Comparative Study of 3D Face Acquisition Techniques

Abstract

Today, communication devices are evolving towards friendly-user interactivity while permanently eyeing towards 3D display technologies. As such, 3D face generation, modelling and animation techniques are in the frontline to design realistic animated 3D talking faces. Simple, reliable and economic, 2D image processing techniques have been widely used to reconstruct 3D faces. This paper focuses on the comparison of different 2D imaging techniques for 3D face generation. Stereo Vision techniques, using either automatic stereo correspondence algorithm or manual feature points location, Orthogonal Views and Photometric Stereo approaches are introduced and applied to acquire face 3D data. In addition, generated reconstruction results are compared qualitatively and quantitatively.

Comparison of Active Structure Lighting Mono and Stereo Camera Systems: Application to 3D Face Acquisition

“…Our prior studies have focused on passive reconstruction techniques [3,4] and our current preliminary investigations have shown that projecting active light structures results in a higher accuracy in facial surface reconstruction [2].…”

A Comparison of Three 3-D Facial Reconstruction Approaches

“…These areas carry most of the audio-visual information expressed by humans and their known locality should be considered in any technical evaluation. However, previously conducted anthropometric quantitative analysis has been inconclusive [1].…”

“…Reconstruction time depends on a compromise between the number of images required (for the case of complex coding 1 -local Dynamic Programming Method (DPM) 1 -global Symmetric Dynamic Programming Stereo (SDPS) 2 -global BVZ (Graph Cut based algorithm) 1 -global Belief-Propagation (BP) 3 -global Chen and Medioni (CM) 2 -local 1 from http://cat.middlebury.edu/stereo/code.html 2 our own implementation 3 from http://people.cs.uchicago.edu/∼pff/bp/, [17] strategies) and subsequent uniqueness in pixel code. From Table 2, the Gray Code algorithm matches codes whereas both of the direct coding techniques project a light pattern that aids the correspondence process in a standard Binocular Stereo algorithm.…”

A Patch-Based CNN Built on the VGG-16 Architecture for Real-Time Facial Liveness Detection

“…The core role of liveness detection is to verify whether the source of a biometric sample is a live human being or a fake representation. This process provides more safety and improvements to traditional facial recognition-based security systems [3], which use a person's unique biometric information, such as their face, to allow that individual to access specific systems or data. However, one of the primary impediments to biometric identification systems is the risk of spoofing attacks [4].…”