一种基于局部信息统计的虹膜分块编码方法

由于虹膜自身的稳定性、非侵犯性、不可更改性等优点,虹膜识别已经成为生物特征身份鉴别领域中的研究热点。但虹膜丰富的纹理和复杂的结构给特征提取和编码带来了很大困难。为尽可能地简化特征提取和编码方法,提高虹膜识别效率,提出了一种基于局部信息统计的虹膜分块编码方法。对原始人眼图像进行虹膜定位等预处理操作,得到归一化的虹膜纹理图像;分别根据虹膜局部信息与全局信息、局部信息与局部信息之间的比较关系进行分块编码;计算了不同虹膜代码之间的汉明(Hamming)距离。根据汉明距离给出识别结果。实验证明该方法有效、可行,具有较高的识别率和识别速度。     

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.     

Region-based SIFT approach to iris recognition

Traditional iris recognition systems transfer iris images to polar (or log-polar) coordinates and have performed very well on data that tends to have a centered gaze. The patterns of an iris are part of a 3-D structure that is captured as a two-dimensional (2-D) image and cooperative iris recognition systems are capable of correctly matching these 2-D representations of iris features. However, when the gaze of an eye changes with respect to the camera lens, many times the size, shape, and detail of iris patterns will change as well and cannot be matched to enrolled images using traditional methods. Additionally, the transformation of off-angle eyes to polar coordinates becomes much more challenging and noncooperative iris algorithms will require a different approach. The direct application of the scale-invariant feature transform (SIFT) method would not work well for iris recognition because it does not take advantage of the characteristics of iris patterns. We propose the region-based SIFT approach to iris recognition. This new method does not require polar transformation, affine transformation or highly accurate segmentation to perform iris recognition and is scale invariant. This method was tested on the iris challenge evaluation (ICE), WVU and IUPUI noncooperative databases and results show that the method is capable of cooperative and noncooperative iris recognition.     

Iris segmentation using variational level set method

Continuous efforts have been made to process degraded iris images for enhancement of the iris recognition performance in unconstrained situations. Recently, many researchers have focused on developing the iris segmentation techniques, which can deal with iris images in a non-cooperative environment where the probability of acquiring unideal iris images is very high due to gaze deviation, noise, blurring, and occlusion by eyelashes, eyelids, glasses, and hair. Although there have been many iris segmentation methods, most focus primarily on the accurate detection of iris images captured in a closely controlled environment. The novelty of this research effort is that we propose to apply a variational level set-based curve evolution scheme that uses a significantly larger time step to numerically solve the evolution partial differential equation (PDE) for segmentation of an unideal iris image accurately, and thereby, speeding up the curve evolution process drastically. The iris boundary represented by the variational level set may break and merge naturally during evolution, and thus, the topological changes are handled automatically. The proposed variational model is also robust against poor localization and weak iris/sclera boundaries. In order to solve the size irregularities occurring due to arbitrary shapes of the extracted iris/pupil regions, a simple method is applied based on connection of adjacent contour points. Furthermore, to reduce the noise effect, we apply a pixel-wise adaptive 2D Wiener filter. The verification and identification performance of the proposed scheme is validated on three challenging iris image datasets, namely, the ICE 2005, the WVU Unideal, and the UBIRIS Version 1.     

基于灰度曲面匹配的虹膜识别方法

目前的虹膜识别都是采用在图像上提取特征点,并将特征点编码为固定长度的特征数据用于匹配的方式。这种方式使虹膜识别系统易受攻击。为了进一步提高虹膜识别系统的安全性和识别速度,提出了一种基于灰度曲面匹配的虹膜识别方法。该方法抛弃了特征提取和编码等传统操作,在特征分析的基础上直接利用了灰度曲面匹配的思想,首先对两幅图像中的像素灰度做差,得到灰度差曲面,然后求出该灰度差曲面的方差。将此方差作为衡量两个虹膜特征曲面之间距离的依据,并据此判定两个虹膜是否来自同一只眼睛。在给定阈值为40的前提下,正确识别率为96.89%,识别时间为53.2 ms。实验结果证明,该方法识别准确率高,识别速度快。     

