Detecção automática de Crateras de Impacto na superfície de Mercúrio

• Ano de defesa: 2015
• Autor(a) principal: Pedrosa, Miriam Maria [UNESP]
• Tipo de documento: Tese
• Tipo de acesso: Acesso aberto
• Idioma: por
• Instituição de defesa: Universidade Estadual Paulista (Unesp)
• Palavras-chave em Português: Cartografia; Crateras de impacto; Sensoriamento remoto; Processamento de imagens - Técnicas digitais; Reconhecimento de padrões; Cartography
• Link de acesso: http://hdl.handle.net/11449/154651; http://www.athena.biblioteca.unesp.br/exlibris/bd/cathedra/06-03-2018/000897073.pdf

Impact craters are geological structures formed by the collision of a meteorite with rocky planetary surfaces. The study of impact craters is a key point in understanding the history of solar system as the collision process has been present from the planet's formation until the present day. Consequently, several works have been carried out involving impact craters. Among them, some have been developed with the aim of identifying these features automatically. However, the vast majority deal only with images from the surfaces of Mars and the Moon. This is directly related to the huge amount of image data available for these two bodies, in comparison with other surfaces. However, just a couple of years ago it became possible to carry out this work for the planet Mercury. Images of the whole of the planet's surface have been obtained only after the Messenger probe. For this reason, few attempts have been made to make automated crater detections on Mercury. This work, presents a methodology to detect impact craters on Mercury from MDISC-NAC images. The images cover three of Mercury's basins and they have different spatial resolutions. The methodology is divided into four main steps. The first consists of creating image mosaics of the areas of interest; the second of finding crater candidates on the mosaics by using morphological operation and shape analysis. Subsequently, the set of features is extracted from the candidates by a Haar-like mask...