Iris segmentation method constrained by three-dimensional eyeball model parameters

Abstract

The present invention relates to an iris segmentation method constrained by three-dimensional eyeball model parameters, comprising: acquiring an eye image, establishing a three-dimensional space coordinate system on the basis of an eyeball, wherein the rightward direction of an X-axis is a positive direction, the upward direction of a Y-axis is a positive direction, and the outward direction of a Z-axis is a positive direction, and acquiring eyeball center and iris radius parameters; establishing a three-dimensional space circle parameter equation; deriving an iris ellipse function on the basis of an iris three-dimensional space circle vector equation; and segmenting the iris on the basis of the iris ellipse function. In the present invention, the position and shape of the iris are constrained by three-dimensional eyeball model parameters, thereby improving the accuracy and robustness of iris segmentation; an ellipse parameter equation is used instead of a circle parameter equation, making it more in line with the actual iris shape; the absolute value of the cross-boundary difference between corresponding points is used instead of the circle mean, which can better reflect changes in the iris boundary; a coarse positioning and fine positioning two-step method is adopted to reduce the search space and computation load, thereby improving the segmentation efficiency and real-time performance.

Description

A method for iris segmentation constrained by three-dimensional eye model parameters Technical Field

[0001] This invention relates to the field of image detection technology, and more specifically, to an iris segmentation method constrained by three-dimensional eye model parameters. Background Technology

[0002] Iris recognition is a biometric identification technology that uses the structural features of the iris of the human eye for identification. Iris recognition boasts advantages such as high accuracy, high stability, and high security, and is considered one of the most reliable biometric identification technologies currently available.

[0003] One of the key steps in iris recognition is iris segmentation, which involves extracting the iris region from an eye image for feature extraction and matching. The accuracy of iris segmentation directly impacts the performance of iris recognition.

[0004] Currently, many iris recognition products internationally use the iris recognition algorithm proposed by Dr. Daugman of the University of Cambridge, UK.

[0005] This algorithm uses an integro-differential operator to detect the inner and outer circular boundaries of the iris, namely:

[0006] Among them, G σ is a Gaussian smoothing function, I(x,y) is the eye image, r,x0,y0 are the radius and center coordinates of the circle, and ds is a infinitesimal element on the circumference.

[0007] The Daugman algorithm is very successful, but it has limitations, mainly in the following aspects:

[0008] The algorithm assumes that the iris is a perfect circular structure, but in reality, due to factors such as the three-dimensional shape of the eyeball and the rotation of the eye, the iris appears as an ellipse in the image.

[0009] This algorithm requires searching and calculating for every possible center and radius, which involves a large amount of computation and has poor real-time performance, making it unsuitable for use in portable devices or real-time applications.

[0010] Summary of the Invention

[0011] To address the shortcomings of existing technologies, this invention proposes an iris segmentation method constrained by three-dimensional eyeball model parameters. This method can segment the iris more accurately, faster, and more robustly, while having lower computational complexity.

[0012] Summary of the Invention

[0013] To address the shortcomings of existing technologies, a clear and concise three-dimensional eye detection technology solution is proposed, including how to calculate the optical axis and the rotation matrix, how to unfold the ellipse into a rectangle, how to match iris texture features, and finally obtain the eye movement rotation matrix.

[0014] The technical solution of the present invention is as follows:

[0015] A method for iris segmentation constrained by three-dimensional eye model parameters, comprising the following steps:

[0016] St10. Acquire eye images, establish a three-dimensional spatial coordinate system based on the eyeball, with the X-axis pointing to the right as positive, the Y-axis pointing upwards as positive, and the Z-axis pointing outwards as positive, and obtain the parameters of the eyeball center and iris radius;

[0017] St20, Establish the three-dimensional spatial parametric equations of a circle;

[0018] St30. Based on the three-dimensional circular vector equation of the iris, derive the iris elliptic function;

[0019] St40, Segment the iris based on the iris elliptic function.

