一种新的基于2维傅里叶谱图像的恒星光谱特征提取方法和深度网络分类应用

天体光谱分类是天文学研究的重要内容之一,其关键是从光谱数据中选择和提取对分类识别最有效的特征构建特征空间.提出一种新的基于2维傅里叶谱图像的特征提取方法,并应用于LAMOST (the Large Sky Area Multi-Object Fiber Spectroscopic Telescope)恒星光谱数据的分类研究中.光谱数据来源于LAMOST Data Release 5(DR5),选取30000条F、 G和K型星光谱数据,利用短时傅里叶变换(Short-Time Fourier Transform, STFT)将1维光谱数据变换成2维傅里叶谱图像,对得到的2维傅里叶谱图像采用深度卷积网络模型进行分类,得到的分类准确率是92.90%.实验结果表明通过对LAMOST恒星光谱数据进行STFT可得到光谱的2维傅里叶谱图像,谱图像构成了新的光谱数据特征和特征空间,新的特征对于光谱数据分类是有效的.此方法是对光谱分类的一种全新尝试,对海量天体光谱的分类和挖掘处理有一定的开创意义.     

基于SE-Inception-v3的星系形态分类模型

随着天文探测技术的快速发展, 海量的星系图像数据不断产生, 能够及时高效地对星系图像进行形态分类对研究星系的形成与演化至关重要. 针对传统的星系形态分类模型特征选择困难、分类速度慢、准确率受限等难题, 提出一种以Inception-v3神经网络为主干结构, 融合压缩激励(Squeeze and Excitation Network, SE)通道注意力机制的星系形态分类模型. 该模型在斯隆数字巡天(Sloan Digital Sky Survey, SDSS)样本的测试集准确率高达99.37%. 旋涡星系、圆形星系、中间星系、雪茄状星系与侧向星系的F1值分别为99.33%、99.58%、99.33%、99.41%与99.16%. 该模型与Inception-v3、MobileNet (Mobile Neural Network)和ResNet (Residual Neural Network)网络模型相比, SE-Inception-v3宽度和深度优势表现出更强的特征提取能力, 可以高效识别不同形态的星系, 为未来大型巡天计划的大规模星系形态分类问题提供了一种新方法.     

全天相机在多设备巡天中的应用

天文观测站夜天空星像星等信息和天区分布信息可用于指导多设备巡天观测.建立全天相机监测系统(Monitoring all-sky system)对本地天区夜天空实时监测,获取的监测图像需要有效的方法进行处理以提取全天图像星像信息.由于全天图像视场大和高阶扭曲的影响,采用天顶等距投影与多项式函数组合的方法计算图像的底片常数.天文定位的均方根残差约为0.15个像素.通过对图像中亮星部分测光得到的星等差,改正大气消光误差.最后使用HEALPix (Hierarchical Equal Area isoLatitude Pixelation)方法实现天区划分和每个天区可观测极限星等值的存储.     

Processing Method of Night-Time Cloudiness for Astronomical Site Selection

The cloudiness is one of the most important factors which affect the quality of an astronomical site, the monitoring and processing of the night- time cloudiness are especially important. The ground-based cloudiness camera is adopted to carry out the monitoring of the all-sky cloudiness, the images taken need to be processed by means of an effective method so as to quantize the cloudiness. The night-time cloudiness images are seriously affected by the moon- light, and therefore, the night-time cloudiness images are processed by dividing them into the moonlight and moonless two sorts. In the light of the condition of moonless night, the processing method of night-time cloudiness is given. The positioning and photometry of the bright stars in the image are conducted to determine their magnitude differences. By referring to the magnitude differences of the bright stars in the clear-night image, the probability of the bright stars of which the magnitude differences are lower than the threshold value are regarded as the probability standard of clear nights. Three sorts of images are selected to test the method. The cloudiness is determined, and the effect of the threshold condition on the result is analyzed. Finally, the applicable range and uncertainty of the method are discussed.     

On the reduction and presentation of data in astronomical two-channel photopolarimetry

Many different methods exist for reducing data obtained when an astronomical source is studied with a two-channel polarimeter, such as a Wollaston prism system. This paper presents a rigorous method of reducing the data from raw aperture photometry, and evaluates errors both by a statistical treatment, and by propagating the measured sky noise from each frame. The reduction process performs a hypothesis test for the presence of linear polarization. The probability of there being a non-zero polarization is obtained, and the best method of obtaining the normalized Stokes Parameters is discussed. Point and interval estimates are obtained for the degree of linear polarization, which is subject to positive bias; and the polarization axis is found.     

