CN103559489B - A non-contact method of extracting features of the palm imaging mode - Google Patents

A non-contact method of extracting features of the palm imaging mode Download PDF

Info

Publication number
CN103559489B
CN103559489B CN201310589834.6A CN201310589834A CN103559489B CN 103559489 B CN103559489 B CN 103559489B CN 201310589834 A CN201310589834 A CN 201310589834A CN 103559489 B CN103559489 B CN 103559489B
Authority
CN
China
Prior art keywords
palm
point
line
gradient
inscribed circle
Prior art date
Application number
CN201310589834.6A
Other languages
Chinese (zh)
Other versions
CN103559489A (en
Inventor
李威
苑玮琦
Original Assignee
沈阳工业大学
Priority date (The priority date is an assumption and is not a legal conclusion. Google has not performed a legal analysis and makes no representation as to the accuracy of the date listed.)
Filing date
Publication date
Application filed by 沈阳工业大学 filed Critical 沈阳工业大学
Priority to CN201310589834.6A priority Critical patent/CN103559489B/en
Publication of CN103559489A publication Critical patent/CN103559489A/en
Application granted granted Critical
Publication of CN103559489B publication Critical patent/CN103559489B/en

Links

Abstract

本发明涉及一种非接触成像方式下手掌特征提取方法,首先提取指根部的手掌内切圆,从内切圆圆心做若干条放射状线段相交于圆周,同时,构造模板计算内切圆内各点的梯度值,根据各线段上各点的梯度值计算质心相对半径,将其定义为特征参量,以此构成特征向量空间。 The present invention relates to a palm at a non-contact imaging mode feature extraction method, first extracting means inscribed circle palm root, do several pieces of the radial line segment intersects the circumference from the cutting center of the circle, while the cut points circular inner configuration template Calculation the gradient value, the gradient value of each point on the radius of each segment relative to the centroid is calculated, which is defined as the characteristic parameter, thereby constituting the feature vector space. 本发明所构造的特征具有平移、旋转和比例缩放不变性,同时,特征提取时间较快等优点。 The construction of the invention characterized in translation, rotation and scaling invariant, while the advantages of rapid feature extraction time.

Description

一种非接触成像方式下手掌特征提取方法 A non-contact method of extracting features of the palm imaging mode

技术领域: FIELD:

[0001] 本发明属于生物特征识别技术领域,涉及一种非接触成像方式下手掌特征提取方法。 [0001] The present invention belongs to the field of biometric technology, relates to a non-contact imaging the palm embodiment feature extraction method.

背景技术: Background technique:

[0002] 随着时代的发展和社会的进步,门禁系统越来越需要高效、可靠的身份认证机制来确定人员对特定区域进出的合法性,生物特征识别技术是一种根据人体所固有的生理特征或行为特征来识别身份的技术。 [0002] With the development of the times and society, access control systems increasingly require efficient and reliable authentication mechanism to determine who out of a specific area of ​​legality, is a biometric technology based on the intrinsic physical human body characteristics or behavioral characteristics to identify the identity of techniques. 因为生物特征具有“人人拥有、人各不同、长期不变”的特点,并且不会被遗忘或丢失,被越来越多的应用在门禁系统中。 Because biometrics have "everyone has, people of all different, long-term change" features, and will not be forgotten or lost, more and more applications in the access control system. 根据已有的相关文献可知, 目前可用于门禁系统的生物特征包括指纹、掌纹、静脉、人脸、虹膜、视网膜、人耳等。 The existing literature known, currently available for biometric access control system includes a fingerprint, palm print, vein, face, iris, retina, and other human ear. 手部特征因为被用户接受程度高以及采集图像时更加方便,在众多生物特征当中得到了更广泛的应用。 Characterized in hand because the high degree of user acceptance and more convenient image acquisition, in many biometrics which has been more widely used.

[0003] 人手平时处于半握拳状态,手掌内的特征相对手指特征和手背特征更不容易被窃取,具有更高的安全性。 [0003] usually in a semi-manual fist state, characterized in opposing fingers in the palm and back of the hand characterized wherein less likely to be stolen, with higher security. 手掌静脉是隐藏于人体内部的活体生物特征,相对手的外部纹理特征,静脉特征更不易于被窃取和复制。 Palm vein is hidden inside the body of the living organism wherein the outer texture features of opposite hand, the vein characteristic less susceptible to being stolen and copied. 但因为手掌自身的生理结构,有些手掌在近红外光下采集的图像中静脉信息几乎目测不到,手掌内的纹理主要由掌纹主线和一些乳突纹组成。 However, because of its physical structure palm, the palm image capture at some near-infrared light vein information is almost not visually, in texture mainly composed of palm and palm several main lines mastoid composition. 因此,本文使用单一采集设备,在近红外光照射下对人的手掌进行拍摄,将得到的手图像中手掌区域所包含的手掌静脉信息和手掌掌纹主线信息定义为手掌特征,将其用于身份识别,如图1所示。 Thus, a single collection equipment used herein, the human hand to shoot under irradiation of near-infrared light, a palm vein information and palm information defining the main palmprint hand image obtained in the palm region of the palm features included, which was used identification, as shown in FIG.

[0004] 手掌静脉识别的最初概念出现在20世纪90年代,因为日本富士通公司手掌静脉识别仪器的推广而在2006年以来得到广泛研究。 [0004] The original concept palm vein recognition appeared in the 1990s, because the promotion of Fujitsu palm vein recognition instrument has been widely studied since 2006. 在2006年到2010年,手采集设备都是基于接触式设计的。 In 2006 and 2010, collecting devices are based on hand-contact design. 接触式采集方式要求用户在采集图像过程中人手与采集设备发生接触,或是握住采集设备的一些外部设备或是将手放在固定摆放位置的固定拴上。 Contacting acquisition mode requires the user to manually brought into contact with the capture device in the image acquisition process, or some external device holding the collecting device or the hand on the fixed placement of fixed fasten. 接触式采集在卫生和安全性方面会带来一些问题,同时,接触式采集仪的传感器表面更容易被污染,尤其是门禁系统经常应用在卫生条件较差的室外环境中,这会导致系统的误拒率上升,同时也缩短了采集仪器的使用寿命。 Contact acquisition health and safety can cause problems, while the contact surface of the sensor type acquisition instrument more easily contaminated, in particular access control system is often used in conditions of poor hygiene outdoor environments, which can lead to system false rejection rate increases, but also shorten the life of the instrument collection.

[0005] 在2010年开始出现针对非接触成像方式的手掌特征识别方法的研究。 [0005] Research for palm recognition method wherein a non-contact imaging modalities began to emerge in 2010. 通过非接触方式实现手掌成像能够使手识别仪在担心疾病传染的人群中得到认可,但手与成像装置之间位置的不确定导致手在图像中存在平移、旋转和比例缩放,为提取稳定的手特征提出了更高的要求。 Non-contact manner to identify the hand palm imaging device can be recognized in the fear of infectious disease in the population, but an undefined position between the hand and the image forming apparatus leads to the presence of the hand in the image translation, scaling and rotation, is extracted by stable hand features a higher requirement.