单摄像机下基于眼动分析的行为识别

眼动信息是识别观看视频、浏览网页等以视觉任务为主的行为的关键信息. 针对传统的可穿戴传感器普遍具有侵入性, 而现有基于视觉的眼动仪存在价格昂贵、校准过程复杂等问题, 本文尝试使用单一的标准网络摄像头获取眼动信息用于行为识别, 并评估了该方法的可行性. 提出一种针对低质量视频图像的虹膜跟踪算法以获取眼动信号, 然后分别从水平和垂直方向的眼动信号中提取出五种新的眼动特征, 并从中选择出最优特征子集, 最后采用支持向量机分类器评价了本文方法的可行性. 针对不同应用背景设计了三组验证实验: 留一交叉验证、k折交叉验证及单独校准测试, 三组实验中, 对不同参与者三种行为的平均识别正确率分别为68.4%, 79.3%及84.1%, 证明了基于视频图像的眼动分析用于行为识别是一种很有希望的传感形式, 并有望用于更为复杂的传感任务. 关键词： 物联网 行为识别 眼睛跟踪 传感     

A non-circular iris localization algorithm using image projection function and gray level statistics

Iris recognition technology identifies an individual from its iris texture with great precision. A typical iris recognition system comprises eye image acquisition, iris segmentation, feature extraction, and matching. However, the system precision greatly depends on accurate iris localization in the segmentation module. In this paper, we propose a reliable iris localization algorithm. First, we locate a coarse eye location in an eye image using integral projection function (IPF). Next, we localize the pupillary boundary in a sub image using a reliable technique based on the histogram-bisection, image statistics, eccentricity, and object geometry. After that, we localize the limbic boundary using a robust scheme based on the radial gradients and an error distance transform. Finally, we regularize the actual iris boundaries using active contours. The proposed algorithm is tested on public iris databases: MMU V1.0, CASIA-IrisV1, and the CASIA-IrisV3-Lamp. Experimental results demonstrate superiority of the proposed algorithm over some of the contemporary techniques.     

非接触式虹膜图像获取及特征提取方法的研究

报道了一种由使用者自行瞄准来实现虹膜图像非接触式获取的新方法,对获得的图像数据进行了虹膜纹理特征提取试验,验证了方法的可行性和有效性。通过设计几何光照明来辅助使用者定位瞄准,使眼睛处于前后合适的位置;并在摄像机镜头前面放置一片半透半反镜,帮助使用者观察自己眼睛的像来判断眼睛是否偏出摄像机的视场范围。通过该方法可以得到虹膜特征识别的图像数据。采用二维伽博小波对虹膜纹理进行特征提取,并计算两个虹膜特征码值中数值相等的特征位个数,对采集到的虹膜图像进行了特征匹配实验,得到的实验数据证实了该系统完全满足虹膜特征识别的要求。     

Localization of iris in gray scale images using intensity gradient

A new method of iris localization based on intensity value analysis is proposed in this paper. Iris recognition systems are mainly dependent on the performance of iris localization processing. Steps after localization involve normalization, feature extraction and matching. These steps are based on the accuracy and efficiency of localization of iris in human eye images. In the proposed scheme, the inner boundary of iris is calculated by finding the pupil center and radius using two methods. In the first method, selected region is adaptively binarized and centroid of the region utilized for obtaining pupil parameters. Edges are processed to detect radius and center of pupil during the second method. For outer iris boundary, a band is calculated within which iris outer boundary lies. Signals in one dimension are picked up along radial direction within determined band at different angles. Three points with maximum gradient are selected from each signal. Redundant points are deleted using Mahalanobis distance and remaining points are used to obtain the outer circle of the iris. Points for upper and lower eyelids are found in the same way as the iris outer boundary. Selected points are then statistically fitted to make parabolas and lastly eyelashes are removed from the image to completely localize the iris. Experimental results show that proposed method is very efficient and accurate.     

虹膜区域选取与识别率对应关系分析

在人眼自然张开状态下,虹膜遭受睫毛、眼睑和反射光斑不同程度地干扰而失去了部分信息,因此需要找出虹膜保留面积与虹膜识别率之间的对应关系,从而根据小同的识别率要求估计不同的人眼最小张开程度.利用相关系数识别方法作为评测手段分别在不同的虹膜图库中进行了实验.结果证明,选取靠近瞳孔的虹膜内环区域与整个虹膜区域的比率在大于37.5%时,或者暴露的虹膜两侧扇形的角度大于70.时,就可以获得区分个体的特征量,达到比较稳定的识别率.实验结果还给出了虹膜内环区域而积与识别率的对应关系曲线以及虹膜两侧扇形区域面积与识别率之间的对应关系曲线.     

