Detecção de transições e classificações de fases para análise de cristais líquidos por processamento digital de imagens

Abstract

This Master dissertation presents methods for identifying the points of phase transitions in nematic liquid crystals and phase identification between each transition. These methods are founded on content based image retrieval that uses visual features to group similar images, and also mathematical morphology to extract properties related to the topology of the images textures of liquid crystals. The algorithms receive as input na image sequence of a liquid crystal typically obtained in a heating process or on function of time and release like output information about the transitions temperatures and phases (reentrant isotropic, discotic, biaxial, etc). The results show that the processing and analysis of textures and colors through computerized procedures allow to find the points of transitions and the correct classification of the phases of studied liquid crystals, which are usually analyzed through the use of polarized light microscopy, a method that uses human skill and makes the process error-prone. Moreover, the methods have the potential to serve as tools to assist in the teaching and research of liquid crystals.