Computer Graphics & Animation

Biography

Biography: Ines Aparecida Gasparotto Boaventura

Abstract

Face recognition technology is a hot topic of research in the field of image processing and computer vision. Face feature has very high reference value in the identification, because it is easy to collect the characteristics. Face recognition technology is widely applied in many system related to information and public safety. In this work it is presented a face recognition algorithm based on a new version of Multi-Scale Local Mapped Pattern Method.
The Local Binary Pattern (LBP) and its extended forms, such as Mean Local Mapped Pattern (LMP) and Multi-Scale Local Binary Pattern (MSLBP), were developed with the purpose of analyzing textures in images. Such methods compare histograms generated by micropatterns extracted from textures. A micropattern may be understood as a structure formed by pixels and its respective gray levels capable of describing or representing a spatial context of some feature found in the image, such as borders, corners, texture and even more complex and abstract patterns, like those found in a face image. In the MSLBP, a histogram is built in each scale with the values generated by image patterns smoothened by the Gaussian filtering. The LMP technique consists of smoothening the image gray levels from the mapping made through a pre-defined function. For each image pixel, the mapping of the region is made on the basis of a specific region of its neighbors.

In the face features description problem, the LMP technique presented excellent results in considering the average of the locally mapped patterns, whereas the MSLBP, working in several scales, also reached higher performance compared with the original LBP. Thus, in this work we propose a new technique combining the LMP method and a new version of the MSLBP method, herein referred to as MSLMP (Multi-Scale Mean Local Mapped Pattern). The proposal of this new approach is to extract micropatterns and to attenuate noisy actions often occurring in digital images.
Therefore, in this talk we will present some results of the method applied on face image of some well known face Database, such as ESSEX, JAFE and ORL. The experiments have been carried out so far suggest that the presented technique provides detections with higher performance than the results presented in the state-of-the-art research in the specialized scientific literature. For the mentioned databases, the results have reached 100% of accuracy, using 7 scales of the proposed method.