Xueqiao Wang

3D Face recognition using Corresponding Point Direction Measure and depth local features

A new scheme for 3D face recognition is presented in this paper. Firstly, we use Iterative Closet Point (ICP) to align all 3D faces with the first 3D face. Secondly, we reduce noise, especially the noise which in front of the face, and remove the spikes. Then we detect the nose tip point. Once the nose tip is successfully found, we crop a region, which is defined by a sphere radius of 100 mm centered at the nose tip. Then we use the Corresponding Point Direction Measure (CPDM) to matching the 3D face with the gallery 3D faces and get the score. At the same time, we use the region to construct depth image, and get the Gabor feature, LBP feature, principle component of the depth image. Finally, we fuse the CPDM result, Gabor feature, LBP feature, and principle component of depth image to finish the recognition. This paper presents a new method for matching 3D face and a new scheme for 3D face recognition. Experiments demonstrated the efficiency and effectiveness of the new method.

Cross-Modality 2D-3D face recognition via multiview smooth discriminant analysis based on ELM

In recent years, 3D face recognition has attracted increasing attention from worldwide researchers. Rather than homogeneous face data, more and more applications require flexible input face data nowadays. In this paper, we propose a new approach for cross-modality 2D-3D face recognition (FR), which is called Multiview Smooth Discriminant Analysis (MSDA) based on Extreme Learning Machines (ELM). Adding the Laplacian penalty constrain for the multiview feature learning, the proposed MSDA is first proposed to extract the cross-modality 2D-3D face features. The MSDA aims at finding a multiview learning based common discriminative feature space and it can then fully utilize the underlying relationship of features from different views. To speed up the learning phase of the classifier, the recent popular algorithm named Extreme Learning Machine (ELM) is adopted to train the single hidden layer feedforward neural networks (SLFNs). To evaluate the effectiveness of our proposed FR framework, experimental results on a benchmark face recognition dataset are presented. Simulations show that our new proposed method generally outperforms several recent approaches with a fast training speed.

Three-dimensional face recognition under expression variation

In this paper, we introduce a fully automatic framework for 3D face recognition under expression variation. For 3D data preprocessing, an improved nose detection method is presented. The small pose is corrected at the same time. A new facial expression processing method which is based on sparse representation is proposed subsequently. As a result, this framework enhances the recognition rate because facial expression is the biggest obstacle for 3D face recognition. Then, the facial representation, which is based on the dual-tree complex wavelet transform (DT-CWT), is extracted from depth images. It contains the facial information and six subregions’ information. Recognition is achieved by linear discriminant analysis (LDA) and nearest neighbor classifier. We have performed different experiments on the Face Recognition Grand Challenge database and Bosphorus database. It achieves the verification rate of 98.86% on the all vs. all experiment at 0.1% false acceptance rate (FAR) in the Face Recognition Grand Challenge (FRGC) and 95.03% verification rate on nearly frontal faces with expression changes and occlusions in the Bosphorus database.

Robust 3D face recognition using learn correlative features

3D images provide several advantages over 2D images for face recognition, especially when considering expression variations. In this paper, a novel framework is proposed for 3D-based face recognition. The key idea in the proposed algorithm is a correlative feature representation of the facial surface, by what is called 3D Local Binary Patterns (3D LBP), which encode relationships in neighboring mesh nodes and own more potential power to describe the structure of faces than individual points. The signature images are then decomposed into their principle components based on Spectral Regression resulting in a huge time saving. Our experiments were based on the CASIA 3D face database. Experimental results show our framework provides better effectiveness and efficiency than many commonly used existing methods for 3D face recognition and handles variations in facial expression quite well.

Expression robust three-dimensional face recognition based on gaussian filter and dual-tree complex wavelet transform

In this paper, a fully automatic framework is proposed for 3D face recognition and its superiority performance is justified by the FRGC v2 data. For 3D data preprocessing, a new face smoothing method is proposed. Meanwhile, 3D facial representation, which is extracted by the Dual-tree Complex Wavelet Transform DT-CWT, is introduced to reflect the facial geometry properties. Low redundancy makes it more effective and efficient to describe the discriminant feature in 2.5D range data. In this paper, DT-CWT is used into 2.5D range data in conjunction with the Linear Discriminant Analysis LDA to form a rejection classifier, which can quickly eliminate a large number of candidate gallery faces. The remaining faces are then verified using sparse representation based classification. Our method achieves the verification rate of 98.66% on All vs. All experiment at an FAR of 0.1%.

A new scheme for 3D face recognition

A novel system for 3D face recognition is presented in this paper. Firstly, we reduce the noise and move spikes from all the 3D faces. Secondly, we use Iterative Closet Point (ICP) to align all 3D face with the first person, and then for each face, we find the nose tip. Once the nose tip is successfully found, we crop a region, which is defined by a sphere radius of 100 mm centered at the nose tip. Depth image are constructed using the region subsequently. Then the depth image is projected into Gabor-based Supervised Locality Sensitive Discriminant Analysis (GISLSDA) space, which is improved by Gabor wavelet and Two-Directional Two Dimensions Principal Component Analysis (2D2PCA). Recognition is achieved by using a Nearest Neighbor (NN) classifier finally. This method is robust to changes in facial expressions and poses. The experimental results show that the new algorithm outperforms the other popular approaches reported in the literature and achieves much higher accurate recognition rate.

2D+3D face recognition using Dual-tree Wavelet Transform

A new automatic framework is proposed for face recognition and its superiority performance is justified by the FRGC v2 data. Adaboost face detecting method is used for facial region extraction. Then 2D and 3D facial representations which are extracted by the Dual-tree Complex Wavelet Transform (DT-CWT) are introduced to reflect the facial geometry properties in this paper. The level four high-frequency components of 2D texture image and 3D depth image are obtained respectively, and then Linear Discriminant Analysis (LDA) is used to get the feature vectors. Cosine distance is developed for establishing two similarity matrixes respectively. Finally a fusion result is established by the two similarity matrixes. The verification rate at an FAR of 0.1% is 97.6% on All vs. All experiment.

An automatic scheme for 3D face recognition

A novel system for 3D face recognition is presented in this paper. For preprocessing, firstly, we reduce the noise and move spikes from all the 3D faces. Secondly, we use Iterative Closet Point (ICP) to align all 3D face with the first person, and then for each face, we find the nose tip. Once the nose tip is successfully found, we crop a region, which is defined by a sphere radius of 100 mm centered at the nose tip. Depth image are constructed using the region subsequently. In the recognition process, we firstly extract intrinsic discriminative information embedded in 3D faces using Gabor filter, and then Patched Weighted Discriminant Locality Preserving Projections (PWDLPP), which is ameliorated by Principle Component Analysis (PCA) to eliminate redundancy, is used into those Gabor faces. Recognition is achieved by using a Nearest Neighbor (NN) classifier finally.