发明内容: SUMMARY:

[0006] 发明目的: [0006] The object of the invention:

[0007] 本发明涉及一种非接触采集方式下手图像中手掌特征提取方法,其主要目的在于提高非接触采集方式下手识别的准确性,解决因手与成像装置间位置和角度的变化所导致的手图像中手平移,旋转和比例缩放问题。 [0007] The present invention relates to a non-contact manner to start collecting the palm image feature extraction method, its main purpose is to improve the accuracy of the non-contact identification start acquisition mode, address the change in the position and the angle between the hand and the image forming apparatus caused by hand in hand image translation, rotation and scaling problems.

[0008] 技术方案: [0008] Technical Solution:

[0009] 本发明是通过以下技术方案来实现的: [0009] The present invention is achieved by the following technical solution:

[0010] —种非接触成像方式下手掌特征提取方法,其特征在于:该方法步骤如下: [0010] - wherein the kind of non-contact imaging modalities palm extraction method, characterized in that: the method steps are as follows:

[0011] (1)掌心稳定参考点的选择:构造一个掌心内切圆来选择掌心稳定的参考点;设计内切圆具备如下条件:分别与手掌两侧轮廓线相切,并通过中指、无名指的指根点;依次分别对位于食指外侧手掌轮廓线上和小手指外侧手掌轮廓线上的手掌边缘点进行扫描,确定内切圆圆心的位置以及内切圆的半径; [0011] (1) Select the palm of a stable reference point: the configuration of a palm palm inscribed circle to select a stable reference point; inscribed circle within the design have the following conditions: tangentially on both sides of the palm contour, and by the middle finger, ring finger It refers to the root node; are sequentially located the index finger side of the hand contour and the outer contour of the palm of the little finger edge of the palm point scanning, determining the location of the inscribed circle center and radius of the inscribed circle;

[0012] (2)手掌不变特征的提取:选取手掌静脉和手掌掌纹两种手掌部特征进行身份识别系统的实现;通过内切圆圆心获得掌心稳定参考点,对手图像进行定位,消除采集图像过程中手掌平移的影响;以掌心参考点为中心点,以角度Qt3为单位射出若干条射线,在掌心参考点与内切圆轮廓线间构造若干条放射状线段,称为特征线段;最后,根据各特征线段上各点的梯度强度值计算线段的质心相对半径,构造特征向量。 [0012] (2) palm invariant feature extraction: selecting a palm vein, palm and palm characteristics for two kinds of portions realized palm identification system; stable reference point obtained by the palm of the inscribed circle center, the opponent image positioning, eliminating collection Effect of the palm image translation process; palm reference point to the center point, angle Qt3 plurality of rays emitted from the unit, the contour line inscribed circle in the palm between the reference point and the inner radial segment configured several pieces, called the characteristic line; finally, the gradient magnitude values ​​of each feature point on the line segment is calculated relative to the radius of the centroid of the feature vector construct.

[0013]上述步骤(2)中构造特征向量的具体步骤如下: [0013] In the above step (2) the specific step of constructing a feature vector as follows:

[0014] 步骤1:梯度图像的生成 [0014] Step 1: generating a gradient of the image

[0015] 选择梯度强度值作为特征的计算依据,在特征提取前将手灰度图像转换成梯度图像: [0015] as the selection gradient intensity value calculated on the basis of the feature, the feature extraction prior to be converted into a grayscale image gradient hand image:

[0016] 1)在与手掌静脉或者手掌掌纹主线相交方向,满足灰度极小值条件,因此,通过四个方向(0^45'90'135°)灰度极小值判断,能够找到手掌静脉或者手掌掌纹候选像素点; 而不满足灰度极小值的像素点一定不是静脉或者掌纹,因此,将该像素点梯度值置零; [0016] 1) in the palm and palm vein or palm intersecting main direction, to meet the minimum gradation value condition, therefore, the four directions (0 ^ 45'90'135 °) intensity minimum value determination, it is possible to find palm vein or palm palm candidate pixels; pixel gray level does not satisfy the minimum value is not constant or palm vein, and therefore, the gradient of the pixel values ​​of zero;

[0017] 2)在满足灰度极小值的像素点中,像素点位于静脉或者掌纹之内,在其垂直方向的两个梯度值都将比较大;当像素点偏离静脉或者掌纹时,梯度值将出现一大一小现象,因此,取符合灰度极小值方向的两个梯度值中的较小值作为该方向的梯度值,再取符合灰度极小值的所有方向梯度值中的最大值作为该像素点的梯度值,该梯度值能够反映沿静脉或掌纹垂直方向梯度变化趋势,即梯度最大值对应于静脉或掌纹的中心位置,随着偏离中心点,梯度逐渐减小; When departing from the vein or palm pixels; [0017] 2) in the pixel point satisfies the minimum gradation value, or intravenously located pixels of the palm, in which two vertical gradient value are relatively large , gradient occurs and a small value and, therefore, takes a smaller value of the two in line with the gray value gradient direction minimum value as the value of the gradient direction, and then take all directions in line with the minimum value of gradation gradient the maximum value of as the gradient value of the pixel, the gradient value can reflect the vertical direction along the vein or palm trend gradient, i.e., the maximum gradient corresponding to the center position of the vein or palm print, offset from the center point with a gradient slowing shrieking;

[0018] 步骤2:特征线段质心点相对半径的提取 [0018] Step 2: centroid point line wherein the radius of the opposing extraction

[0019] 从掌心参考点的零度方向出发,沿逆时针方向将内切圆等分为H个扇形区域,由此获得H条线段K1A2,……,KH;特征线段质心与掌心参考点的距离,即质心半径,用r表示,第j 条特征线段的质心半径为: [0019] From the palm zero reference point direction, the counterclockwise direction H of the inscribed circle is equally divided into two segment areas, thereby obtaining a strip line H K1A2, ......, KH; line from the centroid feature palm reference point , i.e., the centroid radius, denoted by r, j-th line segments centroid radius:

Figure CN103559489BD00041

(1) (1)

[0021] 其中,特征线段即为内切圆半径,为方便计算,将内切圆半径取整数,即公式⑴中的i从1到M,M为内切圆在水平方向上半径包含的像素点个数; [0021] wherein, wherein the line segment is the radius of the inscribed circle, to facilitate calculation, the radius of the inscribed circle rounded, i.e. in the formula ⑴ i from 1 to M, M is the radius of the inscribed circle of pixels contained in a horizontal direction The number of points;

[0022] 除了水平和垂直方向之外,特征线段上的单位点i 一般不会落在像素点上,其梯度值采用插值方法来计算,采用双线性插值方法,利用单位点的4个最近邻像素的梯度值来计算单位点的梯度值; [0022] In addition to horizontal and vertical directions, wherein the single site on the line i does not generally fall on the pixel, the gradient value which is calculated using the interpolation method, a bilinear interpolation method, using a single point 4 nearest o gradient value of the pixel value gradient calculating unit point;

[0023] 为了解决比例缩放问题,对质心点与内切圆圆心的相对半径心进行计算,计算公式如下: [0023] In order to solve the scaling problem, the radius of the centroid point opposite the center of the inscribed circle center calculation, is calculated as follows:

