Similaridade de Formas via Identificação e Caracterização de Saliências

• Autor(a) principal: Pedrosa, Glauco Vitor
• Data de Publicação: 2011
• Tipo de documento: Dissertação
• Idioma: por
• Título da fonte: Repositório Institucional da UFU
• Texto Completo: https://repositorio.ufu.br/handle/123456789/12514
• topic: Processamento de imagens digitais; Recuperação de informação; Recuperação de imagens baseada em conteúdo; Descrição de formas; Pontos de saliência; Digital image processing; Information retrieval; Content based image retrieval; Shape description; Salience points; CNPQ::CIENCIAS EXATAS E DA TERRA::CIENCIA DA COMPUTACAO

The number of images available has grown considerably and, as a consequence there is a growing interest in retrieving images in large databases. For this purpose, the intrinsic image features should be represented in such a way that two images can be perceptually differentiated. In general, the image features analyzed are: color, shape and/or texture. In pattern recognition and related areas, shape is one of the most widely image features exploited in content based image retrieval systems. In this work, we are interested in describing shapes using salience points. Saliences are defined as the points of high curvature along shape contour. These points are very useful for shape description, because they have the ability to represent a shape in a compact manner, invariant to rotation and translation. The main contribution of this work is a new shape descriptor proposed to analyze the similarity between shapes represented by its salience points. This descriptor utilizes three techniques proposed in this work: a salience point detector robust to noise, a salience representation using angular relative position and curvature value, and a distance function to analyze the similarity between two shapes. Experiments were made in order to illustrate the performance of the proposed descriptor while comparing it with other shape-based descriptors in the literature. From the experiments we can note that the proposed descriptor can retrieve images visually similar, requiring low space for storing the extracted features.