Vetores de características de alto nível visando a percepção do usuário

Abstract

The increase in the amount of digital medical images stored recently has created the necessity of algorithms for recovering data in an easier, faster and efficient way. However, one of the problems to use this technique is finding a similarity level based on the user perception, because it is necessary to characterize and quantify this similarity. Therefore, an integration between three elements (characteristic extractor, distance functions and perceptual parameter) which results in the creation of similarity operators is needed. It could be used to approximate the computational system to the final user. The main goal of this research is developing an intelligent learning system able to recovery images according to the user perception.