改进注意力机制的遥感地貌识别算法

doi:10.13474/j.cnki.11-2246.2020.0326

测绘通报 ›› 2020, Vol. 0 ›› Issue (10): 93-96,100.doi: 10.13474/j.cnki.11-2246.2020.0326

改进注意力机制的遥感地貌识别算法

张朕通¹, 单玉刚², 袁杰¹

1. 新疆大学电气工程学院, 新疆乌鲁木齐 830047;
2. 湖北文理学院教育学院, 湖北襄阳 441053

收稿日期:2019-12-27 修回日期:2020-03-16 出版日期:2020-10-25 发布日期:2020-10-29
通讯作者: 单玉刚。E-mail:32748873@qq.com E-mail:32748873@qq.com
作者简介:张朕通(1994-),男,硕士生,主要研究方向为模式识别,计算机视觉。E-mail:1580335323@qq.com
基金资助:
国家自然科学基金（61863033）；湖北省教育厅科学技术研究项目（B2016175）；湖北文理学院博士基金（2015B002）

Remote sensing landform recognition algorithm based on improved attention mechanism

ZHANG Zhentong¹, SHAN Yugang², YUAN Jie¹

1. School of Electrical Engineering, Xinjiang University, Urumqi 830047, China:;
2. School of Education, Hubei University of Arts and Science, Xiangyang 441053, China

Received:2019-12-27 Revised:2020-03-16 Online:2020-10-25 Published:2020-10-29

摘要/Abstract

摘要： 对遥感地貌进行识别，近年来一直是遥感图像应用领域的研究热点。使用深度学习算法识别遥感影像具有比传统方法更高的准确率和稳健性。针对遥感影像中目标复杂度高、特征信息多等问题，本文提出了一种基于改进注意力机制的遥感图像识别算法，即将并联注意力机制（CS）和神经网络模型相结合，借助弱监督学习来辅助训练。同时采用双损失函数来缓解数据过拟合问题。试验结果表明，本文模型总精度为98.35%，Kappa系数达0.95，优于其他深度学习算法，能有效地识别出自然地貌。

关键词: 遥感识别, 深度学习, 并联注意力机制, 弱监督训练, 双损失函数

Abstract: Recognizing remote sensing landforms has been a research hotspot in the field of remote sensing image applications in recent years. Using deep learning algorithms to identify remote sensing images has higher accuracy and robustness than traditional methods. Aimed at the problems of high target complexity and feature information in remote sensing images. This paper proposes a remote sensing image recognition algorithm based on an improved attention mechanism. A parallel attention mechanism (CS) and neural network model are combined to improve training with weakly supervised learning. At the same time, a double loss function is used to alleviate the problem of data overfitting. The experimental results show that the total accuracy of the model in this paper is 98.35%, and the Kappa coefficient is 0.95, which is better than other deep learning algorithms and can effectively recognize natural landforms.

Key words: remote sensing recognition, deep learning, parallel attention mechanism, weakly supervised training, double loss function

中图分类号:

P237

张朕通, 单玉刚, 袁杰. 改进注意力机制的遥感地貌识别算法[J]. 测绘通报, 2020, 0(10): 93-96,100.

ZHANG Zhentong, SHAN Yugang, YUAN Jie. Remote sensing landform recognition algorithm based on improved attention mechanism[J]. Bulletin of Surveying and Mapping, 2020, 0(10): 93-96,100.

