改进贝叶斯ARTMAP的迁移学习遥感影像分类算法

doi:10.3969/j.issn.0372-2112.2016.09.033

电子学报 ›› 2016, Vol. 44 ›› Issue (9): 2248-2253.DOI: 10.3969/j.issn.0372-2112.2016.09.033

改进贝叶斯ARTMAP的迁移学习遥感影像分类算法

韩敏, 杨雪

大连理工大学电子信息与电气工程学部, 辽宁大连 116023

收稿日期:2015-01-14 修回日期:2015-04-29 出版日期:2016-09-25
- 通讯作者:
- 韩敏
- 作者简介:
- 杨雪女,1990年2月出生于河北秦皇岛.现为大连理工大学电子信息与电气工程学部硕士研究生.主要研究方向为遥感影像信息提取.
- 基金资助:
- 国家重点基础研究发展规划 (973计划）项目 (No.2013CB430403）; 国家自然科学基金项目 (No.61374154）

Transfer Learning Using Improved Bayesian ARTMAP for Remote Sensing Image Classification

HAN Min, YANG Xue

Faculty of Electronic Information and Electrical Engineering, Dalian University of Technology, Dalian, Liaoning 116023, China

Received:2015-01-14 Revised:2015-04-29 Online:2016-09-25 Published:2016-09-25

摘要/Abstract

摘要：

遥感分类旨在从图像光谱中提取资源环境监测可用的地理信息，然而基于模式分类的图像处理技术受光谱漂移影响而缺乏历史样本重复利用的有效策略，制约着有限目标样本下遥感分类精度的提高.针对该问题，本文构建了基于改进的贝叶斯ARTMAP神经网络的迁移学习遥感影像分类算法，通过提高谐振匹配性来抑制类别扩散，利用节点的离散增量期望最大化参数更新策略，将历史遥感样本中的地物分类先验信息迁移到目标模型当中.实验结果表明本文方法能有效利用历史遥感数据弥补缺少目标训练数据的不足，相比于其他样本利用策略大幅提高遥感影像分类精度.

关键词: 遥感分类, 信息提取, 迁移学习, ARTMAP, 神经网络

Abstract:

Remote sensing classification aims at extracting available geographic information from image spectrum for resources and environment monitoring,but due to the spectral drift effect,the lack of effective strategies on historical sample reuse for image processing technology based on pattern classification restricts remote sensing classification accuracy with limited target samples.To solve this problem,this paper proposes a transfer learning algorithm for remote sensing classification using improved Bayesian ARTMAP neural network.More productive resonance matching is used to suppress the unattractive property of category proliferation,so that the incremental expectation maximization can be introduced to update parameters adaptively.The classification prior knowledge of the historical samples is transferred to the target model.The experimental results show that this method can effectively compensate for the lack of target training data by reusing the historical samples and significantly improve the accuracy of remote sensing image classification compared with other sample utilization strategy.

Key words: remote sensing classification, information extraction, transfer learning, ARTMAP, neural network

中图分类号:

TP751
TP181

韩敏, 杨雪. 改进贝叶斯ARTMAP的迁移学习遥感影像分类算法[J]. 电子学报, 2016, 44(9): 2248-2253.

HAN Min, YANG Xue. Transfer Learning Using Improved Bayesian ARTMAP for Remote Sensing Image Classification[J]. Acta Electronica Sinica, 2016, 44(9): 2248-2253.

