基于盲信号处理框架的高光谱图像融合及目标检测算法研究

项目名称： 基于盲信号处理框架的高光谱图像融合及目标检测算法研究

项目编号： No.61273245

项目类型： 面上项目

立项/批准年度： 2013

项目学科： 自动化技术、计算机技术

项目作者： 史振威

作者单位： 北京航空航天大学

项目金额： 81万元

中文摘要： 高光谱遥感图像光谱分辨率高，而空间分辨率低；可见光遥感图像空间分辨率高，而光谱信息不足。将这两类图像融合，可获得高空间分辨率的高光谱图像，同时光谱保持不变，并对后续的图像处理，如目标检测更加有利。直接应用传统融合方法，在提高图像空间分辨率的同时，难以保持光谱不变，另外，在融合后的高光谱图像上有效进行目标检测的算法研究工作还鲜有发表。针对这些现状，我们创新性地提出在盲信号处理框架下统一研究高光谱图像融合和目标检测问题。我们将高光谱图像融合问题归结为一类特殊的盲信号分离问题，将高光谱目标检测问题归结为一类特殊的盲信号抽取问题，二者相互耦合并紧密相连，思想和方法可以相互借鉴和提升。基于此，本项目研究典型遥感图像特性；基于盲信号分离的高光谱图像融合；基于盲信号抽取的融合后高光谱图像目标检测等一系列富有挑战性的问题，并研制实用化软件系统。希冀能在科学思想、方向交叉上进行探索研究，产生好的学术成果。

中文关键词： 高光谱图像；高光谱图像融合；高光谱图像目标检测；盲信号处理；

英文摘要： The spectral resolution of hyperspectral image is high, while the spatial resolution is low. By contrast, the visible image has a higher spatial resolution but it lacks spectral information. By fusing such two kinds of images, we could get hyperspectral image with high spatial resolution and preserve the spectral information simultaneously. It is beneficial for the follow-up image processing such as target detection. However, it is hard for traditional fusion methods to improve the spatial resolution of the hyperspectral image while preserving its spectral information. Few researches on target detection in fused hyperspectral images have been done either at home or abroad. Under the circumstance, we innovatively propose new research ideas for hyperspectral image fusion and target detection based on the thoughts and methods of the blind source processing. We treat the hyperspectral image fusion problems and target detection problems as a series of special blind source separation problems and a series of blind signal extraction problems, respectively. These two kinds of problems are closely associated and their thoughts and methods can learn and improve from each other. Based on the thoughts, the project will study a series of challenging problems. These include features of typical hyperspectral images, hyperspect

英文关键词： Hyperspectral image；Hyperspectral image fusion；Hyperspectral image target detection；Blind signal processing；