Figure CN103559489BD00051

(2) (2)

[0025] 利用H条特征线段质心点与内切圆圆心的相对半径构造平移、旋转和比例缩放变换下的不变特征,特征向量为: [0025] H using a strip line segments centroid center of the circle and the endo configuration relative radius translation, rotation and scale-invariant feature transform the feature vectors of:

[0026] F= {δι,δ2,···,δΗ}; [0026] F = {δι, δ2, ···, δΗ};

[0027] 步骤3:特征向量的匹配 [0027] Step 3: matching feature vectors

[0028] 经过上述步骤,在手掌内构造H条能够反映手掌纹理特征的线段,设从注册手图像中提取的特征向量为: [0028] After the above procedure, the palm of the hand structure H strip line to reflect the palm texture feature, provided in the hand image extracted from the registration feature vector as:

Figure CN103559489BD00052

[0030]从待识别手图像中提取的特征向量为: [0030] extracted from the hand image to be recognized in the feature vector as:

Figure CN103559489BD00053

[0032] 判定注册手图像与待识别手图像是否匹配,先判定二者间有多少匹配线段对,在线段对匹配中,计算特征线段间的特征向量距离山作为2幅图像中对应线段质心相对半径的相似性判据,注册手图像和待识别手图像中线段j的特征向量距离计算公式如下: [0032] Registrant hand image to be identified whether the hand image matching, first determine the number of matching pairs of line segments between them, the line segment matching, the eigenvectors are computed between the line segments from the mountain as two image corresponding segment centroid relative similarity criterion radius, the line segment j of the feature vector to be recognized and registered hand images from the hand image is calculated as follows:

[0033] dj=|5ja-5jb (3) [0033] dj = | 5ja-5jb (3)

[0034] 当山小于设定阈值!^时,则认为两条对应线段为匹配线段,如果匹配线段对有g个, 则匹配率为: ! [0034] When less than a set threshold value ^ Hill, is considered to match the two segments corresponding to the segment, if there g of a matching segment, the match rate is:

Figure CN103559489BD00054

(4) (4)

[0036] 当匹配率G大于设定阈值!^时,则认为两幅手图像来源于同一个人。 [0036] When the matching rate is greater than the set threshold value G! ^, Two hand image is considered from the same individual.

[0037] 优点和效果: [0037] The advantages and effects:

[0038] 本发明所构造的特征具有平移、旋转和比例缩放不变性,同时,给出的特征提取方法运行速度较快,更适用于对实时性要求较高的非接触手识别装置。 [0038] The features of the present invention constructed in translation, rotation and scaling invariant, while the feature extraction method given run faster, more suitable for real-time requirements of the non-contact hand recognition means.

附图说明: BRIEF DESCRIPTION OF:

[0039] 图1为非接触方式下采集到的手掌图片; [0039] FIG 1 under a non-contact manner to the palm image capture;

[0040] 图2为手掌内切圆检测不意图; [0040] Figure 2 is not intended to detect an inscribed circle within the palm of the hand;

[0041] 图3为非接触成像方式下手掌特征提取方法流程图; [0041] FIG. 3 wherein the non-contact imaging pattern extracting palm flowchart of a method;

[0042]图4为梯度取值示意图; [0042] FIG. 4 is a schematic diagram of the gradient values;

[0043] 图5为梯度计算模板,图5 (a)为梯度计算模板示意图,图5 (b)为1为2时子模板中心点位置不意图,图5 (c)为1为3时子模板中心点位置不意图; [0043] FIG. 5 is a gradient calculation template, FIG. 5 (a) is a schematic gradient calculation template, FIG. 5 (b) is a sub-template is not intended center point position, FIG. 5 (c) is 1 to 3:00 sub template center point is not intended;

[0044] 图6为基于掌静脉和掌纹特定点位置的手掌不变特征构造图,图6 (a)为内切圆内构造特征线段图,图6 (b)为特征线段示意图,图6 (c)为特征线段上像素点灰度值示意图,图6 (d)为特征线段质心点示意图。 [0044] FIG. 6 is a configuration diagram based on palm and palm particular palm vein invariant feature point position, FIG. 6 (a) is a structure wherein the inner circle tangential line, and FIG. 6 (b) is a characteristic line diagram of FIG. 6 (c) is a schematic view of the pixel gray values ​​on the line segments, FIG. 6 (d) is a schematic line segments centroid point.

具体实施方式: Detailed ways:

[0045] 下面结合附图和具体的实施方式对本发明做进一步的说明: [0045] The present invention will be further illustrated in conjunction with the accompanying drawings and specific embodiments:

[0046] 本发明是一种非接触成像方式下手掌特征提取方法,其特征在于:利用非接触成像方式采集手图像时,手相对成像装置的方向、位置和拍摄距离是不确定的,这导致对于同一个人,每次采集的手图像中手的大小、方向和位置可能不相同。 [0046] The present invention is characterized in the palm of a non-contact mode imaging extraction method, wherein: when using the hand image acquired imaging a non-contact manner, the direction of the hand relative to the image forming apparatus, the position and shooting distance is uncertain, which leads for the same person, the hand image captured in each hand size, orientation, and location may not be the same. 本发明这种方法直接提取与平移、旋转和比例缩放无关的特征,避免了对图像中进行归一化过程中可能会改变原有图像的部分性质和增加了运行时间。 This method of the present invention is directed to extraction with translation, rotation and scaling features independent, it avoids the image normalizing process might change the nature of the original image portion and increasing the operating time. 该方法步骤如下: The steps are as follows:

[0047] (1)掌心稳定参考点的选择:为了利用手掌静脉和掌纹主线上各点与掌心稳定参考点的空间相对位置构造与手掌旋转、平移和比例缩放无关的特征,通过构造一个掌心内切圆来选择掌心稳定的参考点;为了保证内切圆具有唯一性,设计内切圆具备如下条件:分别与手掌两侧轮廓线相切,并通过中指、无名指的指根点;由圆的几何性质可知,圆边界上任意两点的中垂线必经过圆心,且圆的切线必垂直于经过该切点的半径,根据这两条性质, 依次分别对位于食指外侧手掌轮廓线上和小手指外侧手掌轮廓线上的手掌边缘点进行扫描,确定内切圆圆心的位置以及内切圆的半径;步骤如下: [0047] (1) selected palm stable reference point: the relative position is configured to be rotated by the palm and palm space on a palm vein main points and the reference point palm stability, translation, and scaling features unrelated by constructing a palm palm inscribed circle to select a stable reference point; inscribed circle in order to ensure uniqueness, the inscribed circle within the design have the following conditions: the palm sides are tangent to the contour line, and through the middle finger, ring finger base points; by a circle geometric properties seen, any two points on a circular boundary must be perpendicular passes through the center and perpendicular to the tangent to the circle shall radius through the tangent point, according to these two properties, are sequentially positioned outside of the index finger and the palm contour palm palm little finger edge point outside contour is scanned to determine the position of the inscribed circle center and radius of the inscribed circle; the following steps:

[0048] 1)利用基于方向梯度极值的手形轮廓跟踪方法提取手掌轮廓线。 [0048] 1) The method of using the hand contour tracing direction of the gradient extreme value based on the extracted contour line palm.

