A Spectral Mixture Model Based on Spectral Spatial Character of Measured Hyperspectral Data
-
摘要: 为提高高光谱混合像元分解精度, 利用地物多角度二向性反射平台和ASD FieldSpec3 Hi-Res便携式地物波谱仪, 设计等距离/等面积实验, 考虑探测距离远近对混合光谱的影响, 获取不同覆盖条件下叶片与方解石的混合光谱, 找出光谱数据空间特征变化规律, 提出等距离/等面积模型来消除空间位置对混合光谱的影响, 并将模拟混合光谱与实测混合光谱进行对比.通过实测数据分析, 混合反射率分布的权重系数随探测单位面积点与探头距离呈高斯变化规律; 与线性模型和线性改进模型进行光谱混合模拟结果相比, 应用权重系数高斯分布规律以等距离/等面积模型进行混合光谱模拟, 其模拟结果相似度平均增加了1.20%, 均方根误差平均降低了7.78%.等距离/等面积光谱混合模型考虑了光谱空间变化特征, 提高了光谱混合模拟的精度, 为进一步研究高光谱数据混合像元分解提供了新的方法.Abstract: In order to enhance the estimation of mixed spectral model, the equidistant/ homalographic model is established to analyze the spectral spatial character to simulate the mixture spectra. Based on the reflex platform and FieldSpec 3 Hi-Res portable spectrum instrument, the equidistant/ homalographic experiment, which takes the effect of distance between optical fiber probe and detected endmember into account, was designed to acquire the mixed spectral reflectance of calcite and green leaf. The measured mixed spectra analysis shows the weight coefficients of distribution change with the distance between the detected endmember and the probe is in a Gauss distribution. Compared with the linear spectral mixture model and improved linear spectral model, the results simulated by the equidistant/ homalographic model is 1.20% greater in similarity and 7.78% lower in RMSE. Considering the influence exerted by spectral spatial structure on mixed spectral simulation, the equidistant/ homalographic model proves to improve the accuracy of mixed spectral simulation and a new method for unmixing the mixed pixel of hyperspectral data.
-
图 2 等距离/等面积划分探测区域叶片覆盖不同区域与方解石混合光谱
Fig. 2. The mixed spectra of leaf and calcite by different equidistant/homalographic area
图 3 等距离/等面积覆盖不同区域方解石光谱曲线
Fig. 3. The calcite spectra of different equidistant/homalographic area
图 4 每单位面积权重系数随探测点与探头距离远近变化
Fig. 4. The weight coefficient of per unit area with the increase of distance between the detector and detected unit area
图 5 不同方法模拟混合反射率对比
Fig. 5. Comparison of simulated mixed spectra by different mixture model
表 1 等距离/等面积划分探测区域各部分权重系数
Table 1. The weight coefficient of different equidistant/ homalographic area
方解石覆盖区域 DJ1 DJ2 DJ3 DJ4 DJ5 权重系数 0.150 0 0.414 8 0.690 7 0.928 8 0.995 2 方解石覆盖区域 DM1 DM2 DM3 DM4 DM5 权重系数 0.551 5 0.767 0 0.879 1 0.949 0 0.963 1 
下载: 导出CSV
表 2 不同方法模拟混合光谱误差对比
Table 2. The simulated mixed spectra error by different mixture model
模拟方法 DJ1 DJ2 DJ3 DJ4 DJ5 等距离/等面积模拟 0.999 7 0.997 8 0.995 5 0.995 6 0.997 6 相似度 线性模拟 0.999 5 0.993 1 0.977 5 0.966 3 0.999 6 改进模型模拟 0.999 6 0.994 9 0.985 1 0.982 5 0.997 3 等距离/等面积模拟 0.021 6 0.050 8 0.066 9 0.064 7 0.075 1 均方根误差 线性模拟 0.049 2 0.120 8 0.150 0 0.138 4 0.215 0 改进模型模拟 0.051 5 0.127 6 0.156 6 0.122 1 0.077 0 模拟方法 DM1 DM2 DM3 DM4 DM5 等距离/等面积模拟 0.995 3 0.994 4 0.997 3 0.999 3 0.999 0 相似度 线性模拟 0.993 4 0.969 1 0.948 7 0.976 9 0.999 6 改进模型模拟 0.995 6 0.979 1 0.968 7 0.991 1 0.998 5 等距离/等面积模拟 0.086 2 0.060 6 0.037 7 0.022 9 0.032 4 均方根误差 线性模拟 0.116 2 0.174 7 0.169 2 0.125 0 0.181 4 改进模型模拟 0.125 1 0.188 5 0.175 2 0.077 9 0.043 0
下载: 导出CSV
表 3 不同方法模拟混合反射率的误差分析
Table 3. The simulated mixed spectra error by different mixture model
误差类别 线性模拟 改进模型模拟 等距离拟合 等面积拟合 相似度 0.977 6 0.988 5 0.996 4 0.996 1 均方根误差 0.141 6 0.141 7 0.061 7 0.062 2
下载: 导出CSV
-
[1] Borel, C.C., Gerstl, S.A.W., 1994. Nonlinear Spectral Mixing Models for Vegetative and Soil Surfaces. Remote Sensing of Environment, 47(3): 403-416. doi: 10.1016/0034-4257(94)90107-4 [2] Chen, S.B., Chen, L., Liu, Y.L., et al., 2013. Experimental Simulation on Mixed Spectra of Leaves and Calcite for Inversion of Carbonate Minerals from EO-1 Hyperion Data. GIS Science & Remote Sensing, 50(6): 690-703. doi: 10.1080/15481603.2013.866792 [3] Chen, S.B., Li, X.L., Chen, L., 2014. Study on Inversion of Soil Heavy Metal Elements Concentrations in Stream Sediments by In-Situ Hyperspectral Measurements. Journal of Jilin University(Earth