宽波段遥感图像亚像元丰度和温度联合制图技术研究

发布时间：2018-04-10 19:55

本文选题：宽波段高光谱数据 + 空间信息　；参考：《哈尔滨工业大学》2017年硕士论文

【摘要】：随着遥感技术的不断发展,人们获取的高光谱遥感数据不但在空间分辨率以及光谱分辨率上取得了长足的进步,其波段范围更是得到了极大的扩展。宽波段遥感图像覆盖了可见/近红外以及热红外波段,为科研工作者提供了更为完备与充分的遥感信息。由于可见/近红外传感器与热红外传感器固有成像机理的不同,其所获得的遥感图像有着各自的特点,反映出了地面目标不同的物化特性:前者有着较高的空间分辨率,包含了目标丰富的空间、结构以及光谱信息;后者虽然空间分辨率较低,但是由于其热辐射成像的特点,红外数据中包含了地物目标温度以及辐射信息。本文从宽波段遥感数据信息充分利用的角度出发,提出了宽波段遥感图像亚像元丰度以及温度联合制图技术。具体研究内容如下:首先,对于可见光波段遥感图像而言,由于其较高的空间分辨率,可以提供丰富的空间与结构信息。传统的端元提取算法只注重光谱信息来提取端元,而忽视了遥感图像的空间特性。因此,未考虑空间信息的端元提取算法对噪声以及异常信号较为敏感,导致端元提取精度的降低,因而其误差会传递到亚像元制图的过程中去。针对这个问题,本文提出一种基于正交子空间及局部空间相关性的端元提取算法。该算法利用空间信息对提取的端元进行判定、更新以及扩展,从而保证了端元提取算法的有效性,为亚像元制图提供准确的丰度支撑。其次,热红外波段遥感图像蕴含地物丰富的温度以及辐射率信息。本文通过对传统温度反演算法的学习与分析,针对混合像元各组分温度反演难以实现的问题,提出了一种亚像元级温度反演算法。该算法利用丰度信息对纯像元以及混合像元加以区分,并分别进行处理。对于纯像元使用传统温度辐射率分离算法,实现温度与辐射率的估计。对于混合像元,该算法以大气底层辐射线性混合模型为基础,对混合像元中不同地物组分的温度分别进行求解。该部分实现了热红外图像亚像元级温度反演,为亚像元制图提供了地物温度信息。最终,针对宽波段遥感图像,充分结合可见光/近红外遥感图像与热红外遥感图像各自的优势,形成完整的丰度以及温度联合制图算法。利用可见光波段遥感图像丰度的空间信息对像元各个地物组分的丰度进行估计,并实现纯像元与混合像元的定位。利用热红外波段所提供的温度信息,结合所得到的丰度,对纯像元以及混合像元的温度、辐射率信息进行反演。在亚像元制图的过程中,利用像元吸引模型对亚像元空间位置进行分配与调整,使制图结果更加符合真实地物分布状态。
[Abstract]:With the development of remote sensing technology, the hyperspectral remote sensing data not only has made great progress in spatial resolution and spectral resolution, but also has greatly expanded its band range.The wide band remote sensing images cover the visible / near infrared and thermal infrared bands, which provide more complete and sufficient remote sensing information for researchers.Because of the difference of the inherent imaging mechanism between the visible / near infrared sensor and the thermal infrared sensor, the remote sensing images obtained by them have their own characteristics, which reflect the different physical and chemical characteristics of the ground target. The former has higher spatial resolution.The latter contains rich spatial, structural and spectral information. Although the spatial resolution of the latter is low, the infrared data contain object temperature and radiation information due to its thermal radiation imaging characteristics.In this paper, from the point of view of making full use of remote sensing data information in wide band, the technique of sub-pixel abundance and temperature joint mapping of wide band remote sensing image is put forward.The specific research contents are as follows: first, for the visible light band remote sensing image, because of its high spatial resolution, it can provide rich spatial and structural information.The traditional End-element extraction algorithm only pays attention to spectral information, but neglects the spatial characteristics of remote sensing image.Therefore, the End-element extraction algorithm without spatial information is sensitive to noise and abnormal signals, which leads to the reduction of the precision of End-element extraction, so the error will be transferred to the process of sub-pixel mapping.In order to solve this problem, this paper presents an end-component extraction algorithm based on orthogonal subspace and local spatial correlation.The algorithm uses spatial information to judge, update and extend the extracted end elements, thus ensuring the effectiveness of the end element extraction algorithm and providing accurate abundance support for sub-pixel mapping.Secondly, thermal infrared remote sensing images contain rich information of temperature and emissivity.Based on the study and analysis of the traditional temperature inversion algorithm, a subpixel temperature inversion algorithm is proposed to solve the problem that the temperature inversion of each component of mixed pixel is difficult to realize.The algorithm uses abundance information to distinguish pure and mixed pixels, and processes them separately.Traditional temperature emissivity separation algorithm is used to estimate temperature and emissivity for pure pixels.For mixed pixels, the algorithm is based on the linear mixing model of atmospheric bottom radiation, and the temperature of different ground components in the mixed pixel is solved separately.In this part, subpixel temperature inversion of thermal infrared image is realized, which provides ground object temperature information for subpixel mapping.Finally, based on the advantages of visible / near infrared remote sensing images and thermal infrared remote sensing images, a complete joint mapping algorithm of abundance and temperature is formed for the wide band remote sensing images.The spatial information of the abundance of remote sensing image in the visible light band is used to estimate the abundance of each component of the pixel and to realize the localization of pure pixel and mixed pixel.The temperature and emissivity information of pure and mixed pixels are retrieved by using the temperature information provided by the thermal infrared band and the abundance obtained.In the process of sub-pixel mapping, the pixel attraction model is used to allocate and adjust the location of sub-pixel space, so that the mapping results are more consistent with the real distribution state of ground objects.
【学位授予单位】：哈尔滨工业大学
【学位级别】：硕士
【学位授予年份】：2017
【分类号】：TP751

【参考文献】