[0049] 2)提取中指、无名指指根AP1,具体实现方法如下:对于左手成像,其拇指在图像中朝上,无名指位于中指之下,从中指指尖出发,沿步骤1)提取的轮廓线找到第一个曲率变化大的点,该点即为中指与无名指指尖交界处C,再在对应的灰度图像中,从C点出发,运用灰度极小值跟踪方法找到中指与无名指指缝线;设位于指缝线上的任一像素点为Pn,寻找其在垂直指缝方向上的两个点P1^PP13,该两点与点Pn的距离为3个像素宽;分别计算点P11 与点Pi2、点P13的梯度值,将两个梯度值的平均值AVG作为评价点Pn在垂直指缝方向上梯度变化量;以C点为起始点,对位于指缝线上的每个像素点依次计算其梯度变化量AVGjAVG 发生明显减小时,跟踪停止,前一点即为中指、无名指的指根点Pl,将该点标记在手轮廓线二值图像中。 [0049] 2) Extraction middle finger, ring finger base AP1, the specific method is as follows: For the left-hand image, in which image thumbs-up, located below the middle finger, starting from the middle finger, along the extraction step 1) contour found a large change in curvature of the first point, the ring finger and the middle point is at the junction C, and then the gradation corresponding to the image, starting from the point C, the use of gray minimum tracking method finds the middle finger and ring finger suture; line disposed fingers positioned in any point of a pixel Pn, which is looking in the direction of two vertical fingers points P1 ^ PP13, two points away from the point Pn is three pixels wide; each calculation point Pi2 point P11, the point P13 of the gradient value, the average value AVG two gradient values ​​as the change amount of the evaluation point Pn gradient in the vertical direction, the fingers; C to the point as a start point of each line located fingers when the pixels which sequentially calculates the amount of change AVGjAVG gradient occurs significantly reduced, the tracking is stopped, the previous point is the middle finger, ring finger base point Pl, the point in the hand mark contour binary image.

[0050] 3)指根部内切圆提取,手掌内切圆自动检测方法具体实现步骤如下:a)确定内切圆与手掌轮廓线切点的范围,将该范围内的边缘点作为扫描对象。 [0050] 3) means to extract the root inscribed circle, the inscribed circle within the palm of an automatic method for detecting the specific steps are as follows: a) determining the range of the palm of the inscribed circle of the contour tangent point, the edge points as the scanning target range. 首先,通过中指、无名指的指根AP1做直线L3,该直线垂直于中指和无名指间的指缝拟合直线。 First, middle finger, ring finger base AP1 straight line L3, perpendicular to the line between the middle and ring fingers fit straight line. 直线L3相交于手掌轮廓线,取下侧开始的第一个点为P2,第二点为P3。 The straight line L3 intersects palm contour, remove the first starting point P2 side, the second point P3. 其次,从点P3开始向手掌手腕方向做一系列平行线,平行线平行于线L3,每条平行线相交于轮廓线,取下侧开始的第二个点,构成点系列,当点的纵坐标发生跳跃时,取该点为P4。 Secondly, started from the point P3 to the direction of the wrist palm series of parallel lines, parallel to the line L3 parallel lines, each parallel to the lines intersecting contour, remove the second side of the start point, the point constituting the series, when the point of the longitudinal coordinate jump occurs, taking the point P4. 最后,构造两个点集合AdPA2。 Finally, construct two point set AdPA2. 第一个点集合A1 由从点P3到点P4间的边缘线上的点构成;第二个点集合A2由手腕边缘点Q到点P2间的边缘线上的点构成。 The first consists of a set of points from point A1 to point P3 of the edge line between the points P4; a second set point A2 constituted by the edge of the wrist the point Q to point on the edge line between the point P2. 点集合心和知即为内切圆与手掌轮廓线切点的范围,将该范围内的边缘点作为扫描对象,分别执行步骤b)和步骤chb)提取内切圆圆心候选点所在区域。 Heart and set point range known palm inscribed circle tangent point of the contour is the, the edge points as the scanning target range, respectively, step b) and step CHB) extracts a candidate region of the inscribed circle center point is located. 首先,选取集合A1或集合A冲的任意一点,设为AP5。 First, select a set of A1 or A punch set at any point, to AP5. 通过中指、无名指的指根点P1和边缘点内做连线,设线段为L2,对线段1^做中垂线L'2;其次,沿手掌轮廓线在点内左右各取2点,通过这些点做拟合直线L1,过P5点做直线1^的垂线L' i。 By the middle finger, the point P1 and the edge point of the finger ring do connection of line segment L2, on the vertical line 1 ^ do L'2; Secondly, along the contour of the palm of the left and right in two points from each point, by these points make fitting a straight line L1, a straight line through the point P5 of the vertical line 1 ^ L 'i. 直线L' i和直线L' 2相交于点O1。 The straight line L 'i and the straight line L' 2 at point O1. 对集合心和六2内每个点分别作上述操作,提取其相交点〇1,分别得到点集合A' 4PA' 2。 For each point within the set of six center and 2 respectively for the above-described operation, the point of intersection 〇1 extracted, respectively to obtain a set of points A '4PA' 2. 计算集合A' :中的点与A' 2中点的欧式距离,将距离小于ΔΑ的点保留,构成集合A'(本发明取ΔΑ=6),对集合A'中各点的横、 纵坐标进行比较,得到minX,minY,maxX,maxY,构成一个矩形区域,该矩形区域内各像素点为内切圆圆心的候选点。 Calculating a set of A ': the point A' 2 midpoint of the Euclidean distance, the distance point is less than Delta] [alpha retention of the configuration set A '(of the present invention taken ΔΑ = 6), to set A' for each point in the cross-vertical comparing the coordinates, to obtain minX, minY, maxX, maxY, constitute a rectangular area, each of the pixels within the rectangular region is the candidate point of the inscribed circle center. c)确定内切圆圆心和内切圆半径。 c) determining the center of the circle and the inscribed radius of the inscribed circle. 首先计算圆心候选点所在区域内每一个圆心候选点与食指外侧手掌轮廓线和小手指外侧手掌轮廓线的距离,计算方法如下:选取圆心候选点所在区域内的任一圆心候选点,设为0',分别计算圆心候选点0'与集合A^A2中各点的欧式距离,取最小值,设为cU和d2,距离山和办即为像素点0'到两侧手掌轮廓线的距离。 First, calculate the distance of each center candidate point and the outer finger palm contours of center candidate points Area and the little finger side of the hand contour, calculated as follows: Select a center candidate any point in the region of the center of the candidate point is located, is set to 0 ' , calculates the center candidate point 0 'the set a ^ A2 Euclidean distance of each point, a minimum value, and a cU is set d2, and do is the distance mountain 0 pixels' distance from the sides of the palm contour. 然后计算候选点0'与中指、无名指指根点距离为d3。 Then calculating candidate point 0 'and the middle finger, ring finger base point distance d3. 最后计算距离^、办和山的标准差,来评定三个距离相差的程度。 Finally, calculate the distance ^, standard office and the mountain poor, to assess the degree of difference between the three distances. 对每一个圆心候选点分别做上述操作,在得到的标准差系列中选出最小的,对应的候选点即为内切圆圆心0 (Xe,y〇),该点与中指、无名指与手掌相交界点距离即为内切圆半径Ld)以点0(x〇,y〇)为圆心,R为半径,画出手掌内切圆, 剔除圆外信息,即为手掌静脉特征提取区域。 For each candidate center point of the operation are done, to select the smallest, the corresponding cutting candidate point is within the circle center 0 (Xe, y〇) obtained in the standard deviation in the series, the point with the middle, ring and palms the junction of the inscribed circle radius is the distance Ld) 0 point (x〇, y〇) as the center, R is the radius of inscribed circle drawn inside the palm, the outer circle excluding information, i.e. palm vein feature extraction region.

