CN101241550B - A method for judging the quality of iris images - Google Patents

Abstract

The present invention provides a method for judging the quality of an iris image. It first extracts the original normalized iris image through positioning and normalization operations, then performs 4-layer multi-resolution decomposition on the normalized iris image, and by counting the number of points with larger amplitudes of detail components at a resolution of ^2-1 and comparing them with a preset threshold, it can be judged whether the image has eyelid and eyelash occlusion problems, and then calculates the variance of the detail components at a resolution of ^2-2 and compares them with a preset threshold to judge whether the image clarity meets the requirements of the system. The iris image quality judgment method of the present invention has relatively stable performance in different lighting environments, and makes a quick and accurate judgment on the quality of the iris image.

Description

A method for judging the quality of iris images

【Technical field】

The invention belongs to the technical field of image processing, and mainly relates to iris identification technology in biometric identification.

【Background technique】

As computers, ATM machines, mobile phones, access control systems and other electronic devices continue to enter our daily life, it is becoming more and more urgent for personal security and convenient identity authentication technology. However, the existing systems based on smart cards, ID numbers and passwords can only hover between security and convenience. Sufficient security has never been achieved, while better security and inconvenience appear at the same time. In order to achieve higher security, we must use more complex and inconvenient passwords, because if we use the same password for different machines around us, we will increase the security risks while gaining convenience . People have been exploring some safe and convenient solutions for this reason. Biometrics offers a whole new field for this. Biometric technology is the use of human physiological or behavioral characteristics for identification and verification. These physiological characteristics include fingerprints, palm prints, voices, signatures, etc. to identify individuals. Because biological characteristics are neither forgotten like a password nor lost like a key, they are considered a more reliable method of personal identification. Iris identification is an emerging biometric technology that uses the iris as the basis for identification and has the advantages of high uniqueness, high stability, natural anti-counterfeiting and non-invasiveness. See literature for details: Anil K. Jain, Arun Ross, Salil Prabhakar, "An Introduction to Biometric Recognition", IEEE Transaction on Circuits and Systems for Video Technology, Volume 14, No.1, pp.4-20, 2004 and literature: John G. Daugman, "How Iris Recognition Works," IEEE Transaction on Circuits and Systems for Video Technology, Volume 14, Issue 1, pp.21-30, 2004.

Iris image quality assessment is a very important part in the whole automatic iris recognition technology, it ensures that the processed iris images meet the quality requirements of the system. Therefore, the misrecognition and refusal of recognition caused by the quality problem of the iris image itself are avoided. In practice, due to the focal length of the acquisition device during shooting, the rotation of the eyeball at the moment of shooting, and the partial occlusion of the iris by the eyelids and eyelashes, it is often impossible to perform subsequent feature extraction on the collected iris image. An effective iris image quality evaluation model has not been proposed in the existing algorithms, so we aim to establish a general and feasible evaluation model, see the literature for details: Chen Ji, Hu Guangshu, "Iris Image Quality Evaluation based on Wavelet Packet Decomposition", Journal of Tsinghua University (Sci & Tech), Volume 43, No.3, pp.377-380, 2003 and literature: Li Ma, Tieniu Tan, Yunhong Wang, Dexin Zhang, "Efficient Iris Recognition by Characterizing key Local Variations”, described in IEEE Transaction on Image Processing, Volume 13, No.6, pp.739-750, 2004.

At present, the existing iris quality judgment methods are:

(1) Method based on fast Fourier transform. It performs two-dimensional fast Fourier transform on the pixels in two rectangular blocks on the iris area, and then analyzes whether the image is clear and has eyelash occlusion through the statistics of its high-frequency, intermediate-frequency and low-frequency energy. The generality of this model is not strong, and it is easy to misjudge a clear iris image with less texture as a low-quality iris image. See literature for details: Li Ma, Tieniu Tan, Yunhong Wang, Dexin Zhang, "Personal Identification based on Iris Texture Analysis," IEEE Transactions on Pattern Analysis and Machine Intelligence, Volume 25, No.12, pp.1519-1533.