Science Edition), 44(4): 1388-1394 (in Chinese with English abstract). http://www.researchgate.net/publication/285993021_Study_on_inversion_of_soil_heavy_metal_elements_concentrations_in_stream_sediments_by_in-situ_hyperspectral_measurements [4] Du, P.J., Lin, H., Sun, D.X., et al., 2006. On Progress of Support Vector Machine based Hyper Spectral RS Classification. Bulletin of Surveying and Mapping, (12): 37-40 (in Chinese). doi: 10.1080/01431160500242515 [5] Knipling, E.B., 1970. Physical and Physiological Basis for the Reflectance of Visible and Near-Infrared Radiation from Vegetation. Remote Sensing of Environment, 1(3): 155-159. doi: 10.1016/S0034-4257(70)80021-9 [6] Li, X.W., Strahler, A.H., 1985. Geometric-Optical Modeling of a Conifer Forest Canopy. IEEE Trans. Geosci. Remote Sensing, 23(5): 705-721. doi: 10.1109/tgrs.1985.289389 [7] Li, X.W., Strahler, A.H., 1986. Geometric-Optical Bidirectional Reflectance Modeling of a Conifer Forest Canopy. IEEE Trans. Geosci. Remote Sensing, 24: 906-919. http://ieeexplore.ieee.org/document/4072562/citations [8] Lü, J., Liu, X.N., 2012. Hyperspectral Remote Sensing Estimation Model for Cd Concentration in Rice Using Support Vector Machines. Journal of Applied Sciences, 30(1): 105-110 (in Chinese with English abstract). http://en.cnki.com.cn/Article_en/CJFDTOTAL-YYKX201201018.htm [9] Miao, X., Gong, P., Swope, S., et al., 2006. Estimation of Yellow Starthistle Abundance through CASI-2 Hyperspectral Imagery Using Linear Spectral Mixture Models. Remote Sensing of Environment, 101(3): 329-341. doi: 10.1016/j.rse.2006.01.006 [10] Piwowar, J.M., Peddle, D.R., LeDrew, E.F., 1998. Temporal Mixture Analysis of Arctic Sea Ice Imagery: A New Approach for Monitoring Environmental Change. Remote Sensing of Environment, 63(3): 195-207. doi: 10.1016/j.rse.2004.10.008 [11] Raksuntorn, N., Du, Q., 2010. Nonlinear Spectral Mixture Analysis for Hyperspectral Imagery in an Unknown Environment. IEEE Geoscience and Remote Sensing Letters, 7(4): 836-840. doi: 10.1109/lgrs.2010.2049334 [12] Roberts, D.A., Adams, J.B., Smith, M.O., 1990. Predicted Distribution of Visible and Near-Infrared Radiant Flux above and below a Transmittant Leaf. Remote Sensing of Environment, 34(1): 1-17. doi: 10.1016/0034-4257(90)9008006 [13] Savitzky, A., Golay, M.J.E., 1964. Smoothing and Differentiation of Data by Simplified Least Squares Procedures. Analytical Chemistry, 36(8): 1627-1639. doi: 10.1021/ac60214a047 [14] Yang, C.H., Everitt, J.H., Du, Q., 2010. Applying Linear Spectral Unmixing to Airborne Hyperspectral Imagery for Mapping Yield Variability in Grain Sorghum and Cotton Fields. Journal of Applied Remote Sensing, 4(1): 041887. doi: 10.1117/1.3484252 [15] Zhu, F., Gong, H.L., Sun, T.L., et al., 2013. Study on the Nonlinear Characteristics of the Mixed Pixel's Reflectance in Hyperspectral Space. Spectroscopy and Spectral Analysis, 33(3): 737-740 (in Chinese with English abstract). http://europepmc.org/abstract/med/23705444 [16] Zhu, L., Xu, J.F., Huang, J.F., et al., 2008. Study on Hyperspectral Estimation Model of Crop Vegetation Cover Percentage. Spectroscopy and Spectral Analysis, 28(8): 1827-1831 (in Chinese with English abstract). http://europepmc.org/abstract/MED/18975813 [17] 陈圣波, 李鑫龙, 陈磊, 2014. 基于地面实测光谱的水系沉积物重金属含量反演. 吉林大学学报(地球科学版), 44(4): 1388-1394. https://www.cnki.com.cn/Article/CJFDTOTAL-CCDZ201404035.htm [18] 杜培军, 林卉, 孙敦新, 等, 2006. 基于支持向量机的高光谱遥感分类进展. 测绘通报, (12): 37-40. doi: 10.3969/j.issn.0494-0911.2006.12.011 [19] 吕杰, 刘湘南, 2012. 利用支持向量机构建水稻镉含量高光谱预测模型. 应用科学学报, 30(1): 105-110. doi: 10.3969/j.issn.0255-8297.2012.01.016 [20] 朱蕾, 徐俊锋, 黄敬峰, 等, 2008. 作物植被覆盖度的高光谱遥感估算模型. 光谱学与光谱分析, 28(8): 1827-1831. doi: 10.3964/j.issn.1000-0593.2008.08.032 [21] 朱锋, 宫辉力, 孙天琳, 等, 2013. 高光谱空间中混合象元非线性反射特征研究. 光谱学与光谱分析, 33(3): 737-740. doi: 10.3964/j.issn.1000-0593(2013)03-0737-04 -
点击查看大图
图(5) / 表(3)
计量
- 文章访问数: 3011
- HTML全文浏览量: 73
- PDF下载量: 465
- 被引次数: 0
