基于纹理特征与LSSVM的青土湖地物提取

doi:10.12118/j.issn.1000-6060.2018.04.16

摘要/Abstract

摘要： 纹理特征作为一种非光谱信息能够增强地物之间的特征差异，这对于高分辨率遥感影像的地物提取有着重要意义。以青土湖为研究区，以Worldview-2影像为数据源，通过引入权重因子定义联合概率函数来确定最佳窗口尺度，利用灰度共生矩阵提取最佳窗口尺度下的纹理特征，将其与原始遥感影像合成，采用最小二乘支持向量机（LSSVM）进行地物提取，将提取结果与仅利用光谱信息的支持向量机（SVM）提取结果、辅以纹理特征的SVM提取结果对比分析。结果表明：此方法可以更加快速准确地提取青土湖地物，精度高达85.86%，优于仅利用光谱信息的SVM的65.13%，辅以纹理特征的SVM的73.45%，可为地物破碎的干旱区高分辨率遥感影像地物提取提供有益借鉴。

关键词: Worldview-2影像, 纹理特征, 权重, 最小二乘支持向量机, 青土湖

Abstract: In this paper,the study area Qingtu Lake is located at Minqin County,Gansu Province,China.Based on Worldview-2 remote sensing images and the field measured data,using texture features and LSSVM method,the object information of Qingtu Lake in arid area were extracted.The texture features as a non spectral information can enhance the difference between the object features,which has important meaning for object extraction from high resolution remote sensing image,but the key to texture feature extraction is to choose the optimal window scale,which directly influences the accuracy of subsequent extraction.Therefore this paper introduced a weighting factor to define joint probability function to determine the optimal window scale,adopted gray level co-occurrence matrix to extract texture features at the optimal window scale,and then synthesized with the original remote sensing image.Because LSSVM is a good solution to the robustness,sparsity and large-scale computing problems of SVM,the paper adopted the LSSVM method to extract the object information of the Qingtu Lake,and compared the results with that of the feature extraction only using the spectral information of SVM and the extraction of SVM combined with texture feature.The results of comparison show that this method of LSSVM can more quickly and accurately extract the object information of the Qingtu Lake,the total precision was 85.86%,Kappa was 0.822 5,better than SVM only using the spectral information with the total precision 65.13%,and SVM combined with texture features with 73.45%.In general,the LSSVM could be more useful to extract information of the fragmented objects in arid area from high resolution remote sensing image.

Key words: Worldview-2 image, texture feature, weight, LSSVM, Qingtu Lake

中图分类号:

TP79

张华, 王敏. 基于纹理特征与LSSVM的青土湖地物提取[J]. 干旱区地理, 2018, 41(4): 802-808.

ZHANG Hua, WANG Min. Feature extraction of Qingtu Lake based on texture feature and Least Squares Support Vector Machine[J]. 干旱区地理, 2018, 41(4): 802-808.