Unbiased image reconstruction as an inverse problem

An unbiased method for improving the resolution of astronomical images is presented. The strategy at the core of this method is to establish a linear transformation between the recorded image and an improved image at some desirable resolution. In order to establish this transformation only the actual point spread function and a desired point spread function need be known. No image actually recorded is used in establishing the linear transformation between the recorded and improved image.
This method has a number of advantages over other methods currently in use. It is not iterative, which means it is not necessary to impose any criteria, objective or otherwise, to stop the iterations. The method does not require an artificial separation of the image into 'smooth' and 'point-like' components, and thus is unbiased with respect to the character of structures present in the image. The method produces a linear transformation between the recorded image and the deconvolved image, and therefore the propagation of pixel-by-pixel flux error estimates into the deconvolved image is trivial. It is explicitly constrained to preserve photometry and should be robust against random errors.     

Studying the dwarf galaxies in nearby groups of galaxies: Spectroscopic and photometric data

Galaxy evolution by interaction‐driven transformation is probably highly efficient in groups of galaxies. Dwarf galaxies with their shallow potential are expected to reflect the interaction most prominently in their observable structure. The major aim of this series of papers is to establish a data base which allows to study the impact of group interaction onto the morphology and star‐forming properties of dwarf galaxies. Firstly, we present our selection rules for target groups and the morphological selection method of target dwarf member candidates. Secondly, the spectroscopic follow‐up observations with the HET are presented. Thirdly, we applied own reduction methods based on adaptive filtering to derive surface photometry of the candidates. The spectroscopic follow‐up indicate a dwarf identification success rate of roughly 55 %, and a group member success rate of about 33 %. A total of 17 new low surface‐brightness members is presented. For all candidates, total magnitudes, colours, and light distribution parameters are derived and discussed in the context of scaling relations. We point out short comings of the SDSS standard pipeline for surface photometry for these dim objects. We conclude that our selection strategy is rather efficient to obtain a sample of dim, low surface brightness members of groups of galaxies within the Virgo super‐cluster. The photometric scaling relation in these X‐ray dim, rather isolated groups does not significantly differ from those of the galaxies within the local volume. (© 2014 WILEY‐VCH Verlag GmbH & Co. KGaA, Weinheim)     

面向单脉冲信号分类的集成特征选择与评价

受大量射频干扰信号影响, 快速从海量观测数据中准确识别出单脉冲信号已成为天文数据处理的一项重要任务, 而设计和提取有效数据特征, 是利用机器学习进行单脉冲信号高效识别的决定因素. 针对如何选择最优特征, 进而提升单脉冲信号的分类精度这一关键问题, 设计了面向单脉冲信号分类的集成特征选择方法. 方法首先混合单脉冲信号的参数特征、统计特征和抽象特征, 然后分别利用5种单一特征选择方法选出各自的最优特征集, 最后利用贪心策略对5种单一方法获取的最优特征集进行集成筛选, 获取最优集成特征集. 实验表明, 最优特征集合既包含统计特征也包含抽象特征. 在相同特征数量下, 利用集成特征选择比单一特征选择能获得更高的模型精度, 可使F1值最高提升1.8%. 在海量数据背景下, 集成特征选择对减少特征数量、提升分类性能和加快数据处理速度具有重要作用.     

Automatic morphological classification of galaxy images

We describe an image analysis supervised learning algorithm that can automatically classify galaxy images. The algorithm is first trained using manually classified images of elliptical, spiral and edge-on galaxies. A large set of image features is extracted from each image, and the most informative features are selected using Fisher scores. Test images can then be classified using a simple Weighted Nearest Neighbour rule such that the Fisher scores are used as the feature weights. Experimental results show that galaxy images from Galaxy Zoo can be classified automatically to spiral, elliptical and edge-on galaxies with an accuracy of ∼90 per cent compared to classifications carried out by the author. Full compilable source code of the algorithm is available for free download, and its general-purpose nature makes it suitable for other uses that involve automatic image analysis of celestial objects.     