参考文献

[1] 唐晓亮,韩敏.基于改进型模糊ARTMAP网络的CBR遥感图像分类系统研究[J].中国图象图形学报,2007,12(6):1033-1040. TANG Xiao-liang,HAN Min.A study of CBR system for remote sensing classification based on modified fuzzy ARTMAP neural network[J].Journal of Image and Graphics,2007,12(6):1033-1040.(in Chinese)
[2] DAUMÉ H,MARCU D.Domain adaptation for statistical classifiers[J].Journal of Artificial Intelligence Research,2011,26:101-126.
[3] PAN S J,YANG Q.A survey on transfer learning[J].IEEE Transactions on Knowledge and Data Engineering,2010,22(10):1345-1359.
[4] 王皓,高阳,陈兴国.强化学习中的迁移:方法和进展[J].电子学报,2008,36(12):39-43. WANG Hao,CHEN Xing-guo.Transfer of reinforcement learning:the state of the art[J].Acta Electronica Sinica,2008,36(12):39-43.(in Chinese)
[5] 于重重,田蕊,谭励,等.非平衡样本分类的集成迁移学习算法[J].电子学报,2012,40(7):1358-1363. YU Chong-chong,TIAN Rui,TAN Li and Xu-yan.Integrated transfer learning algorithmic for unbalanced samples classification[J].Acta Electronica Sinica,2012,40(7):1358-1363.(in Chinese)
[6] PAN S J,TSANG I W,et al.Domain adaptation via transfer component analysis[J].IEEE Transactions on Neural Networks,2011,22(2):199-210.
[7] 倪彤光,王士同,应文豪,邓赵红.迁移组概率学习机[J].电子学报,2013,41(11):2207-2215. NI Tong-guang,WANG Shi-tong,YING Wen-hao,DENG Zhao-hong.Transfer group probabilities based learning machine[J].Acta Electronica Sinica,2013,41(11):2207-2215.(in Chinese)
[8] BRUZZONE L,MARCONCINI M.Toward the automatic updating of land-cover maps by a domain-adaptation SVM classifier and a circular validation strategy[J].IEEE Transactions on Geoscience and Remote Sensing,2009,47(4):1108-1122.
[9] ZHANG Y S,ZHENG X W,et al.Semi-supervised manifold learning based multigraph fusion for high-resolution remote sensing image classification[J].IEEE Geoscience and Remote Sensing Letters,2014,(11):464-468.
[10] 吴田军,骆剑承,夏列钢,等.迁移学习支持下的遥感影像对象级分类样本自动选择方法[J].测绘学报,2014,43(9):908-916. WU Tianjun,LUO Jiancheng,XIA Liegang,et al.An automatic sample collection method for object-oriented classification of remotely sensed imageries based on transfer learning[J].Acta Geodaetica et Cartographica Sinica,2014,43(9):908-916.(in Chinese)
[11] LIU Y,LI X.Domain adaptation for land use classification:A spatio-temporal knowledge reusing method[J].ISPRS Journal of Photogrammetry and Remote Sensing,2014,98:133-144.
[12] VIGDOR B,LERNER B.The Bayesian ARTMAP[J].IEEE Transactions on Neural Networks,2007,18(6):1628-1644.
[13] TANG X,HAN M.Semi-supervised Bayesian ARTMAP[J].Applied Intelligence,2010,33(3):302-317.
[14] LEIVA-MURILLO J M,GÓMEZ-CHOVA L,et al.Multitask remote sensing data classification[J].IEEE Transactions on Geoscience and Remote Sensing,2013,51(1):151-161.
[15] NG S K,MCLACHLAN G J.On the choice of the number of blocks with the incremental EM algorithm for the fitting of normal mixtures[J].Statistics and Computing,2003,13(1):45-55.
[16] CARPENTER G A,GROSSBERG S,et al.Fuzzy ARTMAP:A neural network architecture for incremental supervised learning of analog multidimensional maps[J].IEEE Transactions on Neural Networks,1992,3(5):698-713.
[17] CARPENTER G A,ROSS W D.ART-EMAP:A neural network architecture for object recognition by evidence accumulation[J].IEEE Transactions on Neural Networks,1995,6(4):805-818.
[18] COETZEE F M.Correcting the Kullback-Leibler distance for feature selection[J].Pattern Recognition Letters,2005,26(11):1675-1683.

改进贝叶斯ARTMAP的迁移学习遥感影像分类算法

Transfer Learning Using Improved Bayesian ARTMAP for Remote Sensing Image Classification

PDF

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

[1]	吴靖, 叶晓晶, 黄峰, 陈丽琼, 王志锋, 刘文犀. 基于深度学习的单帧图像超分辨率重建综述[J]. 电子学报, 2022, 50(9): 2265-2294.
[2]	毛国君, 王者浩, 黄山, 王翔. 基于剪边策略的图残差卷积深层网络模型[J]. 电子学报, 2022, 50(9): 2205-2214.
[3]	袁海英, 曾智勇, 成君鹏. 面向灵活并行度的稀疏卷积神经网络加速器[J]. 电子学报, 2022, 50(8): 1811-1818.
[4]	徐兴荣, 刘聪, 李婷, 郭娜, 任崇广, 曾庆田. 基于双向准循环神经网络和注意力机制的业务流程剩余时间预测方法[J]. 电子学报, 2022, 50(8): 1975-1984.
[5]	邵志文, 周勇, 谭鑫, 马利庄, 刘兵, 姚睿. 基于深度学习的表情动作单元识别综述[J]. 电子学报, 2022, 50(8): 2003-2017.
[6]	王相海, 赵晓阳, 王鑫莹, 赵克云, 宋传鸣. 非抽取小波边缘学习深度残差网络的单幅图像超分辨率重建[J]. 电子学报, 2022, 50(7): 1753-1765.
[7]	张志文, 刘天歌, 聂鹏举. 基于实景数据增强和双路径融合网络的实时街景语义分割算法[J]. 电子学报, 2022, 50(7): 1609-1620.
[8]	吴晓晓, 李刚强, 张胜利. 分布式协作频谱感知网络中恶意节点检测和定位方法研究[J]. 电子学报, 2022, 50(6): 1370-1380.
[9]	张波, 陆云杰, 秦东明, 邹国建. 一种卷积自编码深度学习的空气污染多站点联合预测模型[J]. 电子学报, 2022, 50(6): 1410-1427.
[10]	丁毅, 沈薇, 李海生, 钟琼慧, 田明宇, 李洁. 面向CNN的区块链可信隐私服务计算模型[J]. 电子学报, 2022, 50(6): 1399-1409.
[11]	张云, 化青龙, 姜义成, 徐丹. 基于混合型复数域卷积神经网络的三维转动舰船目标识别[J]. 电子学报, 2022, 50(5): 1042-1049.
[12]	冀振燕, 韩梦豪, 宋晓军, 冯其波. 面向激光光条图像修复的循环相似度映射网络[J]. 电子学报, 2022, 50(5): 1234-1242.
[13]	廖勇, 李玉杰. 一种轻量化低复杂度的FDD大规模MIMO系统CSI反馈方法[J]. 电子学报, 2022, 50(5): 1211-1217.
[14]	袁冠, 邴睿, 刘肖, 代伟, 张艳梅, 蔡卓. 基于时空图神经网络的手势识别[J]. 电子学报, 2022, 50(4): 921-931.
[15]	王延达, 陈炜通, 皮德常, 岳琳. 一种自适应记忆神经网络多跳读取与覆盖度机制结合的药物推荐模型[J]. 电子学报, 2022, 50(4): 943-953.