项目名称: 空间目标观测自动化的关键技术研究
项目编号: No.11503057
项目类型: 青年科学基金项目
立项/批准年度: 2016
项目学科: 天文学、地球科学
项目作者: 杨文波
作者单位: 中国科学院国家天文台长春人造卫星观测站
项目金额: 23万元
中文摘要: 精密测量型地基光电望远镜的主要作用是为空间目标的精密定轨服务。随着观测技术的发展,地基光电望远镜运行模式逐渐趋于以观测自动化、记录数字化、传输网络化为特点的无人职守运行,而快速准确的目标识别和稳定的目标跟踪是实现观测自动化的关键。本项目将要解决的核心问题是研究基于计算机图像视觉的目标识别与跟踪技术,来提高系统观测自动化程度,实现系统的无人职守运行模式。首先,通过对空间目标静态特征和运动特征学习,设计出基于Logistic回归模型的分类器用于空间目标和背景恒星自动分类;其次,提出了一种基于均值漂移与特征匹配相结合的跟踪算法用于目标的精密定轨。项目研究的意义在于它把观测员从繁重观测任务中解放出来,提高台网的运行效率,实现高效、低成本的运行模式。
中文关键词: 目标识别;目标跟踪;Logistic回归模型;均值漂移;特征匹配
英文摘要: A ground-based optical telescope with precision measurement functions mainly for the precise orbit determination on the space target. With the development of observation technology, the operation mode of ground optical telescope tends to be unattended operation featuring observation automation, recording digitalization, transmission network. Fast and accurate target identification and stable target tracking is the key to observation automation. The project centers on the technology of target identification and tracking based on computer vision image to improve observation automation, thus an unattended operation mode is achieved. First of all, a classifier based on Logistic regression model is designed for automatic classification of a space target and background stars by a study on the static and motional feature of a space target. Next, a tracking algorithm based on the combination of Mean Shift and feature matching is proposed for precise orbit determination on a space target. The project serves to free observers from onerous observation routines, speed the operation of station networks, and achieve a high-efficiency and low-cost operation mode.
英文关键词: target identification;target tracking;Logistic regression model;mean shift;feature matching