Mask galaxy: Morphological segmentation of galaxies

Galaxy Morphology and Classification

Classifying galaxies into categories based on their structure has many practical applications in astronomy. In particular, large catalogues of classified galaxy images have been useful in many studies of the universe. However, one of the premier data sources in astronomy, the Sloan Digital Sky Survey (SDSS), does not provide classification information for the 50 million galaxy images it contains. As there are simply too many objects to classify manually, machine learning and classification algorithms are required to automate this process. This research applies the Support Vector Machine (SVM) algorithm to classify galaxy morphologies. The accuracy of the classification is measured on various categories of galaxies from the survey. A three class algorithm is presented that makes use of multiple SVMs. This algorithm is used to assign the categories of spiral, elliptical, and irregular galaxies.

The classification of galaxies according to their shape is a fundamental tool in astronomy. It is through classification schemes that astronomers build a deeper understanding of how galaxies form and evolve. This long-awaited book by one of the pioneers of the field provides a concise and up-to-date summary of current ideas about galaxy morphology and classification. This is the first book dedicated entirely to the shapes and classifications of galaxies. It introduces the most widely used schemes, explains how they have developed and what they can tell us about galaxies. We are also shown how very distant galaxies (seen with the Hubble Space Telescope, for instance) often defy standard classification schemes. Finally, we look at recent work on the use of computers to automatically classify digital images of galaxies. This topical volume provides graduate students and researchers with a unique and indispensable reference on the classification and shape of galaxies.

A galaxy classification system using Optimal Convolution Neural Network has been designed and developed. Broadly there are five different types of galaxies. They are elliptical, spiral, disk, lenticular and irregular. We have proposed an Optimal Deep Convolution Neural Network which learns the features of the different types of galaxies in the form of a set of filters and after the learning is over recognizes the type of the unknown galaxies. With this, the optimal point with data size as well as number of filters of neither overshooting nor undershooting is found out to get the optimal network complexity – which highly improves the performance evaluation. The performance evaluation of the CNN architecture used for galaxy classification in terms of its accuracy, precision, recall and f-score is quite satisfactory. Also the training and testing time is affordable.KeywordsDeep learningConvolution Neural NetworkGalaxyTypes of galaxiesPerformance evaluationHoldout MethodConfusion matrix

It is shown that the galaxies of the Byurakan classes 2 and 5, which belong to the morphological subtypes Sa, Sab, Sb, and Sbc (including the SB galaxies), are on average somewhat bluer than the galaxies of classes 3 and 4 of these same subtypes. Among galaxies of the classes 2, 5, and 2s, objects with excess ultraviolet radiation are found rather more frequently than among galaxies of classes 3 and 4. Whereas Sa, Sab, Sb, and Sbc galaxies of classes 3 and 4 are scarcely different from one another in their mean B -V and U -B, Sc and Scd galaxies of class 4, as also of classes 2 and 5, are on average bluer than class 3. In comparison with the latter, among Sc and Sod ~alaxies of classes 2, 4, and 5, there are rather more objects with negative U -B. These facts, together with the already known indicators of activity in galaxies of the classes 2, 4, 5, and 2s, in their turn point to active processes taking place in galaxies with diffuse, split, semistellar, or starlike nuclei.

High-resolution galaxy spectra contain much information about galactic physics, but the high dimensionality of these spectra makes it difficult to fully utilize the information they contain. We apply variational autoencoders (VAEs), a nonlinear dimensionality reduction technique, to a sample of spectra from the Sloan Digital Sky Survey (SDSS). In contrast to principal component analysis (PCA), a widely used technique, VAEs can capture nonlinear relationships between latent parameters and the data. We find that a VAE can reconstruct the SDSS spectra well with only six latent parameters, outperforming PCA with the same number of components. Different galaxy classes are naturally separated in this latent space, without class labels having been given to the VAE. The VAE latent space is interpretable because the VAE can be used to make synthetic spectra at any point in latent space. For example, making synthetic spectra along tracks in latent space yields sequences of realistic spectra that interpolate between two different types of galaxies. Using the latent space to find outliers may yield interesting spectra: in our small sample, we immediately find unusual data artifacts and stars misclassified as galaxies. In this exploratory work, we show that VAEs create compact, interpretable latent spaces that capture nonlinear features of the data. While a VAE takes substantial time to train (≈1 day for 48,000 spectra), once trained, VAEs can enable the fast exploration of large astronomical data sets.

