3D Face Recognition Based on Multiple Keypoint Descriptors and Sparse Representation

Abstract

Recent years have witnessed a growing interest in developing methods for 3D face recognition. However, 3D scans often suffer from the problems of missing parts, large facial expressions, and occlusions. To be useful in real-world applications, a 3D face recognition approach should be able to handle these challenges. In this paper, we propose a novel general approach to deal with the 3D face recognition problem by making use of multiple keypoint descriptors (MKD) and the sparse representation-based classification (SRC). We call the proposed method 3DMKDSRC for short. Specifically, with 3DMKDSRC, each 3D face scan is represented as a set of descriptor vectors extracted from keypoints by meshSIFT. Descriptor vectors of gallery samples form the gallery dictionary. Given a probe 3D face scan, its descriptors are extracted at first and then its identity can be determined by using a multitask SRC. The proposed 3DMKDSRC approach does not require the pre-alignment between two face scans and is quite robust to the problems of missing data, occlusions and expressions. Its superiority over the other leading 3D face recognition schemes has been corroborated by extensive experiments conducted on three benchmark databases, Bosphorus, GavabDB, and FRGC2.0. The Matlab source code for 3DMKDSRC and the related evaluation results are publicly available at http://sse.tongji.edu.cn/linzhang/3dmkdsrcface/3dmkdsrc.htm.

Deeply Learned Pose Invariant Image Analysis with Applications in 3D Face Recognition

“…It targets to enhance classification accuracies using complementary information obtained from d-MVAHF-based features acquired from synthesized MVAHF images. The results obtained from our proposed methodology are better than the state-of-theart studies [17,19,27,[41][42][43][44] in terms of all the evaluation criteria employed by these studies.…”

“…Therefore, pose invariant face recognition using 3D models is proving to be promising especially in case of in-depth pose variations along x-, y-, and z-axis under unconstrained acquisition scenarios of real world [15]. Encouraged by these lines of evidence, many 3D face recognition approaches have been evolved and experimented in the last few years as given in the work of Bowyer et al [16] and the literature reviews [17][18][19][20]. The existing 3D face recognition approaches can be grouped into holistic, local feature-based, and hybrid domains [20].…”

“…In point based methods, a curvelet transform based study is proposed in the research work [18] to find salient points on the face images for building multiscale local surfaces. Similarly, facial key points are detected exploiting meshSIFT algorithm in the point based methods [19,25]. A prominent curve based method employing Riemannian framework is presented in the paper [26], whereas, the study [27] is a representative region based approach for occlusions and missing data handling problem.…”

“…For example, the algorithms presented in state-of-the-art studies [17,[41][42][43] employed FRGC v2.0 database while those presented in Refs. [19,26,28,[44][45][46] are based on both of FRGC v2.0 and GavabDB databases. Similarly, the studies [26,28,47] employed Gav-abDB, Bosphorus, and FRGC v2.0 databases while the study [27] used UMB-DB to evaluate the algorithm.…”

3D Palmprint Identification Using Block-Wise Features and Collaborative Representation

“…It needs to be noted that in addition to 3D palmprint, there are also some other 3D biometric technologies developed in recent years, such as 3D face [31], [32] and 3D ear [33], [34]. Compared with them, 3D palmprint has some inherent advantages.…”