[0051] (2)手掌不变特征的提取:在近红外光下采集的手图像中,手掌静脉呈现网状,与手掌掌纹主线一起呈现出天然纹理状态,因此一幅手图像可以看成是一幅纹理图像,手掌静脉是手掌纹理特征的重要组成部分。 [0051] (2) palm invariant feature extraction: hand image acquired in the near-infrared light, rendering mesh palm vein, showing a state with natural texture main palm palm, so images can be seen as a hand is a texture image, a palm vein texture feature is an important part of the palm. 但因为手掌自身的生理结构,有些手掌在近红外光下采集的图像中静脉信息几乎目测不到,手掌内的纹理主要由掌纹主线和一些乳突纹组成。 However, because of its physical structure palm, the palm image capture at some near-infrared light vein information is almost not visually, in texture mainly composed of palm and palm several main lines mastoid composition. 掌纹主线和乳突纹不仅具有很强的方向性,而且粗细深浅不一,因此,本发明选取手掌静脉和手掌掌纹两种手掌部特征进行身份识别系统的实现;通过内切圆圆心获得掌心稳定参考点,对手图像进行定位,消除采集图像过程中手掌平移的影响;以掌心参考点为中心点,以角度Θ*3为单位射出若干条射线,在掌心参考点与内切圆轮廓线间构造若干条放射状线段(线段长度等于内切圆半径长度),称为特征线段;最后,根据各特征线段上各点的梯度强度值计算线段的质心相对半径(即质心和内切圆圆心的绝对距离与内切圆半径之比),构造特征向量。 Palmprint mastoid main groove and not only has a strong directivity, and the thickness shades, therefore, the present invention is selected palm vein, palm and palm characteristics for two kinds of portions realized palm identification system; obtained by cutting the center of the circle palm stable reference point, the opponent image positioning, eliminate the influence of the captured image during a translation of the palm; palm reference point to the center point, an angle Θ * 3 units of rays emitted from a plurality of cut lines in the palm of circular profile with the inner reference point inter configured several pieces of the radial line (line segment length is equal to the inscribed circle radius length), referred to as line segments; Finally, the line segment is calculated based on the gradient intensity value of each point on each of the line segments centroid relative radius (i.e., the centroid and endo center of the circle absolute distance ratio of an inscribed circle with radius), the feature vector construct.

[0052] 上述步骤⑵中构造特征向量的具体步骤如下: [0052] The specific steps in the above step ⑵ feature vector configuration as follows:

[0053] 步骤1:梯度图像的生成 [0053] Step 1: generating a gradient of the image

[0054] 手掌静脉和掌纹空间位置的提取并非易事,即使较少的错误也会给识别带来较大的误差。 [0054] Extraction palm and palm vein spatial position is not easy, even fewer errors will bring a large error recognition. 为此,本发明选择梯度强度值作为特征的计算依据,在特征提取前将手灰度图像转换成梯度图像,转换原理如下: To this end, the present invention is selected as a basis for computing the gradient intensity value characteristic, before feature extraction converts an image into a gray-scale gradient hand image conversion works as follows:

[0055] 1)在与手掌静脉或者手掌掌纹主线相交方向,满足灰度极小值条件,因此,通过四个方向(0^45'90'135°)灰度极小值判断,能够找到手掌静脉或者手掌掌纹候选像素点; 而不满足灰度极小值的像素点一定不是静脉或者掌纹,因此,将该像素点梯度值置零; [0055] 1) in the palm and palm vein or palm intersecting main direction, to meet the minimum gradation value condition, therefore, the four directions (0 ^ 45'90'135 °) intensity minimum value determination, it is possible to find palm vein or palm palm candidate pixels; pixel gray level does not satisfy the minimum value is not constant or palm vein, and therefore, the gradient of the pixel values ​​of zero;

[0056] 2)在满足灰度极小值的像素点中,像素点位于静脉或者掌纹之内,在其垂直方向的两个梯度值都将比较大;当像素点偏离静脉或者掌纹时,梯度值将出现一大一小现象,因此,取符合灰度极小值方向的两个梯度值中的较小值作为该方向的梯度值,再取符合灰度极小值的所有方向梯度值中的最大值作为该像素点的梯度值,该梯度值能够反映沿静脉或掌纹垂直方向梯度变化趋势,即梯度最大值对应于静脉或掌纹的中心位置,随着偏离中心点,梯度逐渐减小; When departing from the vein or palm pixels; [0056] 2) in the pixel point satisfies the minimum gradation value, or intravenously located pixels of the palm, in which two vertical gradient value are relatively large , gradient occurs and a small value and, therefore, takes a smaller value of the two in line with the gray value gradient direction minimum value as the value of the gradient direction, and then take all directions in line with the minimum value of gradation gradient the maximum value of as the gradient value of the pixel, the gradient value can reflect the vertical direction along the vein or palm trend gradient, i.e., the maximum gradient corresponding to the center position of the vein or palm print, offset from the center point with a gradient slowing shrieking;

[0057] 步骤2:特征线段质心点相对半径的提取 [0057] Step 2: centroid point line wherein the radius of the opposing extraction

[0058] 从掌心参考点的零度方向出发,沿逆时针方向将内切圆等分为H个扇形区域,由此获得H条线段K1A2,……,KH;特征线段质心与掌心参考点的距离,即质心半径,用r表示,第j 条特征线段的质心半径为: [0058] From the palm zero reference point direction, the counterclockwise direction H of the inscribed circle is equally divided into two segment areas, thereby obtaining a strip line H K1A2, ......, KH; line from the centroid feature palm reference point , i.e., the centroid radius, denoted by r, j-th line segments centroid radius:

Figure CN103559489BD00071

(1) (1)

[0060] 其中,特征线段即为内切圆半径,为方便计算,将内切圆半径取整数,即公式⑴中的i从1到M,M为内切圆在水平方向上半径包含的像素点个数; [0060] wherein, wherein the line segment is the radius of the inscribed circle, to facilitate calculation, the radius of the inscribed circle rounded, i.e. in the formula ⑴ i from 1 to M, M is the radius of the inscribed circle of pixels contained in a horizontal direction The number of points;

