机载时空调制干涉成像光谱数据误差建模及其校正方法研究

Imaging spectrometers collect data over two spatial dimensions and one spectral dimension, which provide spectral information for target detection and categorization. Compared with traditional imaging spectrometer, temporally-spatially modulated Fourier transform imaging spectrometer (TSMFTIS) has many advantages, such as high throughput, multiplex and high stability. As a result, this type of imaging spectrometer can achieve higher spatial resolution than the traditional hyperspectral imaging spectrometer. However, the data acquisition of TSMFTIS involves a constant push-broom process and requires highly stability of the platform. If the attitudes of the platform changes during the data acquisition process, the acquired interferogram will be distorted and lead to serious distortion of the reconstructed spectral information. On the other hand, airborne imagery has emerged as a powerful remote sensing tool for its advantages such as high mobilization flexibility, spatial resolution and temporal flexibility. Even if an additional stabilized-platform was employed for movement compensation, the movement stability during the scanning still cannot fulfill the stability requirement of the traditional spectrum reconstruction algorithm for a TSMFTIS. Therefore, the applications of the TSMFTIS in airborne remote sensing are limited. In this research, we will study the data error and propose the correction methods for airborne TSMFTIS. The data acquisition process and influences of the errors will be modeled and analyzed. Based on the instability character of the airborne platform, the error introduced in the remote sensed data of TSMFTIS will be analyzed and modeled. Combining the direct georeferencing associated with the Global Positioning System (GPS) and an inertial navigation system (INS), we will propose a data correction method based on dynamic reverse tracking of target for airborne application of TSMFTIS. The related key problems of the proposed method will be thoroughly analyzed. Then, fast and precision spectral reconstruction methods will be studied and presented. Finally, experiments, including a flight test, will be performed to verify the feasibility and validity of the proposed models and methods.

时空调制干涉成像光谱仪与传统的成像光谱仪相比，具有高通量、多通道、高稳定度等诸多优点，具备实现高空间分辨率高光谱遥感的能力。由于该类成像光谱仪遥感数据获取与搭载平台运动具有很强的相关性，平台的非平稳运动将导致光谱数据发生严重畸变，制约了该类技术在航空遥感领域的推广应用。本项目拟开展机载时空调制干涉成像光谱仪遥感数据校正及光谱反演方法研究，在分析研究成像光谱仪运动成像及其误差影响理论的基础上，结合机载平台特性进行成像光谱仪遥感数据动态误差建模与分析，针对成像光谱仪结合POS测量系统进行机载遥感探测方法，提出基于目标像点动态逆向追踪进行数据误差校正的思路和方法，并对其中涉及的关键问题进行深入研究，在此基础上，开展快速高精度光谱反演校正方法研究并进行高光谱数据立方体重构，最后，利用机载飞行试验数据验证本项目的理论和方法研究的有效性和可行性。

时空调制干涉成像光谱仪与传统的成像光谱仪相比，具有高通量、多通道、高稳定度等诸多优点，具备实现高空间分辨率高光谱遥感的能力。由于该类成像光谱仪遥感数据获取与搭载平台运动具有很强的相关性，平台的非平稳运动将导致光谱数据发生严重畸变，制约了该类技术在航空遥感领域的推广应用。.为了有效提高TSMFTIS的复原光谱精度，本项目依次理论分析、计算机仿真和实验验证三个层次对机载时空调制干涉成像光谱仪遥感数据校正及光谱复原方法展开了深入研究，主要内容包括：.(1)建立了机载TSMFTIS遥感成像动态成像模型；分析了成像过程中存在的误差源与机载平台的运动规律，建立了遥感动态成像误差模型。.(2)研究了基于严格几何光学模型的逆向动态追踪的直接目标定位方法和基于图像配准的帧间图像位移估计的间接目标定位方法，利用仿真的干涉图像分析和评价了两种方法的效果。.(3)提出了利用相对偏差将灰度信息与距离信息相结合的核函数，建立用于干涉图像重采样的回归模型，利用周围像元灰度值与均值的偏离程度来自适应调制核函数权重，实现定位于像面非整像素的目标干涉强度信息提取。 .(4)提出了基于空间域的小波变换光谱信息重构方法和基于光谱域的非均匀傅里叶变换光谱信息复原方法，利用上述方法重构非等间隔采样的干涉图谱数据，分别对比了算法的光谱复原效果。.(5)开展了数据预处理研究，分别提出了基于数据重构与光谱复原、基于最小梯度信息、基于径向基神经网络和基于WLSSVM的干涉图像坏像元修复方法，并利用仿真图像数据对算法的实施效果进行分析。.(6)利用机载飞行试验数据验证上述本项目提出的理论和方法研究的有效性和可行性。