Composição de dicionários visuais utilizando agrupamento de dados por Florestas de Caminhos Ótimos
2012; Linguagem: Inglês
ISSN
2639-6459
Autores Tópico(s)Image Retrieval and Classification Techniques
ResumoImage categorization by means of bag of visual words has received increasing attention by the image processing and vision communities in the last years. In these approaches, each image is represented by invariant points of interest which are mapped to a Hilbert Space, which is an extension of traditionals Euclidean plane and 3D space having any finite or infinite number of dimensions, representing a visual dictionary which aims at comprising the most discriminative features in a set of images. Notwithstanding, the main problem of such approaches is to find a compact and representative dictionary. Finding such representative dictionary automatically with no user intervention is an even more difficult task. In this work, we propose a method to automatically find such dictionary by employing a recent developed graph-based nao-supervisionado algorithm called OptimumPath Forest, which does not make any assumption about the visual dictionary’s size. Experiments were performed on 3 different databases of different objects in order to compare OPF nao-supervisionado, K-means and Random Selection. The comparison assessed the time for each technique to compute the visual dictionaries and the accuracy rate when such visual dictionaries are used. The experimental results showed that OPF nao-supervisionado is an alternate algorithm for the visual dictionary generation, since accuracy rates are similar, presents a time advantage when high-dimension dictionaries have to be computed and does not require visual dictionary dimension prior its computing.
Referência(s)