可变类空间约束高斯混合模型遥感图像分割

doi:10.11959/j.issn.1000-436x.2017026

通信学报 ›› 2017, Vol. 38 ›› Issue (2): 34-43.doi: 10.11959/j.issn.1000-436x.2017026

可变类空间约束高斯混合模型遥感图像分割

赵泉华,石雪,王玉,李玉

辽宁工程技术大学测绘与地理科学学院遥感科学与应用研究所，辽宁阜新 123000

修回日期:2016-11-23 出版日期:2017-02-01 发布日期:2017-07-20
作者简介:赵泉华（1978-），女，河北承德人，博士，辽宁工程技术大学副教授、博士生导师，主要研究方向为随机几何、空间统计学、模糊集理论等在遥感图像建模、解译及其在海洋环境遥感中的应用。|石雪（1992-），女，辽宁阜新人，辽宁工程技术大学硕士生，主要研究方向为影像信息提取理论与方法。|王玉（1990-），女，辽宁本溪人，辽宁工程技术大学博士生，主要研究方向为图像处理。|李玉（1963-），男，吉林长春人，博士，辽宁工程技术大学教授、博士生导师，主要研究方向为遥感图像处理。
基金资助:
国家自然科学基金资助项目(41301479);国家自然科学基金资助项目(41271435);辽宁省自然科学基金资助项目(The Natural Science Founda-tion of Liaoning Province)

Remote sensing image segmentation based on spatially constrained Gaussian mixture model with unknown class number

Quan-hua ZHAO,Xue SHI,Yu WANG,Yu LI

Institute for Remote Sensing Science and Application,School of Geomatics,Liaoning Technical University,Fuxin 123000,China

Revised:2016-11-23 Online:2017-02-01 Published:2017-07-20
Supported by:
The National Natural Science Foundation of China(41301479);The National Natural Science Foundation of China(41271435)

摘要/Abstract

摘要：

针对传统高斯混合模型（GMM,Gaussian mixture model）难以自动获取类属数和对噪声敏感问题，提出了一种基于可变类空间约束GMM的遥感图像分割方法。首先在构建的GMM中，将像素类属性建模为马尔可夫随机场（MRF,Markov random field），并在此基础上定义其先验概率；结合邻域像素类属性的后验概率和先验概率，定义噪声平滑因子，以提高算法的抗噪性；在参数求解过程中，分别采用可逆跳变马尔可夫链蒙特卡罗（RJMCMC,reversible jump Markov chain Monte Carlo）方法和最大似然（ML,maximum likelihood）方法估计类属数和模型参数；最后以最小化噪声平滑因子为准则获取最终分割结果。为了验证提出的分割方法，分别对模拟图像和全色遥感图像进行了可变类分割实验。实验结果表明提出方法的可行性和有效性。

关键词: 高斯混合模型, 空间约束, 最大似然估计, 可逆跳变马尔可夫链蒙特卡罗, 遥感图像分割

Abstract:

In view of the traditional Gaussian mixture model (GMM),it was difficult to obtain the number of classes and sensitive to the noise.A remote sensing image segmentation method based on spatially constrained GMM with unknown number of classes was proposed.First,in the built GMM,prior probability that represented the membership between a pixel and one class was modeled as a Markov random field (MRF).In order to improve the sensitivity of noise,the smoothing factor was defined by combining the a posterior probability and the prior probability of neighboring pixels.For estimating the number of classes and the parameters of model,the reversible jump Markov chain Monte Carlo (RJMCMC) and maximum likelihood (ML) estimation were employed,respectively.Finally,by minimizing the smoothing factor the final segmentation was obtained.In order to verify the proposed segmentation method,the synthetic and real panchromatic images were tested.The experimental results show that the proposed method is feasible and effective.

Key words: Gaussian mixture model (GMM), spatially constrained, maximum likelihood (ML), reversible jump Markov chain Monte Carlo (RJMCMC), remote sensing image segmentation

中图分类号:

TP391.41

赵泉华,石雪,王玉,李玉. 可变类空间约束高斯混合模型遥感图像分割[J]. 通信学报, 2017, 38(2): 34-43.