(2) The method based on wavelet packet decomposition. It selects the sub-band with the most concentrated distribution of texture high-frequency components as the characteristic sub-band, and uses its energy as the criterion for judging the image quality. The disadvantage of this method is that it cannot judge iris images which are problematic due to eyelash occlusion. See literature for details: Chen Ji, Hu Guangshu, "Iris Image Quality Evaluation based on Wavelet Packet Decomposition", Journal of Tsinghua University (Sci & Tech), Volume 43, No.3, pp.377-380, 2003.

(3) Methods based on image clarity, inner-outer eccentricity, and iris visibility. It establishes three indicators for measuring image quality, namely, image clarity, internal and external eccentricity, and iris visibility, and realizes the requirement of real-time quality evaluation of iris images. The disadvantage of this method is that it is more sensitive to light conditions and less stable. Xing Lei, Shi Pengfei, "A Quality Evaluation Method of Iris Image", Chinese Journal of Stereology and Image Analysis, Volume.8, No.2, pp.108-113, 2003.

The above-mentioned iris image quality judgment algorithms all have problems to a certain extent, such as excessive calculation, sensitivity to light, and poor versatility.

【Content of invention】

The purpose of the present invention is to establish a method for judging the quality of iris images with strong versatility, which can accurately detect iris images blocked by eyelids and eyelashes and iris images with insufficient clarity, and make the algorithm applicable to a wide range of lighting conditions.

The purpose of the present invention is achieved like this:

1, a kind of iris image quality judging method is characterized in that it comprises the following steps:

Step 1, through the camera device, the iris in the human eye is image-acquired, and the normalized iris image f(x, y) whose size is M×N is obtained from the original grayscale image containing the iris image; (x, y) Represents the coordinates of the pixel, f(x, y) represents the gray value of the pixel whose coordinates are (x, y);

和

其中，

是分辨率为2^j下的尺度系数，

是分辨率为2^j下小波系数，4层小波变换j的取值范围为{-1，-2，-3，-4}，i＝{H，V，D}，附加了水平、垂直和对角方向的细节；

为尺度函数，

为小波函数，选取的小波为DMeyer小波；进行二维小波变换的尺度函数

是两个一维尺度函数和的乘积；进行二维小波变换的水平方向敏感的小波函数

为一维小波函数ψ_j，m(x)和一维尺度函数

的乘积；进行二维小波变换的垂直方向敏感的小波函数为一维尺度函数

和一维小波函数ψ_j，n(y)的乘积；进行二维小波变换的对角方向敏感的小波函数

为两个一维小波函数ψ_j，m(x)与ψ_j，n(y)的乘积；Step 2, to the normalized iris image that obtains in step 1, carry out 4 layers of two-dimensional wavelet transform; Specifically, the formula of two-dimensional wavelet transform is:

and

in,

is the scale coefficient at a resolution of 2 ^j ,

is the wavelet coefficient with a resolution of 2 ^j , the value range of the 4-layer wavelet transform j is {-1, -2, -3, -4}, i={H, V, D}, and the horizontal, vertical and Diagonal details;

is a scaling function,

is a wavelet function, and the selected wavelet is DMeyer wavelet; the scaling function for two-dimensional wavelet transformation

are two one-dimensional scaling functions and The product of the horizontal direction sensitive wavelet function for two-dimensional wavelet transform

is the one-dimensional wavelet function ψ _{j, m} (x) and one-dimensional scaling function

The product of the vertical direction sensitive wavelet function for two-dimensional wavelet transform is a one-dimensional scaling function

and the product of the one-dimensional wavelet function ψ _{j, n} (y); the diagonal direction-sensitive wavelet function for two-dimensional wavelet transformation

is the product of two one-dimensional wavelet functions ψ _{j, m} (x) and ψ _{j, n} (y);

垂直方向小波系数

和对角方向小波系数

重构原始归一化虹膜图像在分辨率2^-1下的细节分量

也即是高频细节分量；具体来说，重构公式为：

其中

表示分辨率2^j下坐标(x，y)的细节分量的值，

表示分辨率2^j下的小波系数，

是i方向敏感的二维小波；∑为累加运算符；Step 3, through the horizontal wavelet coefficients obtained in step 2 at a resolution of 2 ^-1