[0061] 除了水平和垂直方向之外,特征线段上的单位点i 一般不会落在像素点上,其梯度值采用插值方法来计算,本发明采用双线性插值方法,利用单位点的4个最近邻像素的梯度值来计算单位点的梯度值; [0061] In addition to horizontal and vertical directions, wherein the single site on the line i does not generally fall on the pixel, the gradient value which is calculated using the interpolation method, a bilinear interpolation method of the present invention, using a single point 4 nearest neighbors to calculate the gradient value of a pixel value gradient of the single site;

[0062] 由于手成像大小不一,计算出的内切圆半径不唯一,质心点相对圆心的距离随手成像大小而变化,为了解决比例缩放问题,对质心点与内切圆圆心的相对半径心(即质心点与内切圆圆心的质心半径^与内切圆半径R之比)进行计算,计算公式如下: [0062] Since the hand image sizes, the calculated radius of the inscribed circle is not unique centroid point opposite the center of the imaging distance readily varies the size, in order to solve the scaling problem, the centroid relative radius center point of the inscribed circle center (i.e., the centroid point of the inscribed circle center ^ centroid radius of the inscribed circle radius and the ratio R) is calculated, is calculated as follows:

Figure CN103559489BD00081

(2) (2)

[0064] 利用H条特征线段质心点与内切圆圆心的相对半径构造平移、旋转和比例缩放变换下的不变特征,特征向量为: [0064] H using a strip line segments centroid center of the circle and the endo configuration relative radius translation, rotation and scale-invariant feature transform the feature vectors of:

[0065] F= {δι,δ2,···,δΗ}; [0065] F = {δι, δ2, ···, δΗ};

[0066] 步骤3:特征向量的匹配 [0066] Step 3: matching feature vectors

[0067] 经过上述步骤,可以在手掌内构造H条可以反映手掌纹理特征的线段,设从注册手图像中提取的特征向量为: [0067] Through the above steps, the strip may be configured to reflect H segment palm texture features in the palm, is provided in the hand image extracted from the registration feature vector as:

[0068] Fa= {δι3,δ23, ··· ,5HaI , [0068] Fa = {δι3, δ23, ···, 5HaI,

[0069] 从待识别手图像中提取的特征向量为: [0069] extracted from the hand image to be recognized in the feature vector as:

[0070] Fb= {3ib,52b,…,δΗ1)}, [0070] Fb = {3ib, 52b, ..., δΗ1)},

[0071] 判定注册手图像与待识别手图像是否匹配时,需要先判定二者间有多少匹配线段对,在线段对匹配中,计算特征线段间的特征向量距离山作为2幅图像中对应线段质心相对半径的相似性判据,注册手图像和待识别手图像中线段j的特征向量距离计算公式如下: When [0071] the hand image registration determining the hand image to be identified are matched, it is determined how many need to match between the two segments, the segment matching the calculated feature vector distance between the feature line as the mountain image corresponding segments 2 similarity criterion relative to a radius of the centroid of the segment j of the feature vector to be recognized and registered hand images from the hand image is calculated as follows:

[0072] dj=|5ja-5jb (3) [0072] dj = | 5ja-5jb (3)

[0073] 当山小于设定阈值!^时,则认为两条对应线段为匹配线段,如果匹配线段对有g个, 则匹配率为: ! [0073] When less than a set threshold value ^ Hill, is considered to match the two segments corresponding to the segment, if there g of a matching segment, the match rate is:

Figure CN103559489BD00082

(4) (4)

[0075] 当匹配率G大于设定阈值!^时,则认为两幅手图像来源于同一个人。 [0075] When the matching rate is greater than the set threshold value G! ^, Two hand image is considered from the same individual.

[0076] 本发明涉及一种非接触成像方式下手掌特征提取方法,对手图像中手掌定位方法以及手掌特征提取方法进行了研究,其主要目的在于提高非接触采集方式下手识别的准确性,解决因手与成像装置间位置和角度的变化所导致的手图像中手平移,旋转和比例缩放问题。 [0076] The present invention relates to a non-contact imaging the palm embodiment feature extraction, image opponents palm and palm positioning feature extraction methods have been studied, its main purpose is to improve the accuracy of the non-contact identification start acquisition mode, address the image change hand position and angle between the hand and the image forming apparatus caused by the hand translation, rotation and scaling problems.

[0077] 本发明以非接触方式采集的手图像进行匹配时,需要将手进行归一化,从而使手掌特征与手成像造成的平移、旋转和比例缩放无关。 When the hand image [0077] In the present invention, a non-contact manner acquired by matching needs to be normalized to the hand, so that the palm of the hand imaging characteristics due to translation, rotation and scaling irrelevant. 由于归一化过程可能会改变原有图像的部分性质,而改变的这部分性质可能会影响特征的提取,同时,归一化过程也会增加运行的时间。 Since the normalization process may change the nature of part of the original image, which is part of the nature of the change may affect extraction features, at the same time, the normalization process will increase the running time. 因此,本发明选择直接构造了一种与平移、旋转和比例缩放无关的手掌特征。 Accordingly, the present invention is directed to select and construct a translation, rotation and scaling features unrelated to the palm.

[0078] 结合附图对具体步骤说明如下: [0078] Specific steps in conjunction with the accompanying drawings as follows:

[0079] (1)图1是在近红外光下通过非接触方式采集的手掌图像。 [0079] (1) FIG. 1 is in the near-infrared light collected by the non-contact manner palm image. 通过对大量的近红外光下采集的手掌图像进行观察,在手掌图像中可以清楚的看见掌纹主线信息,完全可以利用同一采集设备对手掌静脉和掌纹主线进行提取。 Palm image capture by a large number of the near-infrared light observation, the palm image can clearly see the main information palm, palm veins, and can be extracted using the same palm main acquisition device. 从同一幅手图像提取人手的掌静脉特征和掌纹主线特征,可以回避一些多模态生物特征融合的问题,如系统复杂度、成本、用户接受度、数据库管理等。 Extracting from the image of a human hand with the palm of a hand palm vein feature and main features, you can avoid some of the problems multimodal biometric fusion, such as system complexity, cost, user acceptance, and database management.

[0080] (2)图2为手掌内切圆检测示意图。 [0080] (2) FIG 2 is a schematic diagram of an inscribed circle within the palm detected. 对于非接触式的手识别系统来说,采集到的人手图像往往受到镜头距离、手运动、操作不当等因素的影响,造成图像之间存在很大的差异。 For hand recognition of non-contact system, the hand image acquired by the distance from the lens tends to affect movement of the hand, improper operation and other factors, resulting in the presence of a big difference between images. 为了消除图像差异对识别效果的影响,从手掌中提取出合适的基准点,建立新的参考坐标系,对手图像进行定位,以减少图像采集过程中引入的旋转、平移、比例缩放等因素的影响,提高匹配识别算法的鲁棒性。 Effects To eliminate the influence of the image difference of recognition results is extracted from the palm of a suitable reference point, to establish a new reference coordinate system, the opponent image positioning, in order to reduce the rotation of the image acquisition process is introduced, translation, scaling, and other factors to improve the robustness of the matching recognition algorithm.

