Reconhecimento de padrões na investigação dos olhos da face humana aplicada em jogos

Abstract

In relation to the techniques used to draw a human face, the literature describes where and how to position face elements (eg eyes, nose, mouth, head, ear). However, it does not describe the pattern (shape, size, and symmetry) used to define each of these face elements. Therefore, characters/avatars faces of digital games, for example, are created in a random manner and may generate dissatisfaction or non-identification with the character of the game by the players. In other words, they are created without a defined pattern. In this sense, this work investigated the creation of a methodology to elucidate the patterns that define the characteristics of the human face. For this project it was applied to the eye, the protocol being possible to be used for other elements of the face. The methodology was constructed using image processing and pattern recognition techniques to create a system for recognizing these characteristics from a data set with human characteristics. This work is divided in two phases: a) review and define techniques for image processing and extraction of face characteristics; And b) create the set of patterns/characteristics of each element of the face. When reviewing the literature, 47 datasets of facial image data were identified, totaling approximately 225,000 candidate images to be analyzed. Finally, from the obtained patterns will be expected to be able to use in various applications, such as: creating more realistic characters for digital games, create avatars for 3D printing, film, earn in the process of detection and recognition of people as well as application in safety. In addition there will be a set of real data of human face characteristics for the construction of character of your own like and the software for recognition and generation of avatar.