指节纹识别综述*

相比其它生物特征,指节纹具有特征丰富,采集设备价格低,易于结合手形、手指静脉及掌纹组成性能鲁棒的多模态识别系统等优点.文中首先介绍指节纹的定义、数据采集、预处理方法等,之后详细介绍各种指节纹识别算法及多模态识别方案.根据特征提取及匹配方法的不同,将指节纹识别算法分为6类:基于结构的算法、基于子空间学习的算法、基于编码的算法、基于纹理特征的算法、基于相关滤波器的算法和基于局部特征描述子的算法.回顾和总结各种算法的特点,展望未来指节纹识别的发展方向.     

基于特征融合的多模态身份识别方法研究

研究了多模态身份识别问题,结合人脸和掌纹两种不同生理特征,提出了基于特征融合的多模态身份识别方法。对人脸和掌纹图像分别进行Gabor小波、二维主元变换(2DPCA)提取图像特征,根据新的权重算法,结合两种模态的特征,利用最邻近分类器进行分类识别。在AMP、ORL人脸库和Poly-U掌纹图像库中的实验结果表明,两种模态的融合能更多地给出决策分析所需的特征信息相比传统的单一模态的人脸或掌纹识别具有较高的识别率,更具安全性和准确性。     

多模态情感识别研究进展*

情感识别研究热点正从单模态转移到多模态。针对多模态情感特征提取与融合的技术难点,本文列举了目前应用较广的多模态情感识别数据库,介绍了面部表情和语音情感这两个模态的特征提取技术,重点阐述了多模态情感融合识别技术,主要对多模态情感特征融合策略和融合方法进行了综述,对不同算法下的识别效果进行了对比。最后,对多模态情感识别研究中存在的问题进行了探讨,并对未来的研究方向进行了展望。     

基于特征融合的掌纹识别算法应用研究

研究掌纹准确识别问题,由于光照强度、位置移动、采集设备等影响,采集掌纹图像的分辨率较低。单一掌纹特征提取方法难以全面描述掌纹信息,导致掌纹识别率低。为了提高了掌纹识别率,提出一种基于Gabor滤波和LBP算法相融合的掌纹识别方法。首先对采集掌纹进行预处理,然后分别采用Gabor滤波和LBP算法进行特征提取,最后采用神经网络建立掌纹识别器。仿真结果表明,相对于单一特征提取算法,融合特征算法不仅提高了掌纹识别率,同时加快掌纹识别速度,能够很好满足实时掌纹识别系统的要求。     

多模态生物特征提取及相关性评价综述

生物特征识别是身份认证的重要手段,特征提取技术在其中扮演了关键角色,直接影响识别的结果。随着特征提取技术日趋成熟,学者们逐渐将目光投向了生物特征间的相关性问题。本文以单模态和多模态生物识别中的特征提取方法为研究对象,回顾了人脸与指纹的特征提取方法,分析了基于经验知识的特征分类提取方法以及基于深度学习的计算机逻辑采样提取方法,并从图像处理的角度对单模态与多模态方法进行对比。以当前多模态生物特征提取方法和DNA表达过程为引,提出了不同模态的生物特征之间存在相关性的猜想,以及对这一猜想进行建模的思路。在多模态生物特征提取的基础上,对今后可能有进展的各生物特征之间的相关性建模进行了展望。     

基于子空间特征融合的两级掌纹识别算法

针对单一PCA或PCA只能提取掌纹的线性或非线性特征,单一分类器的掌纹识别率低缺陷,提出一种子空间特征融合的两级掌纹识别方法（PCA-KPCA-SVM）。首先采用子空间特征提取方法PCA、KPCA分别提取掌纹图像线性和非线性特征,然后基于融合特征总类间距离最大准则,计算出最佳的融合系数,得到PCA、KPCA的融合掌纹特征,最后将融合特征输入到欧式距离分类器进行掌纹识别,如果拒绝识别,则输入支持向量机进行二次识别。采用Polyu掌纹图像库进行测试实验,结果表明,相对于对比算法,PCA-KPCA-SVM提高了掌纹识别率,有效降低了掌纹的误识率和拒识率。     

