面向对象建筑物目标提取的最优分割尺度选择
|
摘要
面向对象影像分析技术是高分辨率遥感影像自动解译的重要手段,影像分割作为面向对象信息提取的关键预处理步骤,其核心问题在于如何选择合适的影像分割尺度。现有的最优尺度自动确定方法均值方差法、同质性-异质性模型和最大面积法等仅通过建模影像的光谱和几何特征进行最优尺度选取,而忽略了地物纹理信息这一重要特征,难以适用于复杂影像场景的最优尺度选取。针对此问题,提出一种基于灰度共生矩阵的纹理均值法,开展面向对象建筑物目标提取的最优分割尺度选择研究。首先通过生成灰度共生矩阵来提取影像的纹理均值统计量,实现纹理特征和光谱特征的联合建模,继而利用纹理均值与分割尺度之间的变化曲线来自动确定最优尺度。居民住宅区、工业建筑区和教育用地建筑区三类建筑物场景下开展的实验结果表明,本文所提出的方法针对面向对象建筑物目标提取任务能够获得更好的影像分割结果,并且对不同建筑物场景的适用性更强。
Object-oriented image analysis technique is one the most commonly used means for automatic interpretation of high-resolution remote sensing images. As the key pre-processing step of object-based analysis, image segmentation is yet limited to the determination of the proper segmentation scale. It is difficult for the complex image scenes optimal scale selection by means of existing approaches for the automatic selection of optimal segmentation scale, including the mean-variance method, the homogeneity-heterogeneity model, and the maximum area method which used only the spectral or geometric features, and ignored the important feature of texture information. This observation motivates us to develop a novel texture-mean approach based on grey level co-occurrence matrix to conduct the optimal segmentation scale selection for the object-oriented building extraction. The proposed method builds up the grey level co-occurrence matrix followed by a feature extraction step to obtain the statistic of the texture mean, and both the spectral and textural features are therefore jointly utilized. Finally, the change curve of the texture mean and the segmentation scale are created to automatically infer the optimal scale. Experimental results using high-resolution remote sensing images of three types of building scenes indicated that the proposed approach can achieve better image segmentation results for the object-oriented building extraction task in terms of both qualitative and quantitative evaluation, and are more robust to various types of building scenes.
Object-oriented image analysis technique is one the most commonly used means for automatic interpretation of high-resolution remote sensing images. As the key pre-processing step of object-based analysis, image segmentation is yet limited to the determination of the proper segmentation scale. It is difficult for the complex image scenes optimal scale selection by means of existing approaches for the automatic selection of optimal segmentation scale, including the mean-variance method, the homogeneity-heterogeneity model, and the maximum area method which used only the spectral or geometric features, and ignored the important feature of texture information. This observation motivates us to develop a novel texture-mean approach based on grey level co-occurrence matrix to conduct the optimal segmentation scale selection for the object-oriented building extraction. The proposed method builds up the grey level co-occurrence matrix followed by a feature extraction step to obtain the statistic of the texture mean, and both the spectral and textural features are therefore jointly utilized. Finally, the change curve of the texture mean and the segmentation scale are created to automatically infer the optimal scale. Experimental results using high-resolution remote sensing images of three types of building scenes indicated that the proposed approach can achieve better image segmentation results for the object-oriented building extraction task in terms of both qualitative and quantitative evaluation, and are more robust to various types of building scenes.
引文
[1]李德仁,童庆禧,李荣兴,等.高分辨率对地观测的若干前沿科学问题[J].中国科学:地球科学,2012,42(6):805-813.
[2]李德仁,王密,沈欣,等.从对地观测卫星到对地观测脑[J].武汉大学学报:信息科学版,2017,42(2):143-149.
[3]杨冀红,郭蕾,孙家波,等.利用SPOT-6影像提取新增建设用地的方法研究[J].地理信息世界,2014,21(4):54-58.
[4]Shen L,Wu L,Dai Y,et al.Topic Modeling for Object-based Unsupervised Classification of VHR Panchromatic Satellite Image Based on Multiscale Image Segmentation[J].Remote Sensing,2017,9(8):840.
[5]吴政,李成名,武鹏达,等.面向对象的形态学建筑物指数及其高分辨率遥感影像建筑物提取应用[J].测绘学报,2017,46(5):724-733.
[6]慎利,唐宏,王世东,等.结合空间像素模板和Adaboost算法的高分辨率遥感影像河流提取[J].测绘学报,2013,42(3):344-350.
[7]黄慧萍.面向对象影像分析中的尺度问题研究[D].北京:中国科学院研究生院,2003.
[8]Duro D C,Franklin S E,Dube M G.A Comparison of Pixel-based and Object-based Image Analysis with Selected Machine Learning Algorithms for the Classification of Agricultural Landscapes Using SPOT-5 HRG Imagery[J].Remote Sensing of Environment,2012,118:259-272.
[9]Grybas H,Melendy L,Congalton R G.A Comparison o f Un supe rvi sed Se gme nta tion Par ame ter Optimization Approaches Using Moderate and Highresolution Imagery[J].GIScienc&Remote Sensing,2017,54(4):515-533.
