中文摘要：

在常规恒星日滤波的基础上，提出一种顾及分段相似性的恒星日滤波优化算法。在构建滤波算子之前，首先确定用于构建滤波的坐标残差时间序列与地震当天坐标残差时间序列的分段相似性，然后根据相似性度量指标的大小确定权重后再进行滤波。研究结果表明：利用本文提出的顾及分段相似性的恒星日滤波优化方法能够显著改善单历元定位的精度和可靠性，克服常规滤波方法存在的缺陷，有利于恒星日滤波的无人工干预批处理。尤其是通用影像质量评价指数UIQI方法顾及了时间序列间的相关系数以及序列自身的均值和方差，比其他相似性度量更加科学合理，建议采用UIQI作为恒星日滤波优化算法的相似性度量。

英文摘要：

Based on the traditional sidereal filtering method, an improved sidereal filtering method is put forward, accounting for segments＇ similarity, namely before building filtering operator. Firstly determine the segments＇ similarity of the coordinates time series between earthquake day and before that day or after that day, and then based on the similarity measure index to identify weight of each day for filtering. Different similarity measure indicators are analyzed. After that the universal image quality index （UIQI） method is recommended as the similarity measure for the improved sidereal filtering. It~s more scientific and reasonable than other similarity measure which considers the correlation coefficient between two series, own mean value and variance as well. The results show that the improved sidereal filtering can effectively improve the single epoch positioning accuracy and reliability, completely overcome the defects of traditional filtering method, is beneficial to batch processing without manual intervention.