[0081] (3)图3为非接触成像方式下手掌特征提取方法流程图,本方法提出一种能够反映掌脉和掌纹特征且不受旋转和比例缩放影响的手掌特征。 [0081] (3) in FIG. 3 wherein the non-contacting palm palm wherein imagewise flowchart extraction method, the present method provides a pulse to reflect the palm and palm features and is not affected by the rotation and scaling. 首先提取指根部的手掌内切圆, 从内切圆圆心做若干条放射状线段相交于圆周,同时,构造模板计算内切圆内各点的梯度值,根据各线段上各点的梯度值计算质心相对半径,将其定义为特征参量,以此构成特征向量空间。 First extracting means inscribed circle palm root, do several pieces of the radial line segment intersects the circumference from the cutting center of the circle, while the configuration of the template gradient value is calculated for each point of the inscribed circle, the centroid is computed in accordance with the gradient values ​​of the respective points on each segment relative radius, which is defined as the characteristic parameter, thereby constituting the feature vector space.

[0082] (4)图4为梯度取值示意图。 [0082] (4) FIG. 4 is a schematic diagram of the gradient values. 在手掌图像中,在与手掌静脉或者手掌掌纹主线相交方向,满足灰度极小值条件,因此,通过四个方向(Ot3,45°,90°,135°)灰度极小值判断,能够找到手掌静脉或者手掌掌纹候选像素点。 In the palm image, the palm or palm and palm vein intersecting main direction, to meet the minimum gradation value condition, therefore, the four directions (Ot3,45 °, 90 °, 135 °) intensity minimum value determination, You can find palm palm palm vein or candidate pixels. 而不满足灰度极小值的像素点一定不是静脉或者掌纹,因此,将该像素点梯度值置零。 Pixels not meet a certain minimum value than the gray or palm vein, and therefore, the pixel value is set to zero point of the gradient. 在满足灰度极小值的像素点中,像素点位于静脉或者掌纹之内,在其垂直方向的两个梯度值都将比较大;当像素点偏离静脉或者掌纹时,梯度值将出现一大一小现象,因此,取符合灰度极小值方向的两个梯度值中的较小值作为该方向的梯度值,再取符合灰度极小值的所有方向梯度值中的最大值作为该像素点的梯度值,该梯度值能够反映沿静脉或掌纹垂直方向梯度变化趋势,即梯度最大值对应于静脉或掌纹的中;L·、位置,随着偏尚中心点,梯度逐渐减小。 In the pixel point satisfies the minimum gradation value, or intravenously located pixels of the palm, in which two vertical gradient value are larger; when the pixel spot deviates from the vein or palm print, a gradient value appears two smaller value of the gradient value and a small and, therefore, take the minimum value in line with the direction of the gradation as the gradient value in that direction, and then take the maximum value in all directions in line with the minimum value of the gradation value of the gradient of as a gradient value of the pixel, the gradient value can reflect the vertical direction along the vein or palm trend gradient, i.e. a gradient corresponding to the maximum intravenous or palmprint; L ·, position, as the center point is still partial, gradient slowing shrieking.

[0083] (5)图5为梯度计算模板。 [0083] (5) FIG. 5 is a gradient calculation template. 利用构造的梯度模板可以将手灰度图像转换成梯度图像。 Using a gradient configuration template can be converted into a grayscale image gradient hand image.

[0084] (6)图6为基于掌纹和掌静脉特定点位置的手掌不变特征构造的原理图。 [0084] (6) 6 is configured invariant feature based on the position of the specific point and the palm palm palm vein schematic. 手掌特征构造的基本思想如下:在掌心参考点与内切圆轮廓线间构造若干条放射状线段,每条线段可能穿过若干条手掌静脉或掌纹纹线。 The basic idea of ​​the palm features configured as follows: several pieces of configuration segments radially between the reference point and the palm of the inscribed circle contour lines, each line segment may pass through several pieces of a palm vein or palm line patterns. 当用足够多的线段去覆盖内切圆内的手图像时,并结合对应的灰度变化信息就可以用这些线段来描述出内切圆内的手图像中掌纹和掌静脉的分布结构。 When used to cover a sufficient number of segments in the hand image of the inscribed circle, and the combined gray scale information corresponding to the line segments can be used to describe the structure of the distribution of the hand image in an inscribed circle of palm and palm veins.

[0085] 根据一阶不变矩原理可知,特征线段各点梯度值与该点相对掌心参考点距离的乘积之和等于该特征线段各点梯度值之和与质心点相对掌心参考点距离的乘积,由此,可以利用特征线段上各点的梯度值获得特征线段质心点相对掌心参考点的距离,将其定义为质心点的相对半径。 [0085] The first-order moment invariant principle known, line segments each point of the gradient value and the point opposite the palm of the reference point from the sum of products equal to the sum of the centroid point of the gradient values ​​of the respective points of the line segments relative to the palm of the reference point distance product , whereby the gradient may be utilized for each point on the characteristic line segment is obtained wherein the relative centroid point palm reference distance, which is defined as the relative radius of the centroid point.

[0086] 在手图像中,手掌静脉和掌纹主线处相对非纹线处具有更大的灰度突变,即具有更高的梯度值,对线段质心位置的影响起决定作用。 [0086] In the hand image, the palm and palm vein at the opposite main line pattern having a larger non-gradation mutations, i.e., have higher gradient values, the position of the line segment on heart quality determinative. 因此,可以利用特征线段的质心相对半径来描述特征线段上各点的灰度变化,进而体现出内切圆内的手图像中掌静脉和掌纹的分布结构。 Thus, line segments may be utilized in the centroid will be described relative radius of each point on the gradation characteristics change in the line, and further reflects the distribution and structure of palm veins in the palm of the hand image in the inscribed circle.

[0087] 选择线段质心与内切圆圆心的相对半径作为手掌特征,该特征确保手图像比例缩放不变性。 [0087] Selecting the wire relative to the centroid radius of the inscribed circle center of the palm as a feature, the feature ensures that the hand image scaling invariance. 选择掌心参考点到中指、无名指指根点的连线方向为零度方向,将其作为第一条特征线段的方向。 Selecting the reference point to the palm middle finger, ring finger root connection point of the direction of zero degrees direction, the direction of which line segment as the first feature. 由于该点的位置不随手在图像中的位置、方向和大小而变化,因此,第一条特征线段的方向具有唯一性。 Since the position of the point is not readily position in the image, the direction and size changes, therefore, the direction of the first segment of unique features. 根据以上分析,每条特征线段质心相对半径能够反映手掌静脉和掌纹在手空间的位置分布,以此作为手掌特征具有平移,旋转和比例缩放不变性。 Based on the above analysis, the centroid of each segment feature can reflect the relative radial distribution of palm and palm vein space position of the hand, the palm features as translation, rotation and scaling invariance.