参考文献

[1] CHERIYADAT A M. Unsupervised feature learning for aerial scene classification[J].IEEE Transactions on Geoscience and Remote Sensing, 2014,52(1):439-451.
[2] ZHU Q G, ZHONG Y F, ZHANG L D, et al. Scene classification based on the fully sparse semantic topic model[J].IEEE Transactions on Geoscience and Remote Sensing,2017,55(10):5525-5538.
[3] 郭亚宁,林伟,潘泉,等.基于推广流形学习的高分辨遥感影像目标分类[J].自动化学报,2019,45(4):720-729.
[4] HU F,XIA G, WANG S, et al. Unsupervised feature learning via spectral clustering of multidimensional patches for remotely sensed scene classification[J].IEEE Journal of Selected Topics in Applied Earth Observations and Remote Sensing,2015,8(5):2015-2030.
[5] ZHANG F, DU B, ZHANG L P. Saliency-guided unsuper-vised feature learning for scene classification[J].IEEE Transactions on Geoscience and Remote Sensing,2015,53(4):2175-2184.
[6] ZOU Q, NI L H, ZHANG T, et al. Deep learning based feature selection for remote sensing scene classification[J]. IEEE Geoscience and Remote Sensing Letters, 2015,12(11):2321-2325.
[7] 谢梦,刘伟,杨梦圆,等.深度卷积神经网络支持下的遥感影像飞机检测[J].测绘通报,2019(6):19-23.
[8] 王华斌,韩旻,王光辉,等.遥感影像要素提取的可变结构卷积神经网络方法[J].测绘学报,2019,48(5):583-596.
[9] WOO S, PARK J, LEE J Y, et al. CBAM: convolutional block attention module[C]//Proceedings of the European Conference on Computer Vision (ECCV). Munich: [s.n.], 2018:3-19.
[10] XIE S N, GIRSHICK R, DOLLAR P, et al. Aggregated residual transformations for deep neural networks[C]//2017 IEEE Conference on Computer Vision and Pattern Recognition (CVPR). Hawaii: IEEE, 2017:5987-5995.
[11] HE K M, ZHANG X Y, REN S Q, et al. Deep residual learning for image recognition[C]//2016 IEEE Conference on Computer Vision and Pattern Recognition (CVPR). Las Vegas: IEEE,2016:770-778.
[12] MAHAJAN D, GIRSHICK R, RAMANATHAN V, et al. Exploring the limits of weakly supervised pretraining[C]//Proceedings of the European Conference on Computer Vision (ECCV). Munich: Springer,2018:185-201.
[13] 鲁晨光.Shanno公式改造[J].通信学报,1991,12(2):95-96.
[14] SZEGEDY C, VANHOUCKE V, IOFFE S, et al. Rethinking the inception architecture for computer vision[C]//2016 IEEE Conference on Computer Vision and Pattern Recognition (CVPR). Las Vegas: IEEE,2016:2818-2826.
[15] CHENG G, HAN J W, LU X Q. Remote sensing image scene classification: benchmark and state of the art[J]. Proceedings of the IEEE, 2017,105(10):1865-1883.

改进注意力机制的遥感地貌识别算法

Remote sensing landform recognition algorithm based on improved attention mechanism

可视化

摘要/Abstract

引用本文

使用本文

参考文献

相关文章 15

编辑推荐

Metrics

本文评价

[1]	刘立, 董先敏, 刘娟, 文学虎. 人机融合智能的遥感解译生产新方法[J]. 测绘通报, 2022, 0(7): 118-123,137.
[2]	蔡柔丹. 一种基于用户异步轨迹的身份识别智能方法[J]. 测绘通报, 2022, 0(7): 158-162,167.
[3]	滕文鑫, 张建辰, 邵杰. 集成时空邻近与卷积网络车道级高精度定位算法[J]. 测绘通报, 2022, 0(6): 1-5,61.
[4]	张晋赫, 秦育罗, 张在岩, 宋伟东, 朱洪波. 复杂场景下农村道路裂缝分割方法[J]. 测绘通报, 2022, 0(5): 74-78,88.
[5]	师芸, 石龙龙, 牛敏杰, 赵侃. 基于单阶段实例分割网络的黄土滑坡多任务自动识别[J]. 测绘通报, 2022, 0(4): 26-31.
[6]	王雪, 梁珂, 隋立春, 钟棉卿, 朱剑锋. 膨胀卷积与金字塔表达的深度学习模型用于农村建筑物提取[J]. 测绘通报, 2022, 0(4): 61-65.
[7]	高鸣, 周鑫鑫, 刘琦, 杨光迪, 吴长彬. 深度学习网络支持下的农房侵占耕地自动化监测[J]. 测绘通报, 2022, 0(3): 47-53.
[8]	孙琴琴, 蔡国印, 杨柳忠, 张宁, 杜明义. 城市尺度典型地表要素综合提取方法研究[J]. 测绘通报, 2022, 0(2): 62-66.
[9]	张自强, 刘涛, 杜萍, 锁旭宏, 杨国林. 图残差神经网络支持下的建筑物群组模式分类[J]. 测绘通报, 2022, 0(1): 1-7.
[10]	车子杰, 高飞, 吴兆福, 李振轩. 基于改进U-Net网络的多源遥感影像洪涝灾害信息提取与变化分析[J]. 测绘通报, 2022, 0(1): 26-32.
[11]	程瑞, 高建, 邢强, 孙中昶. 一种优化的Faster R-CNN小目标检测方法[J]. 测绘通报, 2021, 0(9): 21-27.
[12]	淳锦, 张新长, 郭海京, 张建国, 金诗程. 基于多源信息与深度学习特征提取的人口空间抽样方法[J]. 测绘通报, 2021, 0(8): 42-47.
[13]	李梓豪, 唐超. 基于高清工业相机的盾构隧道裂缝智能识别算法分析[J]. 测绘通报, 2021, 0(8): 83-87,101.
[14]	吴永静, 吴锦超, 林超, 窦宝成, 黎珂. 基于深度学习的高分辨率遥感影像光伏用地提取[J]. 测绘通报, 2021, 0(5): 96-101.
[15]	梁哲恒, 邓鹏, 姜福泉, 盛森, 魏汝兰, 谢刚生. 基于卷积神经网络的无人机影像违章建筑检测应用[J]. 测绘通报, 2021, 0(4): 111-115.