Large catalogues of classified galaxy images have been useful in many studies of the universe in astronomy. There are too many objects to classify manually in the Sloan Digital Sky Survey, one of the premier data sources in astronomy. Therefore, efficient machine learning and classification algorithms are required to automate the classifying process. We propose to apply the Support Vector Machine (SVM) algorithm to classify galaxy morphologies and Krylov iterative methods to improve runtime of the classification. The accuracy of the classification is measured on various categories of galaxies from the survey. A three-class algorithm is presented that makes use of multiple SVMs. This algorithm is used to assign the categories of spiral, elliptical, and irregular galaxies. A selection of Krylov iterative solvers are compared based on their efficiency and accuracy of the resulting classification. The experimental results demonstrate that runtime can be significantly improved by utilizing Krylov iterative methods without impacting classification accuracy. The generalized minimal residual method (GMRES) is shown to be the most efficient solver to classify galaxy morphologies.

We present an analysis of the relative frequencies of different supernova (SN) types in spirals with various morphologies and in barred or unbarred galaxies. We use a well-defined and homogeneous sample of spiral host galaxies of 692 SNe from the Sloan Digital Sky Survey in different stages of galaxy-galaxy interaction and activity classes of nucleus. We propose that the underlying mechanisms shaping the number ratios of SNe types can be interpreted within the framework of interaction-induced star formation, in addition to the known relations between morphologies and stellar populations. We find a strong trend in behaviour of the NIa/NCC ratio depending on host morphology, such that early spirals include more Type Ia SNe. The NIbc/NII ratio is higher in a broad bin of early-type hosts. The NIa/NCC ratio is nearly constant when changing from normal, perturbed to interacting galaxies, then declines in merging galaxies, whereas it jumps to the highest value in post-merging/remnant galaxies. In contrast, the NIbc/NII ratio jumps to the highest value in merging galaxies and slightly declines in post-merging/remnant subsample. The interpretation is that the star formation rates and morphologies of galaxies, which are strongly affected in the final stages of interaction, have an impact on the number ratios of SNe types. The NIa/NCC (NIbc/NII) ratio increases (decreases) from star-forming to active galactic nuclei (AGN) classes of galaxies. These variations are consistent with the scenario of an interaction-triggered starburst evolving into AGN during the later stages of interaction, accompanied with the change of star formation and transformation of the galaxy morphology into an earlier type.

We perform an analysis of the luminosity functions (LFs) of two types of ringed galaxies – polar-ring galaxies and collisional ring galaxies – using data from the Sloan Digital Sky Survey (SDSS). Both classes of galaxies were formed as a result of interaction with their environment and they are very rare objects. We constructed LFs of galaxies by different methods and found their approximations by the Schechter function. The luminosity functions of both types of galaxies show a systematic fall-off at low luminosities. The polar structures around bright (Mr ≤ −20m) and red (g − r > +0.8) galaxies are about twice as common as around blue ones. The LF of collisional rings is shifted towards brighter luminosities compared to polar-ring galaxies. We analysed the published data on the ringed galaxies in several deep fields and confirmed the increase in their volume density with redshift: up to z ∼ 1 their density grows as (1 + z)m, where m ≳ 5.