Quan-hua ZHAO,Xue SHI,Yu WANG,Yu LI. Remote sensing image segmentation based on spatially constrained Gaussian mixture model with unknown class number[J]. Journal on Communications, 2017, 38(2): 34-43.

图/表 13

图1

表1

图2

图3

图4

表2

表3

图5

图6

图7

图8

图9

图10

参考文献 16

[1]	王春艳, 徐爱功, 杨本臣 . 基于区间二型模糊模型的高分辨率遥感影像分割方法[J]. 仪器仪表学报, 2016,37(3): 658-666.
	WANG C Y , XU A G , YANG B C . High resolution remote sensing image segmentation based on the interval type-2 fuzzy model[J]. Chinese Journal of Scientific Instrument, 2016,37(3): 658-666.
[2]	常生鹏, 马亿旿, 蔡立军 . 一种基于 Hadoop 的高分辨率遥感图像处理方法[J]. 计算机工程与应用, 2015,51(11): 167-171.
	CHANG S P , MA Y W , CAI L J . A kind of high resolution remote sensing image processing method based on Hadoop[J]. Computer Engineering and Applications, 2015,51(11): 167-171.
[3]	王露, 刘庆元 . 高分辨率遥感影像多尺度分割中最优尺度选取方法综述[J]. 测绘与空间地理信息, 2015,38(3): 166-169.
	WANG L , LIU Q Y . The methods summary of optimal segmentation scale selection in high-resolution remote sensing images multi-scale segmentation[J]. Geomatics ＆ Spatial Information Technology, 2015,38(3): 166-169.
[4]	孔凡壮, 蒋艳凰, 刘芸 . 遥感图像分割技术综述[C]// 2005通信理论与技术新进展——第十届全国青年通信学术会议. 2005: 790-792.
	KONG F Z , JIANG Y H , LIU Y . A survey on remote image segmentation methods[C]// 2005 New Progress in Communication Theory and Technology,The 10th National Youth Communication Conference Proceedings. 2005: 790-792.
[5]	赵泉华, 赵雪梅, 李玉 . 结合HMRF模型的模糊ISODATA高分辨率遥感图像分割[J]. 信号处理, 2016,32(2): 157-166.
	ZHAO Q H , ZHAO X M , LI Y . A fuzzy ISODATA approach combining hidden Markov random field model for high resolution remote sensing image segmentation[J]. Journal of Signal Processing, 2016,32(2): 157-166.
[6]	ZHAO Q H , LI Y , WANG Y . SAR image segmentation with unknown number of classes combined Voronoi tessellation and RJMCMC algorithm[J]. ISPRS Annals of Photogrammetry,Remote Sensing and Spatial Information Sciences, 2016,12(19): 119-124.
[7]	MCLACHLAN G J , PEEL D . Finite mixture models[M]. New York: John Wiley ＆ SonsPress, 2000: 7-35.
[8]	卢洁, 杨学志, 郎文辉 . 区域GMM聚类的SAR图像分割[J]. 中国图象图形学报, 2011,16(11): 2088-2094.
	LU J , YANG X Z , LANG W H . SAR image segmentation with re-gion-based GMM[J]. Journal of Image and Graphics, 2011,16(11): 2088-2094.
[9]	王荔霞, 谢维信, 裴继红 . 多高斯模型特征空间覆盖学习的海洋航摄图像分割[J]. 电子学报, 2014,42(10): 2117-2122.
	WANG L X , XIE W X , PEI J H . Aerial ocean image segmentation using multi-Gauss model characteristic space cover leaming[J]. Acta Electronica Sinica, 2014,42(10): 2117-2122.
[10]	SANJAY-GOPAL S , HEBERT T J . Bayesian pixel classification using spatially variant finite mixtures and the generalized EM algorithm[J]. IEEE Transactions on Image Processing, 1998,7(7): 1014-1028.
[11]	NIKOU C , GALATSANOS N P , LIKAS A C . A class-adaptive spatially variant mixture model for image segmentation[J]. IEEE Transactions on Image Processing, 2007,16(4): 1121-1130.
[12]	KATO Z . Segmentation of color images via reversible jump MCMC sampling[J]. Journal Image and Vision Computing, 2006,26(3): 361-371.
[13]	HOU Y , YANG Y , RAO N ,et al. Mixture model and Markov random field-based remote sensing image unsupervised clustering methed[J]. Opto Electronics Review, 2011,19(1): 83-88.
[14]	COMER M , BOUMAN C A , GRAEF M ,et al. Bayesian methods for image segmentation[J]. Journal of the Minerals,Metals and Materials Society, 2011,63(7): 55-57.
[15]	WU N . The maximum entropy method[M]. Springer Science ＆ Business Media,Berlin Heidelberg, 2012: 1-9.
[16]	CHATZIS S P , VARVARIGOU T A . A fuzzy clustering approach toward hidden Markov random field models for enhanced spatially constrained image segmentation[J]. IEEE Transactions on Fuzzy Systems, 2008,16(5): 1351-1361.

