The SI-LBP: A new framework for obtaining 3D local binary patterns from shape-index

Abstract

In this paper, authors have presented a novel and original framework for extracting local binary pattern like facial descriptor from 3D images using Shape- Index, which have been termed as a ‘SI-Local Binary Pattern’ or simply ‘SI-LBP’. This framework could be adapted to all the LBP variations employed in range image based 3D image analysis, as it has been accomplished by processing the depth data from the 3D surface. During this investigation, authors have undertaken only 3D human face images. The SI-LBP framework consists of six different local facial regions with six different surface representations of a single 3D face from which discriminating attributes have been determined. In the paper, authors have described the foundations, implementation and main features of the SI-LBP and reported the characteristics of SI-LBP that confirms the importance of LBP operation on range face images. Furthermore, comparison with state-of-the-art on LBP counterparts applied on 3D images, has also been depicted here for the effectiveness of the proposed framework. The proposed framework has been extended by explaining the case study for face recognition using SI-LBP as well as 2.5DLBP during this investigation.