Vertical wavelet coefficient

and the diagonal wavelet coefficients

Reconstruct the detail component of the original normalized iris image at resolution 2 ^-1

That is, the high-frequency detail component; specifically, the reconstruction formula is:

in

represents the value of the detail component of the coordinate (x, y) at resolution ^2j ,

Indicates the wavelet coefficients at resolution 2 ^j ,

is a two-dimensional wavelet sensitive to the i direction; ∑ is an accumulation operator;

的绝对值

具体来说，如果

如果

$\left|D_{2^{-1}}f(x,y)\right|=-D_{2^{-1}}f(x,y);$ Step 4. Calculate the high-frequency detail components obtained in step 3

absolute value of

Specifically, if

if

$\left|{\mathrm{D.}}_{2^{-1}}f(x,\mathrm{the\; y})\right|=-{\mathrm{D.}}_{2^{-1}}f(x,\mathrm{the\; y});$

的值大于阈值V_o的像素点的数目，将

的值大于阈值V_o的点作为眼睑和睫毛的边界点；V_o为判断像素点是否为眼睑和睫毛边界点的阈值，具体计算公式为：其中M和N分别为原始归一化虹膜图像的宽度和高度，

Step 5, statistics obtained in step 4

The number of pixels whose value is greater than the threshold V _o will be

The point whose value is greater than the threshold V _o is used as the boundary point of the eyelid and eyelashes; V _o is the threshold for judging whether the pixel point is the boundary point of the eyelid and eyelashes, and the specific calculation formula is: Where M and N are the width and height of the original normalized iris image respectively,

值与阈值T_N作比较；如果

则认为该图像为眼睑和睫毛遮挡的图像，如果

则认为该图像为可处理的正常虹膜图像；Step 6, compare the calculated in step 5

The value is compared with the threshold T _N ; if

The image is considered to be an image occluded by eyelids and eyelashes, if

Then it is considered that the image is a normal iris image that can be processed;

垂直方向小波系数

和对角方向小波系数重构原始归一化虹膜图像在分辨率2^-2下的细节分量

Step 7. According to the method in step 3, pass the horizontal direction wavelet coefficients obtained in step 2 under the resolution ^2-2

Vertical wavelet coefficient

and the diagonal wavelet coefficients Reconstruct the detail component of the original normalized iris image at resolution 2 ^-2

的方差；具体的计算公式为：其中，Var为分辨率2^-2下细节分量的方差，M和N分别为原始归一化虹膜图像的宽度和高度，m的计算公式为： $m=\underset{x=1}{\overset{M}{\mathrm{\Σ}}}\underset{y=1}{\overset{N}{\mathrm{\Σ}}}{D}_{2^{-2}f(x,y)}/M\×N;$ Step 8, calculate the obtained in step 7

The variance; the specific calculation formula is: Among them, Var is the variance of the detail component at resolution ^2-2 , M and N are the width and height of the original normalized iris image respectively, and the calculation formula of m is: $m=\underset{x=1}{\overset{m}{\mathrm{\Σ}}}\underset{\mathrm{the\; y}=1}{\overset{N}{\mathrm{\Σ}}}{\mathrm{D.}}_{2^{-2}f(x,\mathrm{the\; y})}/m\×N;$

Step 9. Compare the variance Var obtained in step 8 with the threshold T _v for judging iris clarity, if Var≥T _v , then the iris image clarity meets the system requirements; if Var<T _v , then the iris image is clear The degree does not meet the system requirements.

的虹膜图像匹配，误识率最低时对应的T_N。A kind of iris image quality judging method as above is characterized in that the threshold _TN mentioned in the step 6 is used to judge whether the image is an image that can be processed without eyelids and eyelashes, and the determination of _TN is related to the subsequent matching algorithm, set it to

The iris image matches the corresponding T _N when the false recognition rate is the lowest.

