Análise de formas planas em imagens digitais: uma nova abordagem baseada na Transformada de Hough

• Ano de defesa: 2013
• Autor(a) principal: Souza, Gustavo Botelho de [UNESP]
• Tipo de documento: Dissertação
• Tipo de acesso: Acesso aberto
• Idioma: por
• Instituição de defesa: Universidade Estadual Paulista (Unesp)
• Palavras-chave em Português: Computação gráfica; Processamento de imagens - Tecnicas digitais; Reconhecimento de padrões; Formas (Matemática); Image processing
• Link de acesso: http://hdl.handle.net/11449/110379

Nowadays, given the widespread use of computers by society, the task of recognizing visual patterns are increasingly being automated, in particular to address the vast amount of digital images available. Applications of many areas such as biometrics, content-based image retrieval and medical diagnosis, make use of image analysis techniques in order to identify people, objects, actions, texts, etc. in the images. The basic features that can be extracted and analyzed from digital images are color, texture and shape. This paper presents some fundamental concepts about the shape analysis on digital images as well as some methods proposed in the literature. It also proposes a new shape analysis method called HTS (Hough Transform Statistics), with three versions of shape descriptors, which make use of statistical values extracted from the Hough space in order to characterize the shape of objects. Experiments conducted show that these new descriptors are very effective, presenting excellent performance regarding accuracy and processing time. These methods are good alternatives for the shape analysis task especially when working with large or high-resolution images, and with large databases, a tendency in our days, given the popularization and low costs of sensors and storage media