Submitted by Ricardo Cedraz Duque Moliterno (ricardo.moliterno@uefs.br) on 2017-02-16T20:59:48Z No. of bitstreams: 1 Dissertacao_stanley.pdf: 4369040 bytes, checksum: ed63301078418b326241edb9682820e8 (MD5) / Made available in DSpace on 2017-02-16T20:59:48Z (GMT). No. of bitstreams: 1 Dissertacao_stanley.pdf: 4369040 bytes, checksum: ed63301078418b326241edb9682820e8 (MD5) Previous issue date: 2016-08-26 / Interacting peculiar galaxies are objects that still require further studies because they play an important role in the processes of the evolution of galaxies. The knowledge of its location and properties constitutes a great benefit to the astronomical community. In this dissertation, we present an automatic method for identifying and classifying images of interacting galaxies at low redshifts for the Southern Hemisphere, based on the properties of stellar and interstellar extinction distribution as well as using a pattern recognition software called ?Wndchrm? on images from the 2MASS survey, in the near infrared for the filters: J, H and KS. The training phase was made with images of known interacting galaxies from the Arp &Madore Catalogue, Categories 1 and 2. After training, a validation was performed using images of a region of the sky with 573 square degrees, obtaining a hit of, approximately, 73% in the identification of galaxies identified by visual inspection as interacting. This rate can reach up 88% considering the comparison with previous know galaxy pairs of Category 2 of Arp &Madore Catalogue. The procedure was performed for an area of about 17.836 square degrees of the Southern Hemisphere, finding at least several hundred galaxy pairs as yet uncatalogued. / As gal?xias peculiares interagentes s?o objetos que ainda carecem de maiores estudos. Elas desempenham um papel importante nos processos de forma??o e de evolu??o das gal?xias. O conhecimento de sua localiza??o e propriedades, constitui um grande benef?cio para a comunidade astron?mica. Nesta disserta??o, apresentamos um m?todo autom?tico para identificar e classificar imagens de gal?xias interagentes a baixos redshifts para o Hemisf?rio Sul. Para tal, estabelecemos crit?rios baseados na distribui??o de estrelas e extin??o interestelar, assim como de objetos identificados no Cat?logo de fontes extensas do 2MASS. Usamos um software de reconhecimento de padr?es chamado Wndchrm em imagens do grande levantamento 2MASS no infravermelho pr?ximo para os filtros: J, H e Ks. A fase de treinamento foi feita com imagens de gal?xias interagentes conhecidas, do Cat?logo de Arp&Madore, das categorias 1 e 2. Ap?s o treinamento foi realizada uma valida??o com imagens de uma regi?o do c?u de 573 graus quadrados, obtendo uma taxa de acerto de, aproximadamente 73% na identifica??o de gal?xias previamente identificadas de forma visual como interagentes. Essa taxa aumenta para, aproximadamente 88% levando em conta pares da Categoria 2 previamente identificados no Cat?logo de Arp&Madore como pertencentes ? essa categoria. Executamos, o procedimento para uma ?rea de, aproximadamente 17.836 graus quadrados do Hemisf?rio Sul, encontrando, ao menos, v?rias centenas de pares de gal?xias ainda n?o catalogados.

Submitted by Jadson Francisco de Jesus SILVA (jadson@uefs.br) on 2018-01-29T23:16:03Z No. of bitstreams: 1 Adilson_Almirante_biblioteca.pdf: 10152803 bytes, checksum: 30e4450d727a2f15e626c26a05c598b6 (MD5) / Made available in DSpace on 2018-01-29T23:16:03Z (GMT). No. of bitstreams: 1 Adilson_Almirante_biblioteca.pdf: 10152803 bytes, checksum: 30e4450d727a2f15e626c26a05c598b6 (MD5) Previous issue date: 2017-08-29 / In this work, we present an alternative method for clustering of galaxies, based on the data of galaxies available in large scale astronomical surveys. We have used the Friends of Friends (FoF) algorithm, as the basis of our clustering, with LL (Linking Length) and the minimum value of separation between galaxies, as our parameters. We have used a catalogue with approximately 44 thousand galaxies. After the simulation with six different values of LL, we analysed our clustering by means of the comparison with the obtained one by Tempel et al. (2016). In this comparison, we obtained a similarity of 79%. We also supplemented the validation of our method with an analysis of the silhouette coefficient, obtaining a general coefficient equal to 0.80. We present the characteristics of the groups obtained in our work, their comparison with the obtained one by Tempel et al. (2016) as well as the discussion with the non-grouped galaxies. / Neste trabalho apresentamos um m?todo alternativo para agrupamento de gal?xias, por meio do uso, dos dados de gal?xias dispon?veis em grandes levantamentos astron?micos. Utilizamos o algoritmo FoF, Friends of Friends, como base para o agrupamento, com os par?metros de LL (Linking Length) e o valor m?nimo de gal?xias, em comum para definir a uni?o entre grupos pr?ximos. Utilizamos um cat?logo com aproximadamente 44 mil gal?xias e ap?s a simula??o com seis valores distintos para o LL, e com a compara??o dos respectivos agrupamentos com os obtidos por Tempel et al. (2016), obtivemos, similaridade, de aproximadamente 79%. Complementamos a avalia??o com uma an?lise do Coeficiente se Silhueta, e encontramos agrupamentos com coeficiente geral de 0,80. Apresentamos a discuss?o e caracter?sticas dos grupos obtidos, assim como das gal?xias n?o agrupadas.