Double fusion filtering based multi-view face recognition

Two-dimensional (2D) face recognition by correlation is a key challenge of telecommunication and optical information processing. Although this issue has been the focus of intense research, its utilization still has some drawbacks especially when the face is in rotation. In this paper, we propose an alternative method based on a newly designed optical correlation filter which allows recognizing faces with different view angles. This filter called “Multi-View Binary Phase-Only Filter” is based on a double fusion of reference images allowing an optimisation of the use of the spatial-bandwidth product (SBWP) in the filter Fourier plane. The first fusion is performed in the image (space) domain, and the second one is conducted in the spectral domain. Simulations results with the Pointing Head Pose Image Database illustrate the performance of the designed correlation filter for multi-view face recognition.     

基于核独立成分分析的人脸识别

研究一种基于核独立成分分析的人脸识别方法。利用支持向量机的核函数思想,将原始人脸图像向量映射到高维特征空间,然后在高维特征空间中进行独立成分分析(ICA),提取非线性独立成分作为特征向量进行分类识别。实验结果表明该方法要比常规的基于ICA和PCA的人脸识别算法的识别率要高。     

基于Gabor小波变换的ICA火灾图像纹理识别算法

针对火灾图像纹理识别问题,提出了基于Gabor小波变换的ICA火灾图像纹理识别算法,并根据火灾图像纹理识别特点进行了优化。首先用不同尺度和方向的Gabor滤波器对待识别图像滤波,得到其特征图像,然后将特征图像转化成特征向量作为ICA的输入,得到基矢量子空间,再将测试图像经过Gabor滤波器的特征向量投影到ICA子空间中得到系数向量作为目标识别特征,最后用支持向量机进行识别。通过与Gabor滤波器法和ICA方法的对比实验,表明该算法可以在火灾纹理图像的识别率上比传统方法提高5%以上,为火灾图像识别提供了一种新思路。     

基于多变量鉴别分析的光学畸变不变性图像识别

提出一种基于鉴别分析的光学畸变不变性图像识别方法：对包含各种畸变的训练图像集采用主成分分析得到若干本征图像,作为参考模式与测试图像做光学相关,利用本民输入图像的相关结果作为识别特征,采用最佳鉴别分析做了训练和识别,即可实现对输入图像的畸变为性快速识别。采用非相干光相关器为光学实现硬件,给出了实验结果。     

Genetic optical design for a compressive sensing task

We present a sophisticated optical design method for reducing the number of photodetectors for a specific sensing task. The chosen design parameter is the point spread function, and the selected task is object recognition. The point spread function is optimized iteratively with a genetic algorithm for object recognition based on a neural network. In the experimental demonstration, binary classification of face and non-face datasets was performed with a single measurement using two photodetectors. A spatial light modulator operating in the amplitude modulation mode was provided in the imaging optics and was used to modulate the point spread function. In each generation of the genetic algorithm, the classification accuracy with a pattern displayed on the spatial light modulator was fed-back to the next generation to find better patterns. The proposed method increased the accuracy by about 30 % compared with a conventional imaging system in which the point spread function was the delta function. This approach is practically useful for compressing the cost, size, and observation time of optical sensors in specific applications, and robust for imperfections in optical elements.     

一种提高光学相关模式识别准确性的后处理方法

将基于绝对差算法的最小距离离度量方法用于光学相关器产生的相关结果灰度图像的相假性度量,并以此为依据给出待识别图像与参考图像之间的相似性结论,避免了传统的以相关联峰取阈为判决依据引起误判现象,提高了判决准确率,甚至可以对由灰度图像直接做光学相关得到的低质量相关结果给出准确判断,从而避免了对灰度图像做二值化及编码处理等烦琐的预处理工作,采用非相干相关器为光学实现硬件并给出了实验结果。     

Feature fusion of palmprint and face via tensor analysis and curvelet transform

In order to improve the recognition accuracy of the unimodal biometric system and to address the problem of the small samples recognition, a multimodal biometric recognition approach based on feature fusion level and curve tensor is proposed in this paper. The curve tensor approach is an extension of the tensor analysis method based on curvelet coefficients space. We use two kinds of biometrics: palmprint recognition and face recognition. All image features are extracted by using the curve tensor algorithm and then the normalized features are combined at the feature fusion level by using several fusion strategies. The k-nearest neighbour (KNN) classifier is used to determine the final biometric classification. The experimental results demonstrate that the proposed approach outperforms the unimodal solution and the proposed nearly Gaussian fusion (NGF) strategy has a better performance than other fusion rules.     