The invention adopts a multi-resolution analysis method, and correctly evaluates the quality of the iris image by analyzing the detail components under different resolutions. The invention firstly locates the original iris image to obtain the normalized iris image. By performing 4-layer wavelet transform on the normalized iris image, counting the number of large-amplitude points of the detail component at a resolution of 2 ^-1 and comparing it with the preset threshold, it is judged whether the iris image is covered by eyelids and eyelashes question. Afterwards, the variance of the detail component under the resolution ^2-2 is counted, and compared with the preset threshold to judge whether the iris image is clear or not. It is a characteristic of the present invention to analyze the iris image quality with the idea of multi-resolution. Compared with the general iris image quality evaluation method, the present invention has strong versatility and stability and is not easily affected by light.

【Description of drawings】

Figure 1 is the original image containing the iris;

Among them, 1 represents the pupil; 2 represents the iris; 3 represents the spot in the pupil; 4 represents the inner edge of the iris; 5 represents the outer edge of the iris.

Figure 2 is a positioning result map and a normalized map;

Among them, (a) is the positioning result image; (b) is the iris normalized image.

Fig. 3 is the original normalized iris image and its high-frequency detail components;

Among them, (a) is the normalized iris image with eyelid occlusion, (b) is the high-frequency detail component of (a), (c) is the normalized iris image with eyelash occlusion, (d) is (c) high-frequency detail components.

【Detailed ways】

In order to describe the content of the present invention conveniently, some terms are defined first.

Definition 1: Iris. The center of the eyeball is the black pupil, and the ring-shaped tissue between the outer edges of the pupil is the iris. It presents interlaced texture features resembling spots, filaments, striations, and crypts. The iris of the same person hardly changes throughout a person's life, and the iris of different people is completely different.

Definition 2: Grayscale image. An image that contains only luminance information in the image without any other color information.

Definition 3: Normalized iris image. After positioning the original iris image, in order to eliminate the head rotation during shooting, the inconsistency of the shooting distance, the pupil scaling and other problems, the image obtained after normalization operation, the normalized iris image has the same size.

Definition 4: Wavelet transform. A localized analysis method of time (space) frequency, which gradually refines the signal (function) on multiple scales through stretching and translation operations, and can focus on any details of the signal.

其中，f(x，y)为原始信号，为尺度函数，

为尺度系数。Definition 5: Scale coefficient. In the wavelet transform, the coefficients obtained after the convolution of the original signal and the scale function are used to reconstruct the approximate components of the signal. For two-dimensional wavelet transform, the specific calculation formula of scale coefficient is:

Among them, f(x, y) is the original signal, is a scaling function,

is the scale factor.

组成的展开函数集合，即集合

其中

j，m∈Z。定义5中的二维小波变换的尺度函数

是两个一维尺度函数

和

的乘积。Definition 6: Scaling function. The scaling function is composed of integer translation and real binary scale, square integrable function

The set of expansion functions composed of

in

j, m ∈ Z. The scaling function of the 2D wavelet transform in Definition 5

are two one-dimensional scaling functions

and

product of .

其中，f(x，y)为原始信号，

为小波函数，

为小波系数。Definition 7: Wavelet coefficients. When wavelet transform is performed, the coefficients obtained after the convolution of the original signal and the wavelet function are used to reconstruct the detail components of the signal. For two-dimensional wavelet transform, the specific calculation formula of wavelet coefficient is:

Among them, f(x, y) is the original signal,

is the wavelet function,

is the wavelet coefficient.

为一维小波函数ψ_j，m(x)和一维尺度函数的乘积；进行二维小波变换的垂直方向敏感的小波函数

为一维尺度函数

和一维小波函数ψ_j，n(y)的乘积；进行二维小波变换的对角方向敏感的小波函数为两个一维小波函数ψ_j，m(x)与ψ_j，n(y)的乘积。Definition 8: Wavelet function. The wavelet function is used to describe the difference across two adjacent scale spaces, and it is a set of expansion functions composed of ψ(x), that is, the set {ψ _{j, k} (x)}. where ψ _{j, m} (x) = 2 ^j/2 ψ(2 ^j x m), j, m ∈ Z. The horizontal direction sensitive wavelet function of the two-dimensional wavelet transform in definition 5

is the one-dimensional wavelet function ψ _{j, m} (x) and one-dimensional scaling function The product of the vertical direction sensitive wavelet function for two-dimensional wavelet transform

is a one-dimensional scaling function

and the product of the one-dimensional wavelet function ψ _{j, n} (y); the diagonal direction-sensitive wavelet function for two-dimensional wavelet transformation is the product of two one-dimensional wavelet functions ψ _{j, m} (x) and ψ _{j, n} (y).