Claims (1)

  1. I. 一种非接触成像方式下手掌特征提取方法,其特征在于:该方法步骤如下: (1) 掌心稳定参考点的选择:构造一个掌心内切圆来选择掌心稳定的参考点;设计内切圆具备如下条件:分别与手掌两侧轮廓线相切,并通过中指、无名指的指根点;依次分别对位于食指外侧手掌轮廓线上和小手指外侧手掌轮廓线上的手掌边缘点进行扫描,确定内切圆圆心的位置以及内切圆的半径; (2) 手掌不变特征的提取:选取手掌静脉和手掌掌纹两种手掌部特征进行身份识别系统的实现;通过内切圆圆心获得掌心稳定参考点,对手图像进行定位,消除采集图像过程中手掌平移的影响;以掌心参考点为中心点,以角度为单位射出若干条射线,在掌心参考点与内切圆轮廓线间构造若干条放射状线段,称为特征线段;最后,根据各特征线段上各点的梯度强度值计算线段的质心相对半径 I. A contactless palm wherein the imagewise extraction method, which is characterized in that: the method steps are as follows: (1) selecting a stable reference point palm: a configuration selected inscribed circle palm palm stable reference point; endo Design circle have the following conditions: the palm sides are tangent to the contour line, and through the middle finger, ring finger base point; are sequentially located the index finger side of the hand contour and the outer contour of the palm of the little finger edge of the palm scanning point, determined endo-position and the center of the circle inscribed circle radius; (2) palm invariant feature extraction: selecting a palm vein, palm and palm characteristics for two kinds of portions realized palm identification system; obtained by the inscribed circle center palm stable reference point, the opponent image positioning, eliminate the influence of the captured image during the palm translation; to palm reference point as the center point, in degrees emits a plurality of rays, the configuration of several pieces radially between the palm of the reference point and the inscribed circle contour line, called the characteristic line; Finally, segments based on the gradient strength calculated for each point on each of the line segments centroid relative radius ,构造特征向量。 Construct feature vectors.
CN201310589834.6A 2013-11-19 2013-11-19 A non-contact method of extracting features of the palm imaging mode CN103559489B (en)

Priority Applications (1)

Application Number Priority Date Filing Date Title
CN201310589834.6A CN103559489B (en) 2013-11-19 2013-11-19 A non-contact method of extracting features of the palm imaging mode

Applications Claiming Priority (1)

Application Number Priority Date Filing Date Title
CN201310589834.6A CN103559489B (en) 2013-11-19 2013-11-19 A non-contact method of extracting features of the palm imaging mode

Publications (2)

Publication Number Publication Date
CN103559489A CN103559489A (en) 2014-02-05
CN103559489B true CN103559489B (en) 2017-11-07

Family

ID=50013730

Family Applications (1)

Application Number Title Priority Date Filing Date
CN201310589834.6A CN103559489B (en) 2013-11-19 2013-11-19 A non-contact method of extracting features of the palm imaging mode

Country Status (1)

Country Link
CN (1) CN103559489B (en)

Families Citing this family (9)

* Cited by examiner, † Cited by third party
Publication number Priority date Publication date Assignee Title
CN103870831A (en) * 2014-03-13 2014-06-18 上海云享科技有限公司 Method and device for extracting palm print features and obtaining palm print images
CN104933432A (en) * 2014-03-18 2015-09-23 北京思而得科技有限公司 Processing method for finger pulp crease and finger vein images
CN105095849B (en) * 2014-05-23 2019-05-10 财团法人工业技术研究院 Object identifying method and device
US9400924B2 (en) 2014-05-23 2016-07-26 Industrial Technology Research Institute Object recognition method and object recognition apparatus using the same
CN104318213A (en) * 2014-10-21 2015-01-28 沈阳大学 Method for using human body palm biology information to identify identities
CN104392225B (en) * 2014-12-09 2017-06-16 西安电子科技大学 Template based on the gradient direction of the effective region online fast segmentation method palmprint
CN104636721B (en) * 2015-01-16 2018-09-21 青岛大学 One kind palmprint identification feature edge contour and texture Fusion
CN105426853A (en) * 2015-11-24 2016-03-23 成都四象联创科技有限公司 Human body characteristic identification method based on image
CN107358185A (en) * 2017-07-03 2017-11-17 上海奥宜电子科技有限公司 Palm print and palm vein image recognition method and device based on topology analysis

Citations (2)

* Cited by examiner, † Cited by third party
Publication number Priority date Publication date Assignee Title
CN102156888A (en) * 2011-04-27 2011-08-17 西安电子科技大学 Image sorting method based on local colors and distribution characteristics of characteristic points
US8229178B2 (en) * 2008-08-19 2012-07-24 The Hong Kong Polytechnic University Method and apparatus for personal identification using palmprint and palm vein

Patent Citations (2)

* Cited by examiner, † Cited by third party
Publication number Priority date Publication date Assignee Title
US8229178B2 (en) * 2008-08-19 2012-07-24 The Hong Kong Polytechnic University Method and apparatus for personal identification using palmprint and palm vein
CN102156888A (en) * 2011-04-27 2011-08-17 西安电子科技大学 Image sorting method based on local colors and distribution characteristics of characteristic points

Also Published As

Publication number Publication date
CN103559489A (en) 2014-02-05

Similar Documents

Publication Publication Date Title
Jain et al. An identity-authentication system using fingerprints
Woodard et al. On the fusion of periocular and iris biometrics in non-ideal imagery
Ma et al. Efficient iris recognition by characterizing key local variations
US20060147094A1 (en) Pupil detection method and shape descriptor extraction method for a iris recognition, iris feature extraction apparatus and method, and iris recognition system and method using its
KR100374708B1 (en) Non-contact type human iris recognition method by correction of rotated iris image
Kumar et al. Personal authentication using hand images
Belcher et al. Region-based SIFT approach to iris recognition
You et al. On hierarchical palmprint coding with multiple features for personal identification in large databases
US20080298642A1 (en) Method and apparatus for extraction and matching of biometric detail
Sun et al. Improving iris recognition accuracy via cascaded classifiers
US9141844B2 (en) System and method for three-dimensional biometric data feature detection and recognition
JP5031641B2 (en) Alignment method of pattern matching methods and matching device
Zhao et al. High resolution partial fingerprint alignment using pore–valley descriptors
Jain et al. Facial marks: Soft biometric for face recognition
US8442276B2 (en) Invariant radial iris segmentation
JP4305431B2 (en) The image processing apparatus, a mask generating method, and a program
JP2002109543A (en) Pattern center determination device, pattern direction determination device, pattern alignment device and pattern comparison device
Yu et al. The relative distance of key point based iris recognition
CN101317184A (en) Method for extracting features of an iris in images
CN101398886B (en) Rapid three-dimensional face identification method based on bi-eye passiveness stereo vision
CN101609499B (en) Rapid fingerprint identification method
CN101777128B (en) Fingerprint minutiae matching method syncretized to global information and system thereof
Han et al. Palm vein recognition using adaptive Gabor filter
Xiong et al. Model-guided deformable hand shape recognition without positioning aids
Kumar et al. Combining fingerprint, palmprint and hand-shape for user authentication

Legal Events

Date Code Title Description
C06 Publication
C10 Entry into substantive examination
GR01