基于傅立叶变换的掌纹识别方法

掌纹识别是指由计算机自动识别哪些掌纹图像来自同一只手掌,哪些来自不同的手掌.在掌纹识别中,特征提取算法的优劣至关重要.提出了一种基于傅立叶变换的掌纹特征提取方法.该方法的基本思想是先将掌纹图像应用傅立叶变换转换到频域,然后在频域中进行特征提取和描述.提取出来的特征备用来索引掌纹数据库,以便当一个新的掌纹图像被输入时,可以很快确定该手掌是否已经在掌纹库中注册.该方法可以用来做基于人体生物特征的身份识别,在安全领域有广泛的应用前景.实验验证了该方法的有效性.     

基于视听觉感知系统的多模态情感识别

情绪识别作为人机交互的热门领域,其技术已经被应用于医学、教育、安全驾驶、电子商务等领域.情绪主要由面部表情、声音、话语等进行表达,不同情绪表达时的面部肌肉、语气、语调等特征也不相同,使用单一模态特征确定的情绪的不准确性偏高,考虑到情绪表达主要通过视觉和听觉进行感知,本文提出了一种基于视听觉感知系统的多模态表情识别算法,分别从语音和图像模态出发,提取两种模态的情感特征,并设计多个分类器为单特征进行情绪分类实验,得到多个基于单特征的表情识别模型.在语音和图像的多模态实验中,提出了晚期融合策略进行特征融合,考虑到不同模型间的弱依赖性,采用加权投票法进行模型融合,得到基于多个单特征模型的融合表情识别模型.本文使用AFEW数据集进行实验,通过对比融合表情识别模型与单特征的表情识别模型的识别结果,验证了基于视听觉感知系统的多模态情感识别效果要优于基于单模态的识别效果.     

多文化场景下的多模态情感识别

自动情感识别是一个非常具有挑战性的课题,并且有着广泛的应用价值.本文探讨了在多文化场景下的多模态情感识别问题.我们从语音声学和面部表情等模态分别提取了不同的情感特征,包括传统的手工定制特征和基于深度学习的特征,并通过多模态融合方法结合不同的模态,比较不同单模态特征和多模态特征融合的情感识别性能.我们在CHEAVD中文多模态情感数据集和AFEW英文多模态情感数据集进行实验,通过跨文化情感识别研究,我们验证了文化因素对于情感识别的重要影响,并提出3种训练策略提高在多文化场景下情感识别的性能,包括：分文化选择模型、多文化联合训练以及基于共同情感空间的多文化联合训练,其中基于共同情感空间的多文化联合训练通过将文化影响与情感特征分离,在语音和多模态情感识别中均取得最好的识别效果.     

基于YOLO的多模态特征差分注意融合行人检测

针对可见光模态与热红外模态间的差异问题和如何充分利用多模态信息进行行人检测,本文提出了一种基于YOLO的多模态特征差分注意融合行人检测方法.该方法首先利用YOLOv3深度神经网络的特征提取主干分别提取多模态特征;其次在对应多模态特征层之间嵌入模态特征差分注意模块充分挖掘模态间的差异信息,并经过注意机制强化差异特征表示进而改善特征融合质量,再将差异信息分别反馈到多模态特征提取主干中,提升网络对多模态互补信息的学习融合能力;然后对多模态特征进行分层融合得到融合后的多尺度特征;最后在多尺度特征层上进行目标检测,预测行人目标的概率和位置.在KAIST和LLVIP公开多模态行人检测据集上的实验结果表明,提出的多模态行人检测方法能有效解决模态间的差异问题,实现多模态信息的充分利用,具有较高的检测精度和速度,具有实际应用价值.     