Definition 9: DMeyer wavelet. Discrete form of Meyer wavelet is an effective approximation of Meyer wavelet, which can be regarded as discretized Meyer wavelet with bioorthogonality. It not only maintains the good frequency division characteristics of Meyer wavelet, but also improves the speed of numerical calculation.

Definition 10: Multiresolution analysis. The idea of multi-resolution analysis mainly refers to treating the original image f(x, y) as an approximation at a resolution of 2 ⁰ =1, which is further decomposed into an approximation at a coarse resolution of 2 ^J (J<0) component and a series of asymptotic approximations of detail components at high resolution 2 ^j (j>J).

Definition 11: Detail component. Any image can be decomposed into subject information and detail texture information. According to the idea of multi-resolution analysis, the detail component refers to the detail texture information in different frequency bands.

Definition 12: Matching. Correctly place a specific thing into a category.

Definition 13. False recognition rate. The probability of classifying things of the same class into other classes is often expressed by FMR.

According to the iris image quality judging method of the present invention, it comprises the following steps:

1, a kind of iris image quality judging method is characterized in that it comprises the following steps:

Step 1, through the camera device, the iris in the human eye is image-acquired, and the normalized iris image f(x, y) whose size is M×N is obtained from the original grayscale image containing the iris image; (x, y) Represents the coordinates of the pixel, f(x, y) represents the gray value of the pixel whose coordinates are (x, y);

是分辨率为2^j下的尺度系数，

是分辨率为2^j下小波系数，4层小波变换j的取值范围为{-1，-2，-3，-4}，i＝{H，V，D}，附加了水平、垂直和对角方向的细节；

为尺度函数，

为小波函数，选取的小波为DMeyer小波；Step 2, to the normalized iris image that obtains in step 1, carry out 4 layers of two-dimensional wavelet transform; Specifically, the formula of two-dimensional wavelet transform is: and in,

is the scale coefficient at a resolution of 2 ^j ,

is the wavelet coefficient with a resolution of 2 ^j , the value range of the 4-layer wavelet transform j is {-1, -2, -3, -4}, i={H, V, D}, and the horizontal, vertical and Diagonal details;

is a scaling function,

is a wavelet function, and the selected wavelet is DMeyer wavelet;

垂直方向小波系数

和对角方向小波系数重构原始归一化虹膜图像在分辨率2^-1下的细节分量

也即是高频细节分量；具体来说，重构公式为：

其中

表示分辨率2^j下坐标(x，y)的细节分量的值，

表示分辨率2^j下的小波系数，

是i方向敏感的二维小波；∑为累加运算符；Step 3, through the horizontal wavelet coefficients obtained in step 2 at a resolution of 2 ^-1

Vertical wavelet coefficient

and the diagonal wavelet coefficients Reconstruct the detail component of the original normalized iris image at resolution 2 ^-1

That is, the high-frequency detail component; specifically, the reconstruction formula is:

in

represents the value of the detail component of the coordinate (x, y) at resolution ^2j ,

Indicates the wavelet coefficients at resolution 2 ^j ,

is a two-dimensional wavelet sensitive to the i direction; ∑ is an accumulation operator;

的绝对值

具体来说，如果

如果

$\left|D_{2^{-1}}f(x,y)\right|=-D_{2^{-1}}f(x,y);$ Step 4. Calculate the high-frequency detail components obtained in step 3

absolute value of

Specifically, if

if

$\left|{\mathrm{D.}}_{2^{-1}}f(x,\mathrm{the\; y})\right|=-{\mathrm{D.}}_{2^{-1}}f(x,\mathrm{the\; y});$