From the early stages of Astronomy, the classification of galaxies has been a conundrum that has left astrophysicists in a situation of quandary. Although, previous methods did a phenomenal job in classifying galaxies but while analysing them, certain inefficiencies had been revealed which cannot be overlooked. The objective had been to conduct an analysis of different types of machine learning techniques that have been used to classify galaxies. This analysis had been conducted on the basis of different attributes taken for different types of classification of galaxies. A method had been proposed to classify galaxies with higher accuracy than previous methods. The configuration for the literature analysis used datasets such as ESO-LV and SDSS and discussed the antecedent techniques for classifying galaxies. It had been inferred that a Convolutional Neural Network with certain data augmentation for irregular Galaxies (Irr) gives the best result of all the algorithms that have been discussed in the literature and its analysis. Owing to the aforementioned, an implementation accentuating the use of deep learning algorithms with certain Data Augmentation techniques and certain different activation functions, named daMCOGCNN (data augmentation-based MOrphological Classifier Galaxy Using Convolutional Neural Networks) had been proposed for morphological classification of galaxies. The dataset comprises of 4614 images from SDSS Image Gallery, Galaxy Zoo challenge and Hubble Image Gallery. The efficient implementation of this method gave a testing accuracy of approximately 98% and 97.92% accuracy had been achieved on a dataset taken from different websites such as AstroBin and other such sources. The model introduced here outperforms its earlier contemporaries.

There are several supervised machine learning methods used for the application of automated morphological classification of galaxies; however, there has not yet been a clear comparison of these different methods using imaging data, or an investigation for maximizing their effectiveness. We carry out a comparison between several common machine learning methods for galaxy classification [Convolutional Neural Network (CNN), K-nearest neighbour, logistic regression, Support Vector Machine, Random Forest, and Neural Networks] by using Dark Energy Survey (DES) data combined with visual classifications from the Galaxy Zoo 1 project (GZ1). Our goal is to determine the optimal machine learning methods when using imaging data for galaxy classification. We show that CNN is the most successful method of these ten methods in our study. Using a sample of ∼2800 galaxies with visual classification from GZ1, we reach an accuracy of ∼0.99 for the morphological classification of ellipticals and spirals. The further investigation of the galaxies that have a different ML and visual classification but with high predicted probabilities in our CNN usually reveals the incorrect classification provided by GZ1. We further find the galaxies having a low probability of being either spirals or ellipticals are visually lenticulars (S0), demonstrating that supervised learning is able to rediscover that this class of galaxy is distinct from both ellipticals and spirals. We confirm that ∼2.5 per cent galaxies are misclassified by GZ1 in our study. After correcting these galaxies’ labels, we improve our CNN performance to an average accuracy of over 0.99 (accuracy of 0.994 is our best result).

We extend a recently developed galaxy morphology classification method, Quantitative Multiwavelength Morphology (QMM), to connect galaxy morphologies to their underlying physical properties. The traditional classification of galaxies approaches the problem separately through either morphological classification or, in more recent times, through analysis of physical properties. A combined approach has significant potential in producing a consistent and accurate classification scheme as well as shedding light on the origin and evolution of galaxy morphology. Here we present an analysis of a volume limited sample of 31703 galaxies from the fourth data release of the Sloan Digital Sky Survey. We use an image analysis method called Pixel-z to extract the underlying physical properties of the galaxies, which is then quantified using the concentration, asymmetry and clumpiness (CAS) parameters. The galaxies also have their multiwavelength morphologies quantified using QMM, and these results are then related to the distributed physical properties through a regression analysis. We show that this method can be used to relate the spatial distribution of physical properties with the morphological properties of galaxies.

Context.The classification of galaxy morphology is among the most active fields in astronomical research today. With the development of artificial intelligence technology, deep learning is a useful tool in the classification of the morphology of galaxies and significant progress has been made in this domain. However, there is still some room for improvement in terms of classification accuracy, automation, and related issues.Aims.Convolutional vision Transformer (CvT) is an improved version of the Vision Transformer (ViT) model. It improves the performance of the ViT model by introducing a convolutional neural network (CNN). This study explores the performance of the CvT model in the area of galaxy morphology classification.Methods.In this work, the CvT model was applied, for the first time, in a five-class classification task of galaxy morphology. We added different types and degrees of noise to the original galaxy images to verify that the CvT model achieves good classification performance, even in galaxy images with low signal-to-noise ratios (S/Ns). Then, we also validated the classification performance of the CvT model for galaxy images at different redshifts based on the low-redshift dataset GZ2 and the high-redshift dataset Galaxy Zoo CANDELS. In addition, we visualized and analyzed the classification results of the CvT model based on the t-distributed stochastic neighborhood -embedding (t-SNE) algorithm.Results.We find that (1) compared with other five-class classification models of galaxy morphology based on CNN models, the average accuracy, precision, recall, and F1_score evaluation metrics of the CvT classification model are all higher than 98%, which is an improvement of at least 1% compared with those based on CNNs; (2) the classification visualization results show that different categories of galaxies are separated from each other in multi-dimensional space.Conclusions.The application of the CvT model to the classification study of galaxy morphology is a novel undertaking that carries important implications for future studies.

