抚仙湖1米新真空太阳望远镜（New Vacuum Solar Telescope，简称NVST）是我国新一代地基太阳观测设备之一，其中Halpha成像系统是进行太阳色球观测的主要终端。Halpha实测数据（特别是偏带观测数据）中明显存在由薄膜干涉引起的等厚条纹，而常规平场校正有时无法将其消除，并在后期图像高分辨率统计重建过程中造成严重影响，本人在硕士阶段的工作便是针对成像系统的干涉条纹展开相关的特征分析和去除工作。论文首先介绍了天文薄型CCD产生干涉条纹的基本原理（即薄膜干涉），并可视化分析了薄膜厚度、入射角度以及折射率等参数对干涉条纹的影响。同时对NVST及其多波段成像系统进行了简介，强调了望远镜镜筒弯沉对成像系统所造成的影响（例如系统光轴的摆动）。其次，我们集中分析了Halpha偏带观测数据中干涉条纹的空间分布和时间演化特性。分析发现：干涉条纹的间距或二维空间分布样式随时间固定不变，但条纹的强度随时间变化明显（4个小时左右可见度增加9倍），这是导致常规平场校正无法将其消除的主要原因。根据该现象，结合已有的光轴变化实测结果，推断造成上述时变特性的主要原因是观测过程中入射光的强度以及入射光与CCD靶面位置的相对角度随时间发生了变化，而且产生条纹的薄膜结构与靶面位置的距离比较近（应在40mm以内），为后续从硬件改造的角度抑制干涉条纹提供参考。在上述工作的基础上，我们提出并尝试了两种降低条纹可见度的方法。首先对探测器(pco.4000型CCD)光敏介质的前封窗进行了改造，通过改造前封窗楔角至2度的方案来调制条纹间距至CCD像元尺度，改造之后小范围视场内可识别出可见度仅为0.6%的少量干涉条纹。其次基于图像滤波技术提取干涉条纹图样并生成所谓“条纹平场”，分别采用和比较了傅里叶频域滤波、小波变换滤波和空域中值滤波三种方法，结果基本相同。观测数据再次进行“条纹平场”矫正之后，条纹消除效果明显：可见度可降低8倍左右（即由之前的4.7%降低为0.6%以下）。同时我们也指出多幅图像的积分（累加）可以有效降低精细太阳结构在条纹平场提取中的影响，并给出累加时间的经验值约为20分钟，为下一步的从高分辨统计重建算法角度抑制干涉条纹的工作提供了可借鉴的经验。最后我们讨论了上述滤波方法的局限性，并尝试利用Chae（2004）提出的新算法进一步减弱“条纹平场”中的太阳结构痕迹。我们给出了该方法在NVST上的初步实测结果，为具有强对比结构的活动区数据获取“准实时平场”来修正干涉条纹提供了新的思路。
其他摘要
The 1-meter New Vacuum Solar Telescope is a new generation ground-based solar observation facility of China, located at the FuXian-Lake Solar Observatory of the Yunnan Observatories. One of its instrumentations is a five-channel high resolution image system, while the Halpha-channel is used as the traditional way for solar chromospheric observation. Using a pco.4000 CCD as the detector, we usually observe complicated fringes resulting from thin-film interference in scientific data. These fringes can not be eliminated completely by flat-field calibration and even be intensified after image reconstruction processes.In order to solve this problem,this paper performed study on interference fringes in images observed by the imaging system of NVST.From about 4-hour continuous Halpha off-band observation, we study the temporal-variation of the fringe pattern and amplitude. We find that in contrast to the stable pattern, the fringe amplitude obviously changed with time. The visibility increases 9 times over 4 hours. We speculate it is due to the temporal-variation of the incident light intensity and the direction with respect to the CCD position.Consequently, we suggest two methods to modify or eliminate fringes. On one hand, we suggest to change the CCD inner-glass-window to a wedge-shape with the wedge-angle larger about 2 degree in order to reduce the fringe-interval to the CCD pixel size. It is shown that this change can eliminate the majority fringes. Only few fringes with the visibility of 0.6% can be found in 1/24 field-of-view. On the other hand, we make efforts to eliminate fringes by correcting the data using a so-called "fringe-flat" again after the normal data reduction(flat field and dark field modification). The "fringe-flat" is extracted from the data by using a Fourier filtering technique in frequency domain, a Wavelet analysis technique, a Median-filter in space domain, respectively and all produce a very similar result. We find that corrected data using a "fringe-flat" produce a substantial reduction in the fringe amplitudes with the visibility decreasing to about 1/8. However we have to point out that there inevitably exist some vestiges of long-lived and large scale solar structures in the "fringe-flat". In return, the intensity of these structures will be somehow modified. Consequently it is suggested to integrate the data over a certain time interval in order to enhance the fringes and smooth the high-frequency solar structures. The empirical suggestion of the maximum interval is about 20 minutes.
修改评论