EfficientNetV2-S-Triplet7: 一种改进的星系newline 形态学分类算法

星系的形态与星系的形成和演化息息相关, 其形态学分类是星系天文学后续研究的重要一环. 当前海量天文观测数据的出现使得天文数据自动分析方法越来越得到重视, 针对此问题, 利用先进的深度学习骨干网络EfficientNetV2, 分析不同的注意力机制类型和使用节点对网络性能的影响, 构建了一种命名为EfficientNetV2-S-Triplet7 (即在EfficientNetV2-S stage7的$1\times1$卷积层后加入Triplet模块)的改进算法模型来实现星系形态学的自动分类. 使用第二期星系动物园(Galaxy Zoo 2, GZ2)中超过24万张的测光图像作为初始数据进行实验测试. 在对数据进行预处理时采取了尺寸抖动、翻转、色彩畸变等图像增强手段来解决图像数量的不平衡问题. 在同一系列经典和前沿的深度学习算法模型AlexNet、ResNet-34、MobileNetV2、RegNet进行对比实验后, 得出EfficientNetV2-S-Triplet7算法在分类准确率、查全率和F1分数等指标上具有最好的测试结果. 在9375张测试图像中的3项指标值分别可达到89.03%、90.21%、89.93%, 查准率达到89.69%, 在其他模型中排在第3位. 该结果表明将EfficientNetV2-S-Triplet7算法应用于大规模星系数据的形态学分类任务中有很好的效果.     

A deep learning approach for detecting candidates of supernova remnants

Detecting supernova remnant(SNR) candidates in the interstellar medium is a challenging task because SNRs have weak radio signals and irregular shapes. The use of a convolutional neural network is a deep learning method that can help us extract various features from images. To extract SNRs from astronomical images and estimate the positions of SNR candidates, we design the SNR-Net model composed of a training component and a detection component. In addition, transfer learning is used to initialize the network parameters, which improves the speed and accuracy of network training. We apply a T-T plot(of the different brightness temperatures of map pixels at two different frequencies) to calculate the spectral index of SNR candidates. To accelerate the scientific computing process, we take advantage of innovative hardware architecture, such as deep learning optimized graphics processing units, which increases the speed of computation by a factor of 5. A case study suggests that SNR-Net may be applicable to detecting extended sources in the images automatically.     

A Comparison of BBN,ADTree and MLP in separating Quasars from Large Survey Catalogues

We compare the performance of Bayesian Belief Networks (BBN), Multilayer Perception (MLP) networks and Alternating Decision Trees (ADtree) on separating quasars from stars with the database from the 2MASS and FIRST survey catalogs. Having a train- ing sample of sources of known object types, the classifiers are trained to separate quasars from stars. By the statistical properties of the sample, the features important for classifica- tion are selected. We compare the classification results with and without feature selection. Experiments show that the results with feature selection are better than those without feature selection. From the high accuracy found, it is concluded that these automated methods are robust and effective for classifying point sources. They may all be applied to large survey projects (e.g. selecting input catalogs) and for other astronomical issues, such as the parame- ter measurement of stars and the redshift estimation of galaxies and quasars.     

Random forest algorithm for classification of multiwavelength data

We introduced a decision tree method called Random Forests for multiwavelength data classification. The data were adopted from different databases, including the Sloan Digital Sky Survey (SDSS) Data Release five, USNO, FIRST and ROSAT.We then studied the discrimination of quasars from stars and the classification of quasars,stars and galaxies with the sample from optical and radio bands and with that from optical and X-ray bands. Moreover, feature selection and feature weighting based on Random Forests were investigated. The performances based on different input patterns were compared. The experimental results show that the random forest method is an effective method for astronomical object classification and can be applied to other classification problems faced in astronomy. In addition, Random Forests will show its superiorities due to its own merits, e.g. classification, feature selection, feature weighting as well as outlier detection.     

1.2米地平式望远镜指向模型的研究 (1.全天指向模型的建立)

介绍了云南天文台1.2米地平式望远镜用于天文观测和图像采集处理的方法,建立了新的、独特的全天指向模型,大大提高了该望远镜的指向精度,达到1″,并在多年的实际应用中得到验证。     

How the extinction of extragalactic background light affects surface photometry of galaxies, groups and clusters

