Fast 3D face recognition based on normal map

Abate, Andrea F.; Nappi, Michele; Ricciardi, Stefano; Sabatino, Gabriele

doi:10.1109/icip.2005.1530213

Cited by 14 publications

(7 citation statements)

References 14 publications

Supporting

Mentioning

Contrasting

Order By: Relevance

“…However, the surface normal, which determines (at each point) the orientation of a facial surface, has not been well explored for 3D face representation 1 . To the best of our knowledge, Abate et al [27,28,29,30] introduced normal maps to describe facial surfaces. But this direct use of normal information in the holistic way did not achieve satisfying results.…”

Section: Motivations and Our Solutionsmentioning

confidence: 99%

Expression-robust 3D face recognition via weighted sparse representation of multi-scale and multi-component local normal patterns

Huang

Morvan

et al. 2014

Neurocomputing

View full text Add to dashboard Cite

International audienceIn the theory of differential geometry, surface normal, as a first order surface differential quantity, determines the orientation of a surface at each point and contains informative local surface shape information. To fully exploit this kind of information for 3D face identification, this paper proposes a novel highly discriminative facial shape descriptor, namely Multi-Scale and Multi-Component Local Normal Patterns (MSMC-LNP). Given a registered facial range image, three components of normal vectors are first estimated, leading to three normal images. Then, each normal image is encoded locally to Local Normal binary Patterns (LNP) at different scales. To utilize spatial information of facial shape, each normal image is divided into several patches, and their LNP histograms are computed and concatenated according to facial configuration. Finally, each original facial surface is represented by a set of LNP histograms including both global and local cues. Moreover, to make the proposed solution robust to the variations of facial expressions, being subtle, prototypical or exaggerated, we propose to learn the weight of each local patch under a given encoding scale and normal component. Based on the learned weights and the weighted LNP histograms, we formulate a Weighted Sparse Representation-based Classifier (W-SRC). In contrast to the overwhelming of 3D FR algorithms which were only benchmarked on the FRGC v2.0 dataset, we carried out extensive experiments on the FRGC v2.0, Bosphorus, BU-3DFE and 3D-TEC databases, thus enclosing 3D face data captured under different scenarios through various sensors and depicting in particular different challenges with respect to facial expressions. The experimental results show that the proposed approach consistently achieves competitive rank-one identification rates over those datasets despite their heterogeneous nature, and demonstrates thereby its effectiveness and its generalization ability

show abstract

Section: Motivations and Our Solutionsmentioning

confidence: 99%

Expression-robust 3D face recognition via weighted sparse representation of multi-scale and multi-component local normal patterns

Huang

Morvan

et al. 2014

Neurocomputing

View full text Add to dashboard Cite

show abstract

“…Mohammadzade and Hatzinakos [37] showed that the surface normal vectors of the 3D faces are more useful for recognition purposes, thus, they contain more discriminative information at the sampled points than the coordinates of those points. Abate et al [38] proposed a normal map‐based 3D face recognition method for biometric applications. The normal map is a bi‐dimensional array which represents local curvature of a 3D polygonal mesh in terms of RGB colour data.…”

Section: Gradient‐based Feature Extractionmentioning

confidence: 99%

Use of gradient and normal vectors for face recognition

et al. 2020

View full text Add to dashboard Cite

“…Abate et al developed a simple, fast and accurate recognition method in which a difference map is calculated from the comparison of normal maps of two faces [33]. The surface normals and curvature are represented by a polygonal mesh and then projected as an RGB image which is a 2D representation of a 3D mesh.…”

Section: Related Workmentioning

confidence: 99%