基于2D-Gabor滤波器的虹膜噪声检测方法

人眼在自然张开状态下,虹膜区域会受到部分眼睑、睫毛的遮挡,这些噪声会降低虹膜识别的准确性,因而在识别时应该尽可能将虹膜有效信息提取出来,而将眼睑、睫毛等干扰去除。为此本文提出了一种在归一化后的虹膜图像上利用2D-Gabor滤波器检测和去除虹膜区域内噪声的方法,并将噪声区域标记。在虹膜匹配时,通过将虹膜代码中的干扰位去除来尽量避免噪声对于虹膜识别的影响。实验结果表明本文提出的方法可以准确检测并标记噪声区域,有助于提高识别的可靠性和识别准确率。     

基于K-L变换的虹膜识别方法

提出了一种基于K-L变换和最近邻分类器的虹膜识别方法。该方法采用K-L变换得到一组虹膜图像基,并利用这组基图像构造一子空间,将待识别图像在这个空间上的投影系数作为待识别图像的特征向量。采用最近邻分类器进行了分类。基于CASIA虹膜数据库的试验表明,将K-L变换应用在虹膜特征提取上可以取得较高的识别率。     

虹膜识别技术及应用

虹膜识别技术由于具有稳定性好、错误接受率 (FAR)和错误拒绝率 (FRR)相等的指标最优、识别速度快等优点正逐步受到人们的重视。对虹膜纹理的特征、虹膜识别技术的发展和应用进行了较详细的介绍 ;通过将虹膜识别技术与其它生物识别技术相比较 ,对虹膜识别技术的优点进行了讨论。讨论了虹膜识别系统的构成 ,以及用二维Gabor小波和其它小波变换实现特征编码的方法和叙述了虹膜识别中所采用的一些算法     

基于热红外图像处理技术的农作物冠层识别方法研究

为解决农作物冠层热红外图像边缘灰度级分布不均且噪声较大,而传统图像分割方法难以实现其目标区域有效识别的难题,以苗期红小豆冠层热红外图像为研究对象,将模糊神经网络和仿射变换有机结合,提出了基于热红外图像处理技术的农作物冠层识别模型。首先利用五层线性归一化模糊神经网络的自适应特性,选取高斯隶属度函数,自动计算冠层可见光图像识别的推理规则,有效地分割了可见光图像中的冠层区域。通过分析3种分割指标和熵,定量评价可见光图像冠层分割质量。网络迭代38次时,误差精度为0.000 952,该算法平均有效识别率为96.13%,获取可见光冠层图像的像元信息熵值范围为2.454 4~5.198 7,与标准算法所得冠层图像的像元信息熵仅相差0.245 9。然后以取得可见光图像的冠层有效区域为参考图像,采用仿射变换算法,调整优选平移、旋转、缩放等图像变换因子,配准原始热红外图像,提出了基于仿射变换的冠层热红外图像识别方法。对于初始温度范围值在16.35~19.92 ℃的农作物热红外图像,计算选取旋转幅度为1.0和缩放因子为0.9时,作为异源图像的最优配准参数,获取目标图像的最大温差为3.17 ℃,相对于原图像的平均温度值由18.711 ℃下降至17.790 ℃,进而实现了基于热红外图像处理技术的农作物冠层识别。最后以熵的互信息作为监督指标,对农作物冠层热红外图像识别方法进行评价。提出的冠层热红外图像识别方法,所获取的目标图像与初始热红外图像的平均互信息为4.368 7,标准目标图像和初始热红外图像的平均互信息为3.981 8,二者仅相差0.486 9。同时,两种冠层热红外图像的平均温度差值为0.25 ℃,高效消除了原始热红外图像的背景噪声。结果表明本研究方法的有效性和实用性,能够为应用热红外图像反映农作物生理生态信息特征指标参数提供技术借鉴。     

Co-registration method for photoacoustic imaging and laser speckle imaging

Combining photoacoustic (PA) imaging with laser speckle (LS) imaging (LSI) can simultaneously determine total hemoglobin concentration (HbT), hemoglobin oxygen saturation (SO2), and blood flow rates. Thus,the co-registration of PA and LS images is important in physiological studies and pathological diagnosis.This letter presents a co-registration algorithm combining mutual information with the maximum betweenclass variance segmentation method (Otsu method). The mutual information and Otsu method are used to provide the registration measure criterion and image feature recognition, respectively. The evaluation results show that the registration function possesses a single maximum peak and high smoothness across the global co-registration district, indicating a robust behavior. Moreover, this method has good registration accuracy, and the fusion result simultaneously visualizes the separate functional information of two kindsof images.     