光学小波包变换及其滤波器的研究

基于对光学小波变换必要条件的分析，提出光学小波包变换的概念.选出虹膜图库的联合最优小波包基，利用最优基的线性组合生成相应的复合光学小波包滤波器.将滤波器用于光电混合虹膜识别系统中对待识别输入进行小波包特征提取预处理，模拟结果不仅证明引入该滤波器可明显提升系统的识别效果，也证明了光学小波包变换提出的意义.     

Design of a Label Recognition Filter for Crosstalk Suppression in a Label Recognition Unit using Optical Correlation Assuming the Assistance of Optical Time-Gating

To suppress crosstalk caused by matched filtering in label recognition based on optical correlation, we propose the novel design method of a label recognition filter to improve the discrimination performance. Assuming the assistance of optical time-gating, the label recognition filter is designed by a technique of a multiple-object discriminant filter (MODF) so that the crosstalk can be suppressed only at the center of correlation signals. This method can reduce the requirement for the performance of an optical thresholder. Simulation results show that the designed label recognition filter can successfully suppress the crosstalk.     

利用几何特性及神经网络进行人脸探测技术的研究

在人脸识别过程中 ,首先也是最重要的一个环节是人脸探测 ,因为一旦从图像中定位并提取到了人脸 ,那么下一步的人脸识别工作就变得非常容易。眼睛是人脸图像中最容易探测的部位 ,而且通过探测双眼来发现人脸最符合人的视觉习惯。提出了一种基于几何特征分析和人工神经网络的由粗到细的两级人脸探测方法。在第一级中 ,眼睛和脸是通过测量眼睛的尺寸和眼睛与脸的位置关系探测到的 ,第一级的输出是一个尺寸归一化的人脸 ,但偶尔也伴随着一个或多个因对复杂背景中与眼睛类似的物体的误判而得到的非人脸图像 ;第二级神经网络正是用来过滤掉第一级中被误判的人脸。实验表明 ,这种由粗到细的两级人脸探测系统具有很高的稳定性和探测正确率     

Design of Header Recognition Filter for Binary Phase Shift Keying in Header Recognition Unit using Time-Space Conversion

Optical header recognition is one key function that enables ultrafast optical routing in photonic packet-switched networks. Especially, optical header recognition based on optical correlation processing has attracted much attention. Header signals using optical codes are decoded using optical correlation processing. We improve the decoding method for Binary Phase Shift Keying (BPSK) header signals in order to suppress crosstalk caused by matched filtering in the header recognition unit. We design a header recognition filter using the technique of a multiple-object discriminant filter (MODF). Results of preliminary experiments show that the designed header recognition filter can suppress crosstalk. The intensity ratio of correlation signals of a target header signal to non-target ones is higher than 1.95 : 1 in case of 4-bits BPSK header signals.     

Existence problem of optical correlation based pattern recognition

The existence problem of optical correlation based pattern recognition, namely its range of validity and its limitation, is discussed in this paper conjointly with the function approximation theory of neural networks. The conclusion is that only if the sets to be recognized are linearly separable (which is rare) or the subsets, in which a segmental sample of the targets is involved, are linearly separable, can the classical 4f optical correlation system carry out the task of recognition inerrably. The recognition principle of a joint transform correlator is the same as that of a 4f system, and so is its range of validities. Based on the demonstration of the existence problem of optical correlation based pattern recognition an evaluation on some important problems that were studied in this field over the past 40 years is presented explicitly.     

基于子空间分析的人脸识别方法研究

人脸识别技术是模式识别和机器视觉领域的一个重要研究方向,在众多人脸识别的算法中,基于子空间分析的特征提取方法以其稳定可靠的识别效果成为了人脸识别中特征提取的主流方法之一。本文对目前应用较多的子空间分析方法进行了研究,具体介绍了线性子空间分析方法：主成分分析（PCA）、线性鉴别分析（LDA）、独立主成分分析（ICA）、快速主成分分析（FastICA）等及非线性子空间分析方法：基于核的PCA （KPCA）等的基本思想及其在人脸识别中的研究进展,包括一些新的研究成果。此外,还应用orl及Yale B人脸库对几个基础的子空间方法进行了验证实验。实验结果表明,在几个子空间分析方法中,FastICA算法取得了最高的识别率。最后结合实验结果对各算法的优缺点进行了分析总结。     

基于光学小波微分预处理的联合变换相关目标识别

提出了一种基于光学小波变换微分预处理的光电混合联合变换相关系统,该系统利用同一光路可实时实现对输入图象的小波变换微分预处理和联合变换相关识别.实验结果表明:该系统能得到很尖锐的互相关峰,可显著提高联合变换相关器的识别能力.     