[0020] In summary, the above technical solution has the following beneficial effects: 1. The present invention restricts the position and shape of the iris based on the three-dimensional model parameters of the eyeball, thereby improving the accuracy and robustness of iris segmentation;

[0021] 2. This invention uses an elliptic parametric equation instead of a circular parametric equation, which is more consistent with the actual shape of the iris;

[0022] 3. This invention uses the absolute value of the cross-boundary difference between corresponding points instead of the mean value of the circle, which can better reflect the changes in the iris boundary;

[0023] 4. This invention employs a two-step method of coarse and fine localization, which reduces the search space and computational load, and improves segmentation efficiency and real-time performance;

[0024] 5. This invention employs two methods: the iris center perturbation method and the eyeball center, iris radius, and iris center perturbation method, adapting to different eye image acquisition conditions. Attached Figure Description

[0025] Figure 1 is a flowchart of an iris segmentation method constrained by three-dimensional eyeball model parameters;

[0026] Figure 2 is a schematic diagram of the original image of the eyeball;

[0027] Figure 3 is a schematic diagram of the initial segmentation of the inner and outer boundaries of the iris after cropping and scaling the iris image to 80*60.

[0028] Figure 4 is a first schematic diagram of gradually zooming in on the image to accurately locate the center of the iris and fit the inner and outer boundaries of the iris with an ellipse.

[0029] Figure 5 is a second schematic diagram showing the initial segmentation of the inner and outer boundaries of the iris after cropping and scaling the iris image to 80*60.

[0030] Figure 6 is a second schematic diagram showing how the image is gradually magnified to accurately locate the center of the iris and fit the inner and outer boundaries of the iris with an ellipse.

[0031] Figure 7 is a schematic diagram of the output iris region in the original image;

[0032] Figure 8 is a schematic diagram of the iris inner boundary segmentation results without eyeball parameter constraints. Detailed Implementation

[0033] The present invention will be further described in detail below with reference to the accompanying drawings and embodiments. Identical components are denoted by the same reference numerals. It should be noted that the terms "front," "rear," "left," "right," "upper," and "lower" used in the following description refer to directions in the accompanying drawings, and the terms "bottom surface," "top surface," "inner," and "outer" refer to directions toward or away from the geometric center of a specific component, respectively.

[0034] As shown in Figure 1, an iris segmentation method constrained by three-dimensional eyeball model parameters includes the following process:

[0035] St10. Acquire eye images, establish a three-dimensional spatial coordinate system based on the eyeball, with the X-axis pointing to the right as positive, the Y-axis pointing upwards as positive, and the Z-axis pointing outwards as positive, and obtain the parameters of the eyeball center and iris radius.

[0036] If the eye center and iris radius are known, inputting these parameters will allow for subsequent processing. If the eye center and iris radius are unknown, firstly, based on iris segmentation algorithms, including thresholding methods, edge detection algorithms, and deep learning methods, the iris region in the eye image is detected to obtain the iris center and iris region. The eye center and iris radius data can be obtained using the following three methods.

[0037] St11, Method 1: Based on the iris segmentation algorithm, the iris region in the eye image is detected, the iris center and the iris region are obtained, the coordinates of the iris center at the position where the difference between the minor axis and the major axis of the iris ellipse in the eye movement video is the smallest or the smallest is the coordinate of the eyeball center, and the major axis of the ellipse is taken as the iris radius.

[0038] St12, Method 2: Based on the iris segmentation algorithm, the iris region in the eye image is detected, the iris center and iris region are obtained, the optical axis is calculated according to the iris ellipse parameters, the point with the smallest distance from each optical axis is taken as the eyeball center, and then the iris radius is calculated.

[0039] St13, Method 3: Establish functions with independent variables as the center of the eyeball, the radius of the iris, and the distance from the center of the iris to the center of the eyeball. Search for the center point of the iris using the three-dimensional iris circle formula. The return value is the maximum gradient value of the iris ellipse boundary. Solve for the center of the eyeball, the radius of the iris, and the distance from the center of the iris to the center of the eyeball by minimizing the objective function.

[0040] St20, Establish the three-dimensional spatial parametric equations of the circle.

[0041] St21, The method for establishing the parametric equations of a three-dimensional circle includes the following steps:

[0042] Let the diameter of the three-dimensional circle be r, and let r = a. Let the initial unit normal vector of the circle's plane be V0 = [0,0,1]. After the circle rotates θ around the X-axis for the first time, its projection onto the XY plane is an ellipse. Let its major axis be a and its minor axis be b. The initial unit normal vector of the rotated three-dimensional circular plane is V1 = [0, -sinθ, cosθ].

[0043] A three-dimensional circle is rotated a second time around the Z-axis by an angle δ. Its projection onto the XY plane is an ellipse with major axis a, minor axis b, and rotation angle δ. The normal vector of the rotated three-dimensional circle is o. Then o = [sinθsinδ, -sinθcosδ, cosθ]

[0044] The rotated three-dimensional circle is expressed as: circle=(vcosβ+usinβ)r

[0045] Where v=[cosδ,sinδ,0], u=o*v=[-cosθsinδ,cosθcosδ,sinθ];

[0046] Therefore, a circle in three-dimensional space can be expressed as follows: Circles=((cosδcosβ-cosθsinδsinβ)i+(sinδcosβ+cosθcosδsinβ)j +(sinθsinβ)k)r

[0047] The parametric equation of a circle in three-dimensional space with its center at the origin is expressed as: X = (cosδcosβ - cosθsinδsinβ)r Y = (sinδcosβ + cosθcosδsinβ)r Z = (sinθsinβ)r

[0048] Where a is the major axis of the elliptical shape projected onto the XY plane by the three-dimensional spatial circle, b is the minor axis of the elliptical shape projected onto the XY plane by the three-dimensional spatial circle, θ is the angle of the first rotation of the three-dimensional spatial circle around the X-axis, and δ is the angle of the second rotation of the three-dimensional spatial circle around the Z-axis.

[0049] St30. Based on the circular vector equation of the iris in three-dimensional space, derive the iris elliptic function.

[0050] St31. Based on the three-dimensional circular vector equation of the iris, derive the iris elliptic function; including the following steps:

[0051] Input one or more eye images I(x,y).

[0052] St32, set the center of the eyeball C e = (x0, y0, 0), iris radius is r, iris center C i = (x1, y1, z1), where d is the distance from the center of the iris to the center of the eyeball.

[0053] After the eyeball rotates, the projection of the iris circle is an ellipse. The major axis of the ellipse is a, the minor axis is b, the rotation angle is δ, the unit optical axis is o, and the angle between the optical axis and the Z-axis is θ.

[0054] Where θ is the angle of the first rotation of the three-dimensional circle around the X-axis, and δ is the angle of the second rotation of the three-dimensional circle around the Z-axis.

[0055] St33, Set the parameters C of the 3D eyeball model e ,r,D,C i The perturbation range, for each possible C e ,r,D,C i Calculate the three-dimensional circular vector of the iris. Equation: V iris =C i +(vcosβ+usinβ)r

[0056] Where v = [cosδ, sinδ, 0] is the iris plane unit vector, that is, the unit vector rotated from the initial position X-axis unit vector to the target position, and u = o * v = [-cosθsinδ, cosθcosδ, sinθ] is the iris plane unit vector perpendicular to v, that is, the unit vector rotated from the initial position Y-axis unit vector to the target position.

[0057] St34, when V iris = [X, Y, Z], then X = x1 + (cos δ cos β - cos θ sin δ sin β) r

[0058] St35. Based on the three-dimensional circular vector equation of the iris, the iris elliptic function G(r,β) is derived as: G(r,β)=I(x1+(cos(δ)cos(β)-cos(θ)sin(δ)sin(β))r,y1+(sinδcosβ+cosθcosδsinβ)r)

[0059] For each ellipse equation, calculate the integral at its corresponding boundary point. Select the ellipse equation with the largest integral value as the inner and outer boundaries of the iris, and output the corresponding center of the eyeball, center of the iris, iris radius, and distance from the center of the iris to the center of the eyeball.

[0060] Where Δβ = 2π / n, Δr = 1, and t is the threshold value for the boundary point. A boundary point is a point whose attribute value differs from the specified threshold t by more than 1. Image point attribute values ​​include, but are not limited to, grayscale values, gradient values, and color values. The t value is set according to the attribute values ​​and specific circumstances, and is defaulted to 4% of the average density of image I. Using gradient values ​​as attribute values ​​can better capture the boundary.

[0061] The range of values ​​for 0 (n) can be set to limit the range of values ​​for the iris boundary, thereby reducing the impact of eyelid occlusion.

[0062] Furthermore, the eye image I(x,y) undergoes preprocessing operations including grayscale conversion, histogram equalization, and filtering.

[0063] The default distance from the center of the iris to the center of the eyeball is set to The reason is that if d is unknown, the eyeball radius R is twice the iris radius r, and r 2 +d 2 =R 2 It can be deduced that... Alternatively, the value of d can be set based on the input eyeball model parameters.

[0064] St40, Segment the iris based on the iris elliptic function.

[0065] Iris segmentation based on the iris elliptic function includes the following process:

[0066] St41. For each ellipse equation, calculate the integral and differential value of the absolute value of the cross-boundary difference at the corresponding point:

[0067] Among them, G σ is a Gaussian smoothing function, I(x,y) is the eye image, r,d are the radius of the iris circle and the distance from the center of the iris to the center of the eyeball, x0,y0 are the coordinates of the iris center, and ds is a infinitesimal element on the circumference;

[0068] St42. Select the ellipse equation with the largest integral and differential value as the inner and outer boundaries of the iris, and output the corresponding iris center coordinates and iris region.

[0069] St42, the iris image is cropped and scaled to 80*60, and the inner and outer boundaries of the iris are initially segmented.

[0070] St44. Gradually zoom in on the image to accurately locate the center of the iris and fit the inner and outer boundaries of the iris with an ellipse.

[0071] St45, output the iris region from the original image to obtain the iris inner boundary segmentation result that is not constrained by eye parameters.

[0072] The beneficial effects of this invention are as follows:

[0073] This invention constrains the position and shape of the iris based on the three-dimensional model parameters of the eyeball, thereby improving the accuracy and robustness of iris segmentation;

[0074] This invention uses an elliptic parametric equation instead of a circular parametric equation, which is more consistent with the actual shape of the iris.

[0075] This invention uses the absolute value of the cross-boundary difference between corresponding points instead of the mean value of the circle, which can better reflect the changes in the iris boundary;

[0076] This invention employs a two-step method of coarse and fine localization, which reduces the search space and computational load, and improves segmentation efficiency and real-time performance.

[0077] This invention employs two methods: the iris center perturbation method and the eyeball center, iris radius, and iris center perturbation method, adapting to different eye image acquisition conditions.

[0078] Example 1

[0079] This embodiment uses a known eyeball center and iris radius for iris segmentation. The specific steps are as follows:

[0080] 1. Input an eye image I(x,y);

[0081] 2. If the distance from the center of the iris to the center of the eyeball is unknown, based on the known center of the eyeball C... e Given (x0, y0, 0) and iris radius r, calculate the distance from the center of the iris to the center of the eyeball. Set the iris center perturbation ranges [-Δx, Δx] and [-Δy, Δy], where Δx and Δy can be adjusted according to factors such as image resolution and acquisition angle, for example, the central 50% region of the iris image.

[0082] 3. Within the range of iris center perturbation, for each possible iris center C i= (x1, y1, z1), calculate its circular parametric equation in three-dimensional space, and derive the iris elliptic function.

[0083] 4. For each ellipse equation, calculate the integral of its corresponding boundary point, select the ellipse equation with the largest integral value as the inner and outer boundaries of the iris, and output the corresponding iris center and iris region.

[0084] Example 2

[0085] This embodiment employs an iris segmentation system constrained by three-dimensional eyeball model parameters, the specific structure of which is as follows:

[0086] Image acquisition module: Used to acquire eye images and preprocess them. In this embodiment, a high-definition camera is used as the image acquisition device and connected to the computer via a USB interface. The image acquisition module also includes software components for performing operations such as grayscale conversion, histogram equalization, and filtering on the acquired eye images.

[0087] Eye center localization module: Based on coarse localization through iris segmentation, iris morphology analysis is used to determine the coordinates of the eye center and the iris radius. In this embodiment, a deep learning-based method is used to detect the iris region in the eye image and calculate the center coordinates of the position where the difference between the minor and major axes of the ellipse is small as the eye center coordinates.

[0088] Iris center initialization module: used to initialize the iris based on the coordinates of the eyeball center (Center). eye = [x0, y0, 0], initialize the iris center coordinates Center eye = [x0, y0, 0]. In this embodiment, assuming the iris radius r is 50 pixels, the eyeball radius R is 100 pixels, and the distance d from the iris center to the eyeball center is 86.6 pixels (i.e., √3 multiplied by r), then the iris center coordinates Center are initialized. iris = [x1, y1, z1], where

[0089] Iris boundary detection module: used to detect the boundary of the iris based on the parameters of the 3D eye model Center. eye Center iris Using d and r, the three-dimensional spatial circular parametric equation of the iris is derived, and the projected ellipse of the iris in the XY plane is calculated. In this embodiment, the method and formula described in Embodiment 1 above are used to calculate the parameters and point coordinates of the projected ellipse, and the ellipse is drawn on the eye image.

[0090] Evaluation function calculation module: This module defines an evaluation function to measure the degree of matching between the projected ellipse and the iris region in the eye image. In this embodiment, the method and formula described in Embodiment 1 above are used to calculate the value of the evaluation function E.

[0091] Iris Center Optimization Module: Used to perform a two-step method of coarse and fine positioning to optimize the iris center coordinates (Center). iris The projection ellipse is optimized by maximizing the evaluation function E within the range of [x1, y1, z