[10]Clinton N,Holt A,Scarborough J,et al.Accuracy Assessment Measures for Object-based Image Segmentation Goodness[J].Photogrammetric Engineering&Remote Sensing,2010,76(3):289-299.
[11]Dragut L,Tiede D,Levick S R.ESP:a Tool to Estimate Scale Parameter for Multiresolution Image Segmentation of Remotely Sensed Data[J].International Journal of Geographical Information Science,2010,24(6):859-871.
[12]Yang J,He Y,Weng Q.An Automated Method to Parameterize Segmentation Scale by Enhancing Intrasegment Homogeneity and Intersegment Heterogeneity[J].IEEE Geoscience and Remote Sensing Letters,2015,12(6):1 282-1 286.
[13]何敏,张文君,王卫红.面向对象的最优分割尺度计算模型[J].大地测量与地球动力学,2009,29(1):106-109.
[14]胡文亮,赵萍,董张玉.一种改进的遥感影像面向对象最优分割尺度计算模型[J].地理与地理信息科学,2010,26(6):15-18.
[15]Kuffer M,Pfeffer K,Sliuzas R,et al.Extraction of Slum Areas from VHR Imagery Using GLCM Variance[J].IEEE Journal of Selected Topics in Applied Earth Observations and Remote Sensing,2016,9(5):1 830-1 840.
[16]Haralick R M,Shanmugam K.Textural Features for Image Classification[J].IEEE Transactions on Systems,Man,and Cybernetics,1973 SMC-3(6):610-621.
[17]Cheng J,Bo Y,Zhu Y,et al.A Novel Method for Assessing the Segmentation Quality of High-spatial Resolution Remote-sensing Images[J].International Journal of Remote Sensing,2014,35(10):3 816-3 839.
[2]李德仁,王密,沈欣,等.从对地观测卫星到对地观测脑[J].武汉大学学报:信息科学版,2017,42(2):143-149.
[3]杨冀红,郭蕾,孙家波,等.利用SPOT-6影像提取新增建设用地的方法研究[J].地理信息世界,2014,21(4):54-58.
[4]Shen L,Wu L,Dai Y,et al.Topic Modeling for Object-based Unsupervised Classification of VHR Panchromatic Satellite Image Based on Multiscale Image Segmentation[J].Remote Sensing,2017,9(8):840.
[5]吴政,李成名,武鹏达,等.面向对象的形态学建筑物指数及其高分辨率遥感影像建筑物提取应用[J].测绘学报,2017,46(5):724-733.
[6]慎利,唐宏,王世东,等.结合空间像素模板和Adaboost算法的高分辨率遥感影像河流提取[J].测绘学报,2013,42(3):344-350.
[7]黄慧萍.面向对象影像分析中的尺度问题研究[D].北京:中国科学院研究生院,2003.
[8]Duro D C,Franklin S E,Dube M G.A Comparison of Pixel-based and Object-based Image Analysis with Selected Machine Learning Algorithms for the Classification of Agricultural Landscapes Using SPOT-5 HRG Imagery[J].Remote Sensing of Environment,2012,118:259-272.
[9]Grybas H,Melendy L,Congalton R G.A Comparison o f Un supe rvi sed Se gme nta tion Par ame ter Optimization Approaches Using Moderate and Highresolution Imagery[J].GIScienc&Remote Sensing,2017,54(4):515-533.
[10]Clinton N,Holt A,Scarborough J,et al.Accuracy Assessment Measures for Object-based Image Segmentation Goodness[J].Photogrammetric Engineering&Remote Sensing,2010,76(3):289-299.
[11]Dragut L,Tiede D,Levick S R.ESP:a Tool to Estimate Scale Parameter for Multiresolution Image Segmentation of Remotely Sensed Data[J].International Journal of Geographical Information Science,2010,24(6):859-871.
[12]Yang J,He Y,Weng Q.An Automated Method to Parameterize Segmentation Scale by Enhancing Intrasegment Homogeneity and Intersegment Heterogeneity[J].IEEE Geoscience and Remote Sensing Letters,2015,12(6):1 282-1 286.
[13]何敏,张文君,王卫红.面向对象的最优分割尺度计算模型[J].大地测量与地球动力学,2009,29(1):106-109.
[14]胡文亮,赵萍,董张玉.一种改进的遥感影像面向对象最优分割尺度计算模型[J].地理与地理信息科学,2010,26(6):15-18.
[15]Kuffer M,Pfeffer K,Sliuzas R,et al.Extraction of Slum Areas from VHR Imagery Using GLCM Variance[J].IEEE Journal of Selected Topics in Applied Earth Observations and Remote Sensing,2016,9(5):1 830-1 840.
[16]Haralick R M,Shanmugam K.Textural Features for Image Classification[J].IEEE Transactions on Systems,Man,and Cybernetics,1973 SMC-3(6):610-621.
[17]Cheng J,Bo Y,Zhu Y,et al.A Novel Method for Assessing the Segmentation Quality of High-spatial Resolution Remote-sensing Images[J].International Journal of Remote Sensing,2014,35(10):3 816-3 839.