高分辨率遥感卫星的多模型自适应颤振检测研究

Given the recent development in high-resolution (HR) remote sensing satellites, attitude jitter (AJ) has become increasingly the primary factor affecting on the HR images quality. The recent research shows that the use of short-time asynchronous images and image-registration method can effectively detect the AJ of HR remote sensing satellites, and consequently compensate the image AJ deterioration and improve the image quality by using those detecting results. However, how to achieve engineering applications of AJ detection system for the different HR satellite with different parameters, time and accuracy requirements, has become an open issue. On the basis of previous work, the applicant proposes the multiple model-based adaptive AJ detection system to solve the actually problems of engineering application for AJ processes in the image processing system of HR satellites. To construct AJ detection system, this work will study on the multiple AJ detection models, and use data dimensionality reduction method, parallel processing technology in the typical working area and other methods to improve processing efficiency for each detection model. Consequently, this work will establish multi model-based adaptive control strategy by studying on underlying efficiency evaluation function and hierarchically supervision control, and discuss the feedback-based looping control strategy by additionally constructing AJ compensation and evaluation system. Finally, this work will use a large number of detection results to in-depth study on the model and long-term trend of AJ. In summary, the work of this study for jitter detection will propose a framework for addressing the jitter of HR remote sensing satellites.

近些年来随着高分辨率遥感卫星的发展，卫星平台颤振逐渐变成影响高分辨率影像质量的主要因素。最新研究表明，利用非同时成像影像匹配的方法可以有效的检测出高分辨率卫星的平台颤振，依据检测结果可以有效的补偿颤振降质，从而提高影像质量。然而对于高分辨率卫星不同载荷参数，时段和精度要求，如何实现颤振检测系统的工程化应用成为目前的研究热点。结合前期工作，申请者提出基于多模型的自适应颤振检测研究，解决颤振检测系统在高分辨率卫星影像地面处理中的实际应用问题。为构建颤振检测系统，本研究将开展颤振检测多模型的算法研究，利用数据降维、典型工作区并行处理等手段提高检测模型的效率，通过研究底层的性能评价函数和顶层的监督分层控制，建立多模型自适应切换的控制策略，探讨基于颤振补偿及质量评价体系的反馈闭环控制策略，并通过大量检测结果分析颤振的模型和变化趋势，为解决高分辨率遥感卫星颤振问题奠定基础。

近些年来随着高分辨率遥感卫星的发展，卫星平台颤振逐渐变成影响高分辨率影像质量的.主要因素。最新研究表明，利用非同时成像影像匹配的方法可以有效的检测出高分辨率卫星的平台颤振，依据检测结果可以有效的补偿颤振降质，从而提高影像质量。然而对于高分辨率卫星不同载荷参数，时段和精度要求，如何实现颤振检测系统的工程化应用成为目前的研究热点。本项目结合前期工作成果，开展了基于多模型的自适应颤振检测研究，解决颤振检测系统在高分辨率卫星影像地面处理中的实际应用问题。为构建颤振检测系统，本研究开展颤振检测和优化的算法研究，提出了RPCS辐角相关匹配、K-means-SVM高频颤振检测误差消除算法、基于Harr小波的颤振数据优化等算法，开展了基于新型设备星载角位移传感器的高频颤振量检测研究。在颤振检测系统中，建立了多模型自适应切换的控制策略，探讨基于颤振补偿及质量评价体系的反馈闭环控制策略，并通过大量检测结果分析颤振的模型和长期变化趋势，为解决高分辨率遥感卫星颤振问题奠定基础。