较大空谱分辨率差异下的多源遥感影像亚像元变化检测研究

多分辨率多源遥感影像变化检测是国内外研究的热点和难点问题，尤其是在多源影像间存在较大的空间分辨率和光谱分辨率差异的情况下，如高分五号的VIMS成像仪空间分辨率为30米、光谱波段为330个，而SPOT-7多光谱空间分辨率为6米、光谱波段为4个。针对低空间分辨率、高光谱分辨率的影像与高空间分辨率、低光谱分辨率的影像之间的变化检测难题，本项目在多光谱影像空间尺度基准下，以“光谱响应可信估计-类别自适应映射-亚像元变化检测联合模型构建”为研究主线，发展稳健、快速的多分辨率多源遥感影像亚像元变化检测方法，解决了多源影像间存在的波段匹配难、地物类别映射难、空间尺度转换难的问题，实现了较大空谱分辨率差异下的多源遥感影像亚像元变化检测。研究为多源影像精细化分析提供了新的思路，能够有效提升具有不同分辨率特性的多源影像应用潜力，有重要的理论与应用意义。

Change detection for multi-source remotely sensed images with different resolutions has attracted worldwide attention from researchers, especially when there is a big difference on the spatial and spectral resolutions for image pairs. For example, the VIMS sensor on GF-5 satellite can acquire hyperspectral images which have a spatial resolution of 30 meters and 330 spectral bands. While for the Multi-spectral instrument on SPOT-7 satellite, the spatial resolution is 6 meters and 4 spectral bands. Given the spatial scale of multispectral image as the baseline, this project develops a robust and efficient sub-pixel change detection framework for hyperspectral images with lower spatial resolution and higher spectral resolution to multispectral images with higher spatial resolution and lower spectral resolution. The proposed framework involves spectral response function estimation, adaptive endmember projection and the joint sub-pixel change detection modeling to solve the difficulties of spectral band matching, land cover relating and spatial scale converting for image pairs. By this way, the sub-pixel change detection result can be obtained for multi-source remote sensing images with a big difference on spatial and spectral resolutions. This research provides an innovative method for multi-source image analyzing, and contributes to improving the capability of multi-source remotely sensed images with different resolutions in applications.