参考文献

[1] 胡玉福,邓良基,匡先辉,等.基于纹理特征的高分辨率遥感图像土地利用分类研究[J].地理与地理信息科学,2011,27(5):42-45.[HU Yufu,DENG Liangji,KUANG Xianhui,et al.Study on land use classification of high resolution remote sensing image based on texture feature[J].Geography and Geo-Information Science,2011,27(5):42-45.]
[2] HACKCLFORD A K,DAWS C H.A combined fuzzy pixel-based and object-based approach for classification of high-resolution multispectral data over urban[J].IEEE Transactions Geo-Science and Remote Sensing,2003,41(10):2354-2363.
[3] GLADA S,DE GROEVE T,EHDICH D,et al.Information extraction from very high-resolution satellite imagery over Lukole refugee camp,Tanzania[J].International Journal of Remote Sensing,2003,24(22):4251-4260.
[4] 黄惠萍.面向对象影像分析中的尺度问题的研究[D].北京:中国科学院遥感应用研究所,2003.[HUANG Huiping.Scale issues in object-oriented image analysis[D].Beijing:Institute of Remote Sensing Applications Chinese Academy of Sciences,2003.]
[5] GAMANYA R,DE MAEYER,PHILIPPE,et al.Precision change detection:Based on knowledge-based and object-oriented satellite image analysis in central Zimbabwe[C].ASPR Annual Conference,2006:97.
[6] 刘伟,赵庆展,汪传建,等.基于最小二乘支持向量机的无人机遥感影像分类[J].江苏农业科学,2017,45(9):187-191.[LIU Wei,ZNAO Qingzhan,WANG Chuanjian,et al.Remote sensing image classification of UAV based on least squares support vector machines[J].Jiangsu Agricultural Sciences,2017,45(9):187-191.]
[7] DIHKAN Mustafa,GUNEROGLU Nilgun,KARSLI Fevzi,et al.Remote sensing of tea plantations using an SVM classifier and pattern-based accuracy assessment technique[J].International Journal of Remote Sensing,2013,34(23):8549-8565.
[8] GALAL Omer,ONISIMO Mutanga,ABDEL-RAHMAN Elfatih M,et al.Performance of support vector machines and artificial neural network for mapping endangered tree species using Worldview-2 data in Dukuduku forest,south Africa[J].IEEE Journal of Selected Topics in Applied Earth Observations & Remote Sensing,2015,8(10):4825-4840.
[9] SUYKEN S J A K,VANDEWALLE J.Least squares support vector machine classifiers[J].Neural Processing Letter,1999,9:293-300.
[10] 牟丹,王祝文,黄玉龙,等.基于最小二乘支持向量机测井识别火山岩类型:以辽河盆地中基性火山岩为例[J].吉林大学学报:地球科学版,2015,45(4):639-648.[MOU Dan,WANG Zhuwen,HUANG Yulong,et al.Appliacation of least squares support vector machine to lithology identification:Taking intermediate/basaltic rocks of Liaohe as an example[J].Journal of Jinlin University:Earth Science Edition,2015,45(4):639-648.]
[11] 胡根生,吴问天,黄文江,等.粒子群优化的最小二乘支持向量机在小麦白粉病监测中的应用[J].遥感技术与应用,2017,32(2):299-304.[HU Gensheng,WU Wentian,HUANG Wenjiang,et al.Application of PSO-LSSVM in wheat powdery mildew monitoring[J].Remote Sensing Technology and Application,2017,32(2):299-304.]
[12] 杨佳佳,姜琦刚,陈永良,等.基于最小二乘支持向量机和高分辨率遥感影像的大尺度区域岩性划分[J].中国石油大学学报(自然科学版),2012,36(1):60-67.[YANG Jiajia,JIANG Qigang,CHEN Yongliang,et al.Lithology division for large-scale region segmentation based on LS-SVM and high resolution remote sensing[J].Journal of China University of Petroleum,2012,36(1):60-67.]
[13] 魏轩,周立华,马永欢,等.民勤绿洲50余年水利建设的生态经济影响[J].干旱区地理,2015,38(5):1014-1021.[WEI Xuan,ZHOU Lihua,MA Yonghuan,et al.Ecological and economic impacts of water conservancy construction in Minqin Oasis of more than 50 years[J].Arid Land Geography,2015,38(5):1014-1021.]
[14] 郭树江,杨自辉,王多泽,等.石羊河流域下游青土湖近地层风尘分布特征[J].干旱区地理,2016,39(6):1255-1262.[GUO Shujiang,YANG Zihui,WANG Duoze,et al.Dust distribution characteristics of Qingtu Lake surface layer at downstream Shiyang River[J].Arid Land Geography,2016,39(6):1255-1262.]
[15] 赵莹,王环,方圆.基于Worldview-2的遥感影像预处理[J].测绘与空间地理信息,2014,37(6):165-170.[ZHAO Ying,WANG Huan FANG Yuan.Remote sensing image preprocessing based on Worldview-2[J].Geomatics & Spatial Information Technology,2014,37(6):165-170.]
[16] HARALICK R M,SHANMUGAM K,DINSTEIN I H.Textural features for image classification[J].IEEE Transactions on Systems,Man and Cybernetics,1973,3(6):610-621.
[17] ZHANG Q,WANG J,GONG P,et al.Study of urban spatial patterns from SPOT panchromatic imagery using textural analysis[J].International Journal of Remote Sensing,2003,24(21):4137-4160.
[18] LU C S,CHUN P C,CHEN C F.Unsupervised texture segmentation via wavelet transform[J].Pattern Recognition,1997,30(5):729-742.
[19] INGLADA J.Automatic recognition of man-made objects in high resolution optical remote sensing images by SVM classification of geometric image features[J].ISPRS Journal of Photogrammetric and Remote Sensing,2007,62(3):236-248.
[20] SUYKANS J A K,DE BRABANTER J,LUKAS L,et al.Weighted least squares support vector machines:Robustness and sparse approximatiom[J].Neuro-computing,2002,48:85-105.
[21] VAPNIK V.The natural of statistical learning theory[M].New York:Springer-Verlag,1995.
[22] MELGANI F,BRUZZONE L.Classification of hyperspectral remote sensing images with support vector machine[J].IEEE Trans Geosci Remote Sens,2004,42(21):1778-1790.
[23] FOODY G M,MATHUR A.A relative evalution of multiclass image classification by support vector machines[J].IEEE Trans Geosci Remote Sens,2004,6(42):1335-1343.
[24] PAL M,MATHER P M.Support vector machines for classification in remote sensing[J].Int J Remote Sens,2005,3(26):1007-1011.
[25] 陈文倩,丁建丽,王娇,等.基于高分一号的土地覆被分类方法初探[J].干旱区地理,2016,39(1):182-189.[CHEN Wenqian,DING Jianli,WANG Jiao,et al.Classification method of land cover based on GF-1 image[J].Arid Land Geography,2016,39(1):182-189.]