参数	I	II	III	IV	V
μ	50	100	120	150	200
σ	10	7	5	12	10

指标	I	II	III	IV	V
μ_e/e_μ	50.0%/0	100.4%/0.4%	120.4%/0.3%	150.3%/0.2%	199.9%/0.05%
σ_e/e_σ	9.9%/1%	7.0%/0	4.6%/8%	11.3%/6%	9.9%/1%

精度指标	I	II	III	IV	V
用户精度	100%	99.99%	99.66%	99.90%	99.92%
产品精度	100%	99.83%	99.94%	99.90%	99.70%
总精度			99.87%
Kappa			99.85%

可变类空间约束高斯混合模型遥感图像分割

Remote sensing image segmentation based on spatially constrained Gaussian mixture model with unknown class number

在线阅读

PDF下载

可视化

被引次数

摘要/Abstract

引用本文

使用本文

图/表 13

参考文献 16

相关文章 11

Metrics

推荐阅读 0

[1]	张小贝, 张权, 方明, 秦琨, 张琦. 宽带数字阵列雷达通道固定时延误差估计技术的研究[J]. 通信学报, 2023, 44(3): 24-32.
[2]	胡建伟,车欣,周漫,崔艳鹏. 基于高斯混合模型的增量聚类方法识别恶意软件家族[J]. 通信学报, 2019, 40(6): 148-159.
[3]	唐鑫,马兆丰,钮心忻,杨义先. 基于变分贝叶斯学习的音频水印盲检测方法[J]. 通信学报, 2015, 36(1): 121-128.
[4]	崔玮,吴成东,张云洲,贾子熙,程龙. 基于高斯混合模型的非视距定位算法[J]. 通信学报, 2014, 35(1): 99-106.
[5]	陈明生,梁光明,孙即祥,刘东华,赵键. 复杂背景下H.264压缩域运动目标检测算法[J]. 通信学报, 2011, 32(3): 91-97.
[6]	赵凌,冯镔,邱锦波. 基于自适应分块外观模型的视觉跟踪[J]. 通信学报, 2011, 32(10): 166-173.
[7]	张丙杰,胡捍英,王大鸣. 基于交织循环前缀的MIMO-OFDM系统多符号定时同步[J]. 通信学报, 2010, 31(8A): 184-187.
[8]	何业军,何牧,朱光喜,刘德明. 基于最大似然估计的OFDM信号精确定位判决[J]. 通信学报, 2008, 29(11A): 72-76.
[9]	王津申,戴跃伟,王执铨. 对QIM水印滤波攻击的估计[J]. 通信学报, 2007, 28(6): 99-104.
[10]	唐洪,邱天爽,张文蓉. 脉冲噪声环境中自适应阵列对信号的跟踪[J]. 通信学报, 2005, 26(8): 22-27.
[11]	陈晨,李建东,赵林靖. OFDM整数倍频偏ML估计方法的性能分析[J]. 通信学报, 2005, 26(4): 7-13.