应用非负矩阵分解和RBPNN模型的掌纹识别方法

提出一种基于非负矩阵分解（NMF）和径向基概率神经网络的掌纹识别方法。NFM是一种有效的图像局部特征提取算法,用于图像分类时能得到较高的识别率。考虑PolyU掌纹图像数据库,应用NMF、局部NMF（LNMF）、稀疏NMF（SNMF）和具有稀疏度约束的NMF（NMFSC）算法分别对掌纹图像进行特征提取,并对提取到的局部特征基图像进行分析对比;在特征提取的基础上,应用径向基概率神经网络（RBPNN）模型对掌纹特征进行分类,分类结果表明了RBPNN模型对掌纹特征具有较好的识别能力。实验对比结果证明了基于RBPNN的NMF掌纹识别方法在掌纹识别中的有效性,具有一定的理论研究意义和实用性。     

Concavity-orientation coding for palmprint recognition

Efficient feature extraction strategies play an important role in palmprint recognition systems. Among various feature extraction methods, orientation methods such as Competitive Code and Half Orientation Code are the baseline ones. They encode responses of a bank of orientational filters into a binary representation and can match a test palmprint sample in real-time with a relatively good accuracy. However, they use the orientation information based upon this idea that palmprints encompass only straight lines with different orientations, whereas in reality, the majority of palm’s lines are curved. This observation naturally brings the idea that the concavity and orientation features as different aspects of palmprints curves might provide more reliable and discriminative representations in palmprint recognition. Motivated by this idea, in this work we investigate the use of the concavity feature in different orientations for palmprint recognition. The experimental results, which are applied on PolyU II, 2D/3D PolyU, and blue and near infrared range images from Multispectral PolyU palmprint databases prove the efficiency of this idea compared to other coding-based methods.     

Robust palmprint identification based on directional representations and compressed sensing

In this paper, we propose a novel approach for palmprint recognition, which contains two interesting components: directional representation and compressed sensing. Gabor wavelets can be well represented for biometric image for their similar characteristics to human visual system. However, these Gabor-based algorithms are not robust for image recognition under non-uniform illumination and suffer from the heavy computational burden. To improve the recognition performance under the low quality conditions with a fast operation speed, we propose novel palmprint recognition approach using directional representations. Firstly, the directional representation for palmprint appearance is obtained by the anisotropy filter, which is robust to drastic illumination changes and preserves important discriminative information. Then, the principal component analysis (PCA) is used for feature extraction to reduce the dimensions of the palmprint images. At last, based on a sparse representation on PCA feature, the compressed sensing is used to distinguish palms from different hands. Experimental results on the PolyU palmprint database show the proposed algorithm have better performance than that of the Gabor based methods.     

小波分解与PCA方法的掌纹特征提取方法*

提出了一种新的掌纹特征提取方法,其目的在于在不降低识别率的情况下,提高掌纹特征提取速度。首先将原始掌纹图像进行小波分解,获得低分辨率的掌纹图像;其次通过主成分分析（PCA）方法获得一个低维子空间,即“特征掌”;最后通过将训练、测试样本在该“特征掌”上投影来提取掌纹特征。实验结果表明,所提出方法与单一PCA方法比较,在同样识别率情况下,特征提取速度明显提高。     

刚性变形成像条件下掌纹识别方法研究

目前广泛使用的掌纹图像采集装置是非接触式,这种方式适应了掌纹识别生活化的实用要求。但是构成了不稳定的成像环境,拍摄过程中会产生平移、旋转、扭曲,我们将这些不会影响掌纹线结构特征的变形称之为刚性变形。本文从手掌长度和宽度两个角度衡量掌纹图像刚性变形程度,设计了一种归一化校正方法。建立不同变形程度的掌纹图库,对掌纹特征匹配结果进行比较实验;实验结果表明,这种方法能够降低由于刚性变形对识别率产生的影响。     