The faint regions of galaxies, groups and clusters hold important clues about how these objects formed, and surface photometry at optical and near-infrared wavelengths represents a powerful tool for studying such structures. Here, we identify a hitherto unrecognized problem with this technique, related to how the night sky flux is typically measured and subtracted from astronomical images. While most of the sky flux comes from regions between the observer and the target object, a small fraction – the extragalactic background light (EBL) – comes from behind. We argue that since this part of the sky flux can be subjected to extinction by dust present in the galaxy/group/cluster studied, standard reduction procedures may lead to a systematic oversubtraction of the EBL. Even very small amounts of extinction can lead to spurious features in radial surface brightness profiles and colour maps of extended objects. We assess the likely impact of this effect on a number of topics in extragalactic astronomy where very deep surface photometry is currently attempted, including studies of stellar haloes, starburst host galaxies, disc truncations and diffuse intragroup/intracluster light. We argue that EBL extinction may provide at least a partial explanation for the anomalously red colours reported for the haloes of disc galaxies and for the hosts of local starburst galaxies. EBL extinction effects also mimic truncations in discs with unusually high dust opacities, but are unlikely to be the cause of such features in general. Failure to account for EBL extinction can also give rise to a non-negligible underestimate of intragroup and intracluster light at the faintest surface brightness levels currently probed. Finally, we discuss how EBL extinction effects may be exploited to provide an independent constraint on the surface brightness of the EBL, using a combination of surface photometry and direct star counts.     

Difference imaging photometry of blended gravitational microlensing events with a numerical kernel

The numerical kernel approach to difference imaging has been implemented and applied to gravitational microlensing events observed by the PLANET collaboration. The effect of an error in the source-star coordinates is explored and a new algorithm is presented for determining the precise coordinates of the microlens in blended events, essential for accurate photometry of difference images. It is shown how the photometric reference flux need not be measured directly from the reference image but can be obtained from measurements of the difference images combined with the knowledge of the statistical flux uncertainties. The improved performance of the new algorithm, relative to isis2 , is demonstrated.     

Low-cost high performance distributed data storage for multi-channel observations

The New Vacuum Solar Telescope (NVST) is a 1-m solar telescope that aims to observe the fine structures in both the photosphere and the chromosphere of the Sun. The observational data acquired simultaneously from one channel for the chromosphere and two channels for the photosphere bring great challenges to the data storage of NVST. The multi-channel instruments of NVST, including scientific cameras and multi-band spectrometers, generate at least 3 terabytes data per day and require high access performance while storing massive short-exposure images. It is worth studying and implementing a storage system for NVST which would balance the data availability, access performance and the cost of development. In this paper, we build a distributed data storage system (DDSS) for NVST and then deeply evaluate the availability of real-time data storage on a distributed computing environment. The experimental results show that two factors, i.e., the number of concurrent read/write and the file size, are critically important for improving the performance of data access on a distributed environment. Referring to these two factors, three strategies for storing FITS files are presented and implemented to ensure the access performance of the DDSS under conditions of multi-host write and read simultaneously. The real applications of the DDSS proves that the system is capable of meeting the requirements of NVST real-time high performance observational data storage. Our study on the DDSS is the first attempt for modern astronomical telescope systems to store real-time observational data on a low-cost distributed system. The research results and corresponding techniques of the DDSS provide a new option for designing real-time massive astronomical data storage system and will be a reference for future astronomical data storage.     

VLBI观测信号相关处理中的小波保角变换

Daubechies小波可以表征为幂函数或e指数函数的形式 ,保持相关相位条纹的幅角不变。在VLBI观测信号相关处理过程中引入了小波分析方法 ,以探讨小波理论在VLBI技术中的应用特点。两个台站的VLBI观测信号经过时间补偿和干涉条纹旋转后 ,利用静态小波算法在线性空间中的保角变换特征可提取VLBI信号的干涉相位特征 ,并完整保留相位的时序关系。     

Astronomical Image Co-adding Method based on Wide-field Image Deconvolution

Due to the atmospheric turbulence, the static aberration, tracking and pointing errors of telescopes, the point spread functions (PSFs) in different fields of view are different. Meanwhile, there are different PSFs in the images obtained by different telescopes. The quality of co-adding image is limited by the image with the poorest quality, and finally the resolution and sensitivity of the quad-channel telescope will also be affected. Dividing the image into some regions with the same type of PSF, and deconvolving these regions can improve the quality of the co-adding image. According to this theory, an image restoration algorithm based on the PSF clustering is proposed. Firstly, this paper makes the PSF clustering analysis by using Self-Organizing Maps, and makes the image segmentation based on the result of the PSF clustering analysis, then using the clustered PSFs to make deconvolutions on the sub-images. Then, the restored sub-images after deconvolution are joined together. Finally, by through the image registration and co-adding, the image with a high signal to noise ratio can be obtained. The result shows that the signal to noise ratio of the astronomical images are improved with our method, and the detection capability on faint stars is also improved.