的值大于阈值V_o的像素点的数目，将

的值大于阈值V_o的点作为眼睑和睫毛的边界点；V_o为判断像素点是否为眼睑和睫毛边界点的阈值，具体计算公式为：

其中M和N分别为原始归一化虹膜图像的宽度和高度， Step 5, statistics obtained in step 4

The number of pixels whose value is greater than the threshold V _o will be

The point whose value is greater than the threshold V _o is used as the boundary point of the eyelid and eyelashes; V _o is the threshold for judging whether the pixel point is the boundary point of the eyelid and eyelashes, and the specific calculation formula is:

Where M and N are the width and height of the original normalized iris image respectively,

值与阈值T_N作比较；如果

则认为该图像为眼睑和睫毛遮挡的图像，如果

则认为该图像为可处理的正常虹膜图像；Step 6, compare the calculated in step 5

The value is compared with the threshold T _N ; if

The image is considered to be an image occluded by eyelids and eyelashes, if

Then it is considered that the image is a normal iris image that can be processed;

垂直方向小波系数和对角方向小波系数

重构原始归一化虹膜图像在分辨率2^-2下的细节分量

Step 7. According to the method in step 3, pass the horizontal direction wavelet coefficients obtained in step 2 under the resolution ^2-2

Vertical wavelet coefficient and the diagonal wavelet coefficients

Reconstruct the detail component of the original normalized iris image at resolution 2 ^-2

的方差；具体的计算公式为：

其中，Var为分辨率2^-2下细节分量的方差，M和N分别为原始归一化虹膜图像的宽度和高度，m的计算公式为： $m=\underset{x=1}{\overset{M}{\mathrm{\&Sigma;}}}\underset{y=1}{\overset{N}{\mathrm{\&Sigma;}}}{D}_{2^{-2}f(x,y)}/M\&times;N;$ Step 8, calculate the obtained in step 7

The variance; the specific calculation formula is:

Among them, Var is the variance of the detail component at resolution ^2-2 , M and N are the width and height of the original normalized iris image respectively, and the calculation formula of m is: $m=\underset{x=1}{\overset{m}{\mathrm{\&Sigma;}}}\underset{\mathrm{the\; y}=1}{\overset{N}{\mathrm{\&Sigma;}}}{\mathrm{D.}}_{2^{-2}f(x,\mathrm{the\; y})}/m\&times;N;$

Step 9. Compare the variance Var obtained in step 8 with the threshold T _v for judging iris clarity, if Var≥T _v , then the iris image clarity meets the system requirements; if Var<T _v , then the iris image is clear The degree does not meet the system requirements.

Through the above steps, we can judge whether the image meets the requirements of the system by analyzing the normalized iris image extracted from the original image containing iris.

It should be noted:

1. The iris normalization operation in step 1 must be performed after iris positioning.

是两个一维尺度函数

和

的乘积。进行二维小波变换的水平方向敏感的小波函数

为一维小波函数ψ_j，m(x)和一维尺度函数

的乘积；进行二维小波变换的垂直方向敏感的小波函数

为一维尺度函数

和一维小波函数ψ_j，n(y)的乘积；进行二维小波变换的对角方向敏感的小波函数

为两个一维小波函数ψ_j，m(x)与ψ_j，n(y)的乘积。2. Scaling function for two-dimensional wavelet transform in step 2

are two one-dimensional scaling functions

and

product of . Horizontal sensitive wavelet function for two-dimensional wavelet transform

is the one-dimensional wavelet function ψ _{j, m} (x) and one-dimensional scaling function

The product of the vertical direction sensitive wavelet function for two-dimensional wavelet transform

is a one-dimensional scaling function

and the product of the one-dimensional wavelet function ψ _{j, n} (y); the diagonal direction-sensitive wavelet function for two-dimensional wavelet transformation

is the product of two one-dimensional wavelet functions ψ _{j, m} (x) and ψ _{j, n} (y).

