基于遥感影像的卫星颤振探测分析的关键技术与实现

Research on Key Technology and Implementation of Detection and Analysis of Satellite Attitude Jitter using Remote Sensing Imagery

姿态颤振是高分辨率卫星普遍存在的复杂现象,随着卫星的复杂性和敏捷性的提高,故其对成像质量的影响机制和相应的探测与补偿方法,成为高分辨率卫星对地观测研究的关键问题。本文介绍了基于遥感影像的卫星颤振探测原理和关键技术,利用传感器的构造特点,通过不同波段的多光谱影像,以及不同视角的全色立体影像间的密集匹配精确量测和获取影像视差差异值,反演卫星姿态颤振的规律和特性,从而对规律性的颤振现象进行建模和补偿。鉴于上述理论,在Visual C++的平台下,研制了卫星颤振探测分析系统,设计了以计算机性能优化的图像分块并行算法。测试结果表明,该软件能有效探测出0.1像元偏移的卫星姿态颤振,并可对规律性颤振进行建模和补偿,有较好的稳定性和并行效率。其已成功应用于资源三号卫星地面应用系统,并可扩展于其他卫星系统的颤振探测分析。

With the development of high resolution satellites, the attitude jitter, which is a common and complex phenomenon, has become a key problem in the applications of high resolution remote sensing images. It leads to the degradation of image quality, and non-steady and lower precision for the attitude measurement data. There- fore, in this paper, a jitter detection and analysis method based on remotely sensed imagery is proposed. It took the advantage of the imagery data to detect jitter without additional involvement of high-performance sensors or a large quantity of GCPs. Two different jitter detection methods for different types of sensors are presented, which consist of： （1）jitter detection methods based on multispectral images utilizing parallax between two differ- ent bands in the image, and （2） jitter detection methods based on stereo images using triplet stereo images and ra- tional polynomial coefficients （RPCs）. With the jitter detection methods, the relative registration error is ob- tained. Afterwards, we can retrieve the time-dependent attitude jitter. Consequently, the frequency and amplitude of the attitude jitter can be achieved by analyzing and modeling the jitter using cosine functions. To reduce the jit- ter effect on geo-positioning process, the true attitude information should be estimated from the detected jitter components. An improved sensor model is introduced to take into account the compensation for the attitude jit- ter. The performance of the improved sensor model is assessed in direct georeferencing. Experiments are con- ducted with images acquired from the first civilian three-line-array stereo mapping satellite ZY-3. A periodic pat- tern with a frequency of about 0.65Hz has been detected. Based on the theories and methods mentioned above, the detection and analysis of satellite attitude jitter system based on remote sensing imagery has been developed using VC＋＋. An image sub-blocking and a parallel computation scheme based on the computer performance is performed, which can improve the efficiency. Test results show that the system can efficiently detect the attitude jitter of 0.1 pixels. Meanwhile, it can model the regular jitter and correct the jitter effect on geo-positioning pro- cess. The accuracies for the plain and height directions are improved from 2.1799 m and 3.2497 m to 1.4396 m and 0.7453 m, respectively. This system has a good stability and robustness and has been successfully applied to the ZY-3 ground application system.