可见-近红外高光谱成像结合化学计量学分辨人面部信息

人的面部信息与指纹和虹膜一样可以用于人的身份鉴别,并且相比之下更容易实现远距离的分辨和识别。利用高光谱成像技术可以应用到人脸识别领域并获取丰富的信息和庞大的成像数据量,需要采用化学计量学方法才能充分提取其中包含的有效信息,并为计算机识别奠定基础。研究了可见-近红外高光谱成像技术对人的面部信息进行分析的可行性。结果表明,多元曲线分辨-交替最小二乘方法不同于主成分分析,能够通过主成分纯光谱和相对浓度等具有具体物理化学意义的数据表征人的面部信息,而且可以方便地根据成像数据的特点施加运算中的约束。另外,采用偏最小二乘判别分析的方法实现了对不同肤色的皮肤信号光谱进行分类。白种人和黄种人的面部高光谱信息特征相似,分类难度高于深色皮肤人种。     

高精度虹膜识别光学镜头的研制

针对虹膜识别系统中难于获取高分辨率的虹膜图像问题,设计一款高精度虹膜图像采集摄像镜头。镜头焦距14.2 mm, 畸变小于0.15%,像方数值孔径为0.26,像面直径为5.35 mm,工作距离130 mm,物像共轭距为150 mm。光学系统可获得高分辨率图像,在0.7视场光学传递函数MTF在120 lp/mm时大于30%。镜头由4片球面透镜构成,结构简单,适合产业化生产,各项光学指标检测结果达到设计要求。     

Human eye localization using the modified Hough transform

The precise localization of parts of a human face such as mouth, nose or eyes is important for their image understanding and recognition. The developed successful computer method of eyes and eyelids localization using the modified Hough transform is presented in this paper. The efficiency of this method was tested on two publicly available face images databases and one private face images database with the location correctness better than 96% for a single eye or eyelid and 92% for eye and eyelid couples.     

A method on calculating high-dimensional mutual information and its application to registration of multiple ultrasound images

Using mutual information as a criterion for medical image registration, which requires no prior segmentation or preprocessing, has been both theoretically and practically proved to be an effective method in these years. However, this technique is confined in registering two images and hard to apply to multiple ones. The reason is that unlike mutual information between two variables, high-dimensional mutual information is ill defined. In textbooks and theoretical essays, three-dimensional mutual information is proposed based on Venn diagram. Unfortunately, mutual information defined in this way is not necessarily nonnegative. In order to overcome the problem, in this paper, we introduce the mutual information matrix. By calculating its eigenvalues, high-dimensional mutual information is defined. This definition is nonnegative, bounded, and could be extended to higher dimensions, thus enables us to register more than three images. In the end, this definition is tested and proved to be effective on registration of multiple US images through simulation.     

透射和反射高光谱成像的马铃薯损伤检测比较研究

针对马铃薯损伤部位随机放置会影响检测精度的问题,提出从正对相机、背对相机及侧对相机三个方向,应用透射和反射高光谱成像技术采集马铃薯图像,进行透射和反射高光谱成像的马铃薯损伤检测比较研究。对透射和反射高光谱图像进行独立成分(IC)分析和特征提取,利用所得特征对反射图像进行二次IC分析,对透射和反射光谱进行变量选择,最终分别建立基于反射图像、反射光谱、透射光谱的马铃薯损伤定性识别模型;对识别准确率高的模型做进一步优化,采用子窗口排列分析(SPA)算法对透射光谱的特征做二次选择得到3个光谱变量,并建立任意放置的马铃薯损伤识别最优模型。试验结果表明,基于反射图像、反射光谱建立的模型识别准确率较低,其中基于反射图像的马铃薯碰伤,侧对相机识别准确率最低为43.10%;基于透射光谱信息建立的模型识别准确率较高,损伤部位正对、背对相机的识别准确率均为100%,侧对相机为99.53%;马铃薯损伤识别最优模型对任意放置的损伤识别准确率为97.39%。应用透射高光谱成像技术可以检测任意放置方向下的马铃薯损伤,该研究可为马铃薯综合品质的在线检测提供技术支持。