3. In steps 3, 4, and 5, select the detail component under resolution ^2-1 to analyze the occlusion of eyelids and eyelashes, because there are obvious grayscale changes on the grayscale image at the edge of the eyelids and eyelashes, according to the multi-resolution Based on the idea of rate analysis, the points at these positions correspond to the larger magnitudes of the high-frequency detail components.

大于V_o的点就是眼睑和睫毛的边缘点，是因为在眼睑和睫毛的边缘位置处，存在灰度值的明显跳变，对应着高频细节分量上幅值较大的点；而虹膜的纹理边缘位置处，灰度变换较为缓慢，对应着高频细节分量上幅值较小的点。当高频细节分量幅值

大于某一值时，该点就是眼睑和睫毛的边缘点。4. The threshold V _o in step 5 is used to judge whether the point on the high-frequency detail component is the edge point of the eyelid and eyelashes. The magnitude of the high-frequency detail component

Points greater than V _o are the edge points of the eyelids and eyelashes, because at the edge of the eyelids and eyelashes, there is an obvious jump in the gray value, corresponding to a point with a larger amplitude on the high-frequency detail component; while the iris At the edge of the texture, the grayscale transformation is relatively slow, corresponding to the point with a smaller amplitude on the high-frequency detail component. When the amplitude of high-frequency detail components

Above a certain value, the point is the edge point of the eyelids and eyelashes.

的虹膜图像匹配，误识率最低时对应的T_N。5. The threshold _TN mentioned in step 6 is used to judge whether the image is an image without eyelids and eyelashes that can be processed. The determination of T _N is related to the subsequent matching algorithm, we set it as

The iris image matches the corresponding T _N when the false recognition rate is the lowest.

进行虹膜清晰度的分析，是因为虹膜纹理在灰度图像上变化较为缓慢，根据多分辨率分析的思想，这一变化多体现在中频细节分量

虹膜图像越清晰，中频细节分量的方差Var越大；虹膜图像越模糊，中频细节分量的方差Var越小。6. In steps 7 and 8, select the detail component of the normalized iris image at resolution ^2-2

The analysis of iris clarity is because the iris texture changes slowly on the grayscale image. According to the idea of multi-resolution analysis, this change is mostly reflected in the intermediate frequency detail component

The clearer the iris image, the larger the variance Var of the intermediate frequency detail component; the blurrier the iris image, the smaller the variance Var of the intermediate frequency detail component.

7. The threshold T _v in step 9 is used to judge whether the iris image is clear, and the determination of T _v is related to the subsequent matching algorithm.

The present invention adopts the method of multi-resolution analysis, first by extracting the iris in the original grayscale image and normalizing it; then performing multi-resolution decomposition on the normalized iris image to obtain resolution ^2-1 and resolution 2 respectively The detail component under ^-2 ; finally judge whether the image has the problem of eyelid and eyelash occlusion according to the number of larger amplitude points of the detail component under resolution 2 ^-1 , and judge the iris image according to the variance of the detail component under resolution 2 ^-2 Is it clear. By adopting the method based on multi-resolution analysis proposed by the present invention, the iris image quality can be effectively evaluated, and the problem that traditional algorithms are sensitive to light is avoided.

Adopt the method of the present invention, at first use C language to write the iris image quality evaluation program; Then adopt CMOS or CCD camera to take the original image of iris automatically; Then the iris original image that is taken is input to the iris image on the PC platform as source data It is processed in the quality evaluation program; after iris positioning, normalization and image quality evaluation, it gives the judgment whether the image meets the system requirements. Using 2,400 gray-scale iris images taken by different people, including different lighting conditions and different shooting postures, as source data, the results of the program judgment are compared with the results of subjective judgment. The error probability is 1.2%. Processing time <150ms.

To sum up, the method of the present invention makes full use of the texture information of the iris, combined with the method of multi-resolution analysis, so as to realize the fast and accurate judgment of the quality of the iris image.

