Descritores de forma baseados em tensor scale

In the past few years, the number of image collections available has increased. In this scenery, there is a demand for information systems for storing, indexing, and retrieving these images. One of the main adopted solutions is to use content-based image retrieval systems (CBIR), that have the ability to, for a given query image, return the most similar images stored in the database. To answer this kind of query, it is important to have an automated process for content characterization and, for this purpose, the CBIR systems use image descriptors based on color, texture and shape of the objects within the images. In this work, we propose shape descriptors based on Tensor Scale. Tensor Scale is a morphometric parameter that unifies the representation of local structure thickness, orientation, and anisotropy, which can be used in several computer vision and image processing tasks. Besides the shape descriptors based on this morphometric parameter, we present a study of algorithms for Tensor Scale computation. The main contributions of this work are: (i) study of image descriptors based on color, texture and shape descriptors; (ii) study of algorithms for Tensor Scale computation; (iii) proposal and implementation of a contour salience detector based on Tensor Scale; (iv) proposal and implementation of new shape descriptors based on Tensor Scale; and (v) validation of the proposed descriptors with regard to their use in content-based image retrieval systems, comparing them, experimentally, to other relevant shape descriptors, recently proposed.