针对遥感混合像元分解中不能有效利用空间邻域信息的问题，提出一种基于超分辨率重建的分解方法. 通过小波系数双线性插值获得遥感影像的超分辨率影像，对超分辨率影像进行监督分类生成超分辨率分类图，最后通过窗口统计得到原始分辨率下各地物的丰度图. 广州城区的模拟TM遥感影像试验表明，该方法的分解精度
在3种方法中最优，能够较充分利用空间邻域信息，提高混合像元分解精度，为混合像元分解提供了新的途径.
This paper proposes a wavelet coefficient interpolation method, which uses the neighboring information in the spatial domain for spectral unmixing of remote sensing images. A super-resolution image is first generated using bilinear interpolation of wavelet coefficients. The new image is then classified to derive
a super-resolution classification map. Finally, an abundance map at the original spatial resolution is obtained using a counting window on the super-resolution classification map. This way, the original image is unmixed. A simulated TM image of Guangzhou City is used to verify the proposed method. It is found that the method performs best among three methods as it can make use of neighboring information in the space to improve unmixing accuracy.
[1]CHEN C H, HO P G P. Statistical pattern recognition in remote sensing [J]. Pattern Recognition, 2008, 41(9): 2731-2741.
[2]DURAN O, PETROU, M. Spectral unmixing with negative and superunity abundances for subpixel anomaly detection [J]. IEEE Geoscience and Remote Sensing Letters, 2009, 6(1): 152-156.
[3]ECHES O, DOBIGEON, N, MAILHES C. Bayesian estimation of linear mixtures using the normal compositional model [J]. IEEE Transactions on Image Processing, 2010, 19(6): 1403-1413.
[4]CHEN J, JIA X P, YANG W, MATSUSHITA B. Generalization of subpixel analysis for hyperspectral data with flexibility in spectral similarity measures [J]. IEEE Transactions on Geoscience and Remote Sensing, 2009, 47(7): 2165-2171.
[5]RAKSUNTORN N, DU Q. Nonlinear spectral mixture analysis for hyperspectral imagery in an unknown environment [J]. IEEE Geoscience and Remote Sensing Letters, 2010, 7(4): 836-840.
[6]LEE S, LATHROP R G. Subpixel analysis of Landsat ETM+ using self-organizing map (SOM) neural networks for urban land cover characterization [J]. IEEE Transactions on Geoscience and Remote Sensing, 2006, 44(6): 1642-1654.
[7]PLAZA J, PLAZA A, PEREZ R, MARTINEZ P. On the use of small training sets for neural network-based characterization of mixed pixels in remotely sensed hyperspectral images [J]. IEEE Transactions on Geoscience and Remote Sensing, 2009, 42(11): 3032-3045.
[8]WANG L, JIA X P. Integration of soft and hard classifications using extended support vector machines [J]. IEEE Geoscience and Remote Sensing Letter, 2009, 6(3): 543-547.
[9]TAKEDA H, FARSIU S, MILANFAR P. Kernel regression for image processing and reconstruction [J]. IEEE Transactions on Image Processing, 2007, 16(2): 349-366.
[10]YANG J C, WRIGHT J, HUANG T S, MA Y. Image super-resolution via sparse representation [J]. IEEE Transactions on Image Processing, 2010, 19(11): 2861-2873.
[11]KIM K I, KWON Y. Single-image super-resolution using sparse regression and natural image prior [J]. IEEE Transactions on Pattern Analysis and Machine Intelligence, 2010, 32(6): 1127-1133.
[12]刘笑宙，涂国防. 小波双线性插值应用于光学遥感图像 [J]. 中国科学院研究生院学报, 2003, 20(1): 39-43.
LIU Xiaozhou, TU Guofang. Wavelet bilinear interpolation in remote sensing image [J]. Journal of the Graduate School of the Chinese Academy of Sciences, 2003, 20(1): 39-43. (in Chinese)
[13]Liu W G, Wu E Y. Comparison of non-linear mixture models: sub-pixel classification [J]. Remote Sensing of Environment, 2005, 94(2): 145-154.
[14]POWELL R L, ROBERTS D A, DENNISON P E, HESS L L. Sub-pixel mapping of urban land cover using multiple endmember spectral mixture analysis: Manaus, Brazil [J]. Remote Sensing of Environment, 2007, 106(2): 253-267.
[15]BEN R Z, FARAH I R, MERCIER G, SOLAIMAN B. A new method to change illumination effect reduction based on spectral angle constraint for hyperspectral image unmixing [J]. IEEE Geoscience and Remote Sensing Letters, 2011, 8(6): 1110-1114.