基于光学相关的红外目标检测

光学图像处理技术能以复振幅、并行、高速度和巨量互连来处理信号。利用光学方法的半色调网屏技术实现了红外图像的假彩色编码,提高了红外图像的可识别性;利用光学相关识别技术构造了一种基于4f系统的光学图像实时相关识别系统,以实现对红外图像的实时处理和识别。     

基于傅里叶-极坐标变换的相关结果后处理技术

在光学相关识别中,对相关输出结果进行后处理是保证图像识别准确性的关键之一。研究者通常采用神经网络技术对相关输出结果进行处理,并取得了较好的效果,但它需要预先准备大量的相关峰和噪声峰训练样本。根据极坐标变换能将笛卡尔坐标系下的旋转转变成平移的性质以及傅里叶变换的平移不变性,提出一种基于傅里叶-极坐标变换的相关结果后处理方法。验证结果表明:目标的旋转图像与目标自身的傅里叶-极坐标变换的相似度较大,而干扰图像与目标的傅里叶-极坐标变换的相似度较小。本文的方法在后处理阶段可对目标和干扰进行有效的分类识别,而且还能避免对相关峰和噪声峰训练样本的收集,从而使得光学相关识别系统的应用更加便利。     

A hierarchical structure with improved OMP sparse representation used with face recognition

With the rapid development of the face recognition technology, more and more optical products are applied in people's real life. The recognition accuracy can be improved by increasing the number of training samples, but the colossal training samples will result in the increase of computational complexity. In recent years, sparse representation method becomes a research hot spot on face recognition. In this paper we propose an energy constrain orthogonal matching pursuit (ECOMP) algorithm for sparse representation to select the few training samples and a hierarchical structure for face recognition. We filter the training samples with ECOMP algorithm and then we compute the weights by all selected training samples. At last we find the closest recovery sample to the test sample. Simultaneously the experimental results in AR, ORL and FERET database also show that our proposed method has better recognition performance than the LRC and SRC_OMP method.     

Fast Face-Recognition Optical Parallel Correlator Using High Accuracy Correlation Filter

We designed and fabricated a fully automatic fast face recognition optical parallel correlator [E. Watanabe and K. Kodate: Appl. Opt. 44 (2005) 5666] based on the VanderLugt principle. The implementation of an as-yet unattained ultra high-speed system was aided by reconfiguring the system to make it suitable for easier parallel processing, as well as by composing a higher accuracy correlation filter and high-speed ferroelectric liquid crystal-spatial light modulator (FLC-SLM). In running trial experiments using this system (dubbed FARCO), we succeeded in acquiring remarkably low error rates of 1.3% for false match rate (FMR) and 2.6% for false non-match rate (FNMR). Given the results of our experiments, the aim of this paper is to examine methods of designing correlation filters and arranging database image arrays for even faster parallel correlation, underlining the issues of calculation technique, quantization bit rate, pixel size and shift from optical axis. The correlation filter has proved its excellent performance and higher precision than classical correlation and joint transform correlator (JTC). Moreover, arrangement of multi-object reference images leads to 10-channel correlation signals, as sharply marked as those of a single channel. This experiment result demonstrates great potential for achieving the process speed of 10000 face/s.     

Face recognition based on PCA and logistic regression analysis

Face recognition is an important research hotspot. More and more new methods have been proposed in recent years. In this paper, we propose a novel face recognition method which is based on PCA and logistic regression. PCA is one of the most important methods in pattern recognition. Therefore, in our method, PCA is used to extract feature and reduce the dimensions of process data. Afterwards, we present a novel classification algorithm and use logistic regression as the classifier for face recognition. The experimental results on two different face databases are presented to illustrate the efficacy of our proposed method.     

一种抗姿态与表情变化的三维人脸识别方法

为了提高人脸在姿态和表情变化下的识别率,结合局部平面距离(DLP)对曲面局部凹凸性优良的判断能力,提出了一种采用人脸的等距不变表示形式来匹配的人脸识别方法。首先,对深度摄像头采集到的深度图像进行距离约束、位置约束、转换等操作,得到干净完整的三维人脸,利用三维人脸上每一点DLP值确定鼻尖点,利用聚类的思想确定鼻根点;其次,采用改进的快速推进算法计算人脸的测地距矩阵,设置阈值并切割出有效的人脸区域;最后,计算有效的人脸区域的高阶矩特征,作为人脸的特征向量进行匹配。实验结果表明,对于不同的数据库,本文算法的识别率接近97%;将本文算法与基于轮廓线特征的人脸识别算法以及基于Gabor特征的人脸识别算法进行比较,其识别率分别提高了14.1%和8.3%,同时有着较高的运算效率。