基于近邻三值模式和协作表示的三维掌纹识别

针对二维掌纹图像存在易伪造、抗噪能力差的问题，提出一种基于近邻三值模式（NTP）和协作表示的三维掌纹识别方法。首先，利用形状指数把三维掌纹的表面几何信息映射成二维数据，以弥补常用均值或高斯曲率映射无法精确描述三维掌纹特征的缺陷；其次，对形状指数图作分块处理，利用近邻三值模式提取分块形状指数图的纹理特征；最后，利用协作表示的方法进行特征分类。在三维掌纹库上和经典算法进行的对比实验中，该方法的识别率为99.52%，识别时长为0.6738 s，优于其他算法；在识别率方面，与经典的局部二值模式（LBP）、局部三值模式（LLTP）、CompCode、均值曲率图（MCI）法相比分别提高了7.77%、6.02%、5.12%和3.97%；在识别时间方面，与Homotopy、对偶增广拉格朗日法（DALM）、SpaRSA方法相比分别降低了6.7 s、15.9 s和61 s。实验结果表明，所提算法具有良好的特征提取和分类能力，能够有效地提高识别精度并减少识别时间。     

Stereo-based palmprint recognition in various 3D postures

In order to increase performance in palmprint recognition systems, various devices are normally used to restrict the movement of the hand. These can cause problems, especially for those users with physical disabilities. They also cause significant hygiene problems in multi-user systems. Recently, studies on palmprint recognition systems have progressed towards the development of unconstrained, contactless and unrestricted background techniques. The most common problem encountered in these studies is the alignment arising from the free movement of the hand. Despite 3D hand-acquisition devices which offer extra recognition features to overcome this problem, the applicability of these devices is low because of their increased cost. In this study, a stereo camera was proposed. Although due to matching problems, it is difficult to achieve precise, distinct feature extraction in the unrestricted 3D environment used for palmprint recognition, the orientation of the hand in 3D space can be determined by obtaining depth information. In this study, the depth information was extracted by using the binocular stereo approach. First, the orientation of the hand was estimated by fitting a surface model associated with the eigenvectors of the depth information. Pose correction was then accomplished by establishing a relationship between the orientation and the images. The pose correction greatly relieved the perspective distortion that usually occurs within the various poses of the hands. Next, the region of interest was determined by performing segmentation on the corrected images using the Active Appearance Model (AAM). The palmprint features were then extracted via Gabor-based Kernel Fisher Discriminant Analysis. In order to demonstrate the performance of the proposed approach, a new dataset was compiled from stereo images within various scenarios collected from 138 different individuals. As a result of these experimental studies, the EER values, especially on the images captured from different hand orientations in 3D, were reduced from around 14–0.75%. With the help of this suggested approach, the palmprint recognition system was transformed into a more portable form by removing the closed-box mechanisms and equipment restricting movement of the hand. This system can automatically perform pose estimation, hand segmentation and recognition processes without any special intervention.     

Double-cohesion learning based multiview and discriminant palmprint recognition

Palmprint recognition has been widely used in security authentication. However, most of the existing palmprint representation methods are focused on a special application scenario using the hand-crafted features from a single-view. If the features become weak as the application scenario changes, the recognition performance will be degraded. To address this problem, we propose to comprehensively exploit palmprint features from multiple views to improve the recognition performance in generic scenarios. In this paper, a novel double-cohesion learning based multiview and discriminant palmprint recognition (DC_MDPR) method is proposed, which imposes a double-cohesion strategy to reduce the inter-view margins for each subject and the intra-class margins for each view. In this way, for each subject, the features from different views can be closer to each other in the binary-label space. Meanwhile, for each view, the features sharing the same label information can move towards each other by imposing a neighbor graph regularization. The proposed method can be flexibly applied to any type of palmprint feature fusion. Moreover, it presents the multiview features in a low-dimensionality sub-space, effectively reducing the computational complexity. Experimental results on various palmprint databases have shown that the proposed method can always achieve the best recognition performance compared to other state-of-the-art algorithms.