel algorithms such as GCN (GPU based Cube-CNN) cannot make full use of the parallel capabilities of the GPU, and the algorithm acceleration effect is not ideal. In order to further improve the efficiency of the algorithm, the GGCN (GPU based Cube-CNN improved by GEMM) parallel acceleration algorithm based on the general matrix multiply (GEMM) algorithm is proposed. G-PNPE(GEMM based Parallel Neighbor Pixels Extraction) reorganizes and arranges the input data and convolution kernel to achieve parallel calculation of convolution, which effectively improves the utilization of GPU and increases the training efficiency of the algorithm. By analyzing the experimental results on the three datasets, the classification accuracy of the improved algorithm is consistent with the original algorithm, and the training time of the CNN network is shortened by about 30%, which proves the effectiveness and superiority of the algorithm.

广告组1 - 空间光调制器+DMD
补充资料

中图分类号:TP751

DOI:10.3788/LOP57.201101

所属栏目:成像系统

基金项目:国家自然科学基金、上海市科委部分地方院校能力建设项目、上海市青年科技英才扬帆计划;

收稿日期:2019-12-16

修改稿日期:2020-02-25

网络出版日期:2020-10-01

作者单位    点击查看

张明华:上海海洋大学信息学院, 201306
邹亚晴:上海海洋大学信息学院, 201306
宋巍:上海海洋大学信息学院, 201306
黄冬梅:上海海洋大学信息学院, 201306上海电力大学电子与信息工程学院, 上海 200090
刘智翔:上海海洋大学信息学院, 201306

联系人作者:黄冬梅(dmhuang@shou.edu.cn)

备注:国家自然科学基金、上海市科委部分地方院校能力建设项目、上海市青年科技英才扬帆计划;

【1】Zhang B. Advancement of hyperspectral image processing and information extraction [J]. Journal of Remote Sensing. 2016, 20(5): 1062-1090.
张兵. 高光谱图像处理与信息提取前沿 [J]. 遥感学报. 2016, 20(5): 1062-1090.
Zhang B. Advancement of hyperspectral image processing and information extraction [J]. Journal of Remote Sensing. 2016, 20(5): 1062-1090.
张兵. 高光谱图像处理与信息提取前沿 [J]. 遥感学报. 2016, 20(5): 1062-1090.

【2】Bioucas-Dias J M, Plaza A, Camps-Valls G, et al. Hyperspectral remote sensing data analysis and future challenges [J]. IEEE Geoscience and Remote Sensing Magazine. 2013, 1(2): 6-36.

【3】Camps-Valls G, Bruzzone L. Kernel-based methods for hyperspectral image classification [J]. IEEE Transactions on Geoscience and Remote Sensing. 2005, 43(6): 1351-1362.

【4】Belgiu M, Dragu? L. Random forest in remote sensing: a review of applications and future directions [J]. ISPRS Journal of Photogrammetry and Remote Sensing. 2016, 114: 24-31.

【5】LeCun Y, Bengio Y, Hinton G. Deep learning [J]. Nature. 2015, 521(7553): 436-444.

【6】Yan M, Zhao H D, Li Y H, et al. Multi-classification and recognition of hyperspectral remote sensing objects based on convolutional neural network [J]. Laser & Optoelectronics Progress. 2019, 56(2): 021702.
闫苗, 赵红东, 李宇海, 等. 基于卷积神经网络的高光谱遥感地物多分类识别 [J]. 激光与光电子学进展. 2019, 56(2): 021702.

【7】Liu Y Z, Jiang Z Q, Ma F, et al. Hyperspectral image classification based on hypergraph and convolutional neural network [J]. Laser & Optoelectronics Progress. 2019, 56(11): 111007.
刘玉珍, 蒋政权, 马飞, 等. 基于超图和卷积神经网络的高光谱图像分类 [J]. 激光与光电子学进展. 2019, 56(11): 111007.

【8】Leng J B, Li T, Bai G, et al. Cube-CNN-. SVM: , 2016, 1027-1034.

【9】Li P, Guan G X, Wu T X, et al. GPU optimization of hyperspectral real-time anomaly detection based on Cholesky decomposition [J]. Transducer and Microsystem Technology. 2019, 38(3): 7-10.
李萍, 关桂霞, 吴太夏, 等. 基于Cholesky分解的高光谱实时异常探测的GPU优化 [J]. 传感器与微系统. 2019, 38(3): 7-10.

【10】Jiang X R, Wei W. Application of CUDA technology in digital image matching Modern Information Technology[J]. 0, 2019(18): 61-63.
姜雪茸, 魏威. CUDA技术在数字图像匹配中的应用 现代信息科技[J]. 0, 2019(18): 61-63.

【11】Shen T, Hu F. Acceleration of CNN on GPU Applications of IC[J]. 0, 2017(6): 18-22.
沈恬, 胡飞. 卷积神经网络在图形处理GPU芯片上的优化 集成电路应用[J]. 0, 2017(6): 18-22.

【12】Dong H, Li T, Leng J B, et al. GCN: , 2017, 41-49.

【13】Nakasato N. A fast GEMM implementation on the cypress GPU [J]. ACM SIGMETRICS Performance Evaluation Review. 2011, 38(4): 50-55.

引用该论文

Zhang Minghua,Zou Yaqing,Song Wei,Huang Dongmei,Liu Zhixiang. GGCN: GPU-Based Hyperspectral Image Classification Algorithm[J]. Laser & Optoelectronics Progress, 2020, 57(20): 201101

张明华,邹亚晴,宋巍,黄冬梅,刘智翔. GGCN:基于GPU的高光谱图像分类算法[J]. 激光与光电子学进展, 2020, 57(20): 201101

您的浏览器不支持PDF插件,请使用最新的(Chrome/Fire Fox等)浏览器.或者您还可以点击此处下载该论文PDF