(abridged) We present measurements of galaxy-galaxy weak lensing from early commissioning imaging data from the Sloan Digital Sky Survey (SDSS). We measure a mean tangential shear around a stacked sample of foreground galaxies in three bandpasses out to angular radii of 600'', detecting the shear signal at very high statistical significance. The shear profile is well described by a power-law. A variety of rigorous tests demonstrate the reality of the gravitational lensing signal and confirm the uncertainty estimates. We interpret our results by modeling the mass distributions of the foreground galaxies as approximately isothermal spheres characterized by a velocity dispersion and a truncation radius. The velocity dispersion is constrained to be 150-190 km/s at 95% confidence (145-195 km/s including systematic uncertainties), consistent with previous determinations but with smaller error bars. Most strikingly, our detection of shear at large angular radii sets a 95% confidence lower limit on the outer scale of >150'', corresponding to a physical radius of 275/h kpc, implying that the dark halos of typical luminous galaxies extend to very large radii. We also present a preliminary determination of the galaxy-mass correlation function finding a correlation length similar to the galaxy autocorrelation function and consistency with a low matter density universe with modest bias. The full SDSS will cover an area 44 times larger and provide spectroscopic redshifts for the foreground galaxies, making it possible to greatly improve the precision of these constraints, to measure additional parameters such as halo shape and halo concentration, and to measure the properties of dark matter halos separately for many different classes of galaxies.

We measure the relation between galaxy luminosity and disk circular velocity (the Tully-Fisher [TF] relation), in the g, r, i, and z bands, for a broadly selected sample of galaxies from the Sloan Digital Sky Survey, with the goal of providing well-defined observational constraints for theoretical models of galaxy formation. The input sample of 234 galaxies has a roughly flat distribution of absolute magnitudes in the range -18.5 > Mr > -22, and our only morphological selection is an isophotal axis ratio cut b/a < 0.6 to allow accurate inclination corrections. Long-slit spectroscopy from the Calar Alto and MDM observatories yields usable Hα rotation curves for 162 galaxies (69%), with a representative color and morphology distribution. We define circular velocities V80 by evaluating the rotation curve at the radius containing 80% of the i-band light. Observational errors, including estimated distance errors due to peculiar velocities, are small compared to the intrinsic scatter of the TF relation. The slope of the forward TF relation steepens from -5.5 ± 0.2 mag (log10 km s-1)-1 in the g band to -6.6 ± 0.2 mag (log10 km s-1)-1 in the z band. The intrinsic scatter is σ ≈ 0.4 mag in all bands, and residuals from either the forward or inverse relations have an approximately Gaussian distribution. We discuss how Malmquist-type biases may affect the observed slope, intercept, and scatter. The scatter is not dominated by rare outliers or by any particular class of galaxies, although it drops slightly, to σ ≈ 0.36 mag, if we restrict the sample to nearly bulgeless systems. Correlations of TF residuals with other galaxy properties are weak: bluer galaxies are significantly brighter than average in the g-band TF relation but only marginally brighter in the i band; more concentrated (earlier type) galaxies are slightly fainter than average, and the TF residual is virtually independent of half-light radius, contrary to the trend expected for gravitationally dominant disks. The observed residual correlations do not account for most of the intrinsic scatter, implying that this scatter is instead driven largely by variations in the ratio of dark to luminous matter within the disk galaxy population.