Claims (2)

1. a method for judging iris image quality, is characterized in that it comprises the following steps: Step 1, through the camera device, the iris in the human eye is image-acquired, and the normalized iris image f(x, y) whose size is M×N is obtained from the original grayscale image containing the iris image; (x, y) Represents the coordinates of the pixel, f(x, y) represents the gray value of the pixel whose coordinates are (x, y); Step 2, to the normalized iris image that obtains in step 1, carry out 4 layers of two-dimensional wavelet transform; Specifically, the formula of two-dimensional wavelet transform is: and

in,

is the scale coefficient at a resolution of 2 ^j ,

is the wavelet coefficient with a resolution of 2 ^j , the value range of the 4-layer wavelet transform j is {-1, -2, -3, -4}, i={H, V, D}, and the horizontal, vertical and Diagonal details;

is a scaling function,

is a wavelet function, and the selected wavelet is DMeyer wavelet; the scaling function for two-dimensional wavelet transformation are two one-dimensional scaling functions

and The product of the horizontal direction sensitive wavelet function for two-dimensional wavelet transform

is the one-dimensional wavelet function ψ _{j, m} (x) and one-dimensional scaling function The product of the vertical direction sensitive wavelet function for two-dimensional wavelet transform

is a one-dimensional scaling function

and the product of the one-dimensional wavelet function ψ _{j, n} (y); the diagonal direction-sensitive wavelet function for two-dimensional wavelet transformation

is the product of two one-dimensional wavelet functions ψ _{j, m} (x) and ψ _{j, n} (y); Step 3, through the horizontal wavelet coefficients obtained in step 2 at a resolution of 2 ^-1 Vertical wavelet coefficient and the diagonal wavelet coefficients

Reconstruct the detail component of the original normalized iris image at resolution 2 ^-1 That is, the high-frequency detail component; specifically, the reconstruction formula is:

in

represents the value of the detail component of the coordinate (x, y) at resolution ^2j ,

Indicates the wavelet coefficients at resolution 2 ^j ,

is a two-dimensional wavelet sensitive to the i direction; ∑ is an accumulation operator; Step 4. Calculate the high-frequency detail components obtained in step 3

absolute value of

Specifically, if

if

$\left|{\mathrm{D.}}_{2^{-1}}f(x,\mathrm{the\; y})\right|=-{\mathrm{D.}}_{2^{-1}}f(x,\mathrm{the\; y});$ Step 5, statistics obtained in step 4 The number of pixels whose value is greater than the threshold V _o will be

The point whose value is greater than the threshold V _o is used as the boundary point of the eyelid and eyelashes; V _o is the threshold for judging whether the pixel point is the boundary point of the eyelid and eyelashes, and the specific calculation formula is:

Where M and N are the width and height of the original normalized iris image respectively, Step 6, compare the calculated in step 5

The value is compared with the threshold T _N ; if

The image is considered to be an image occluded by eyelids and eyelashes, if

Then it is considered that the image is a normal iris image that can be processed; Step 7. According to the method in step 3, pass the horizontal direction wavelet coefficients obtained in step 2 under the resolution ^2-2 Vertical wavelet coefficient

and the diagonal wavelet coefficients

Reconstruct the detail component of the original normalized iris image at resolution 2 ^-2

Step 8, calculate the obtained in step 7

The variance; the specific calculation formula is:

Among them, Var is the variance of the detail component at resolution ^2-2 , M and N are the width and height of the original normalized iris image respectively, and the calculation formula of m is: $m=\underset{x=1}{\overset{m}{\mathrm{\&Sigma;}}}\underset{\mathrm{the\; y}=1}{\overset{N}{\mathrm{\&Sigma;}}}{\mathrm{D.}}_{2^{-2}f(x,\mathrm{the\; y})}/m\&times;N;$ Step 9: Compare the variance Var obtained in step 8 with the threshold T _v for judging iris clarity, if Var≥T _v , then the iris image clarity meets the system requirements; if Var<T _v , then the iris image is clear The degree does not meet the system requirements. 2. a kind of iris image quality judging method according to claim 1, it is characterized in that the threshold _TN mentioned in step 6 is to be used for judging whether this image is the image that can be processed without eyelids and eyelashes blocking, _TN The determination of is related to the subsequent matching algorithm, which is set as

The iris image matches the corresponding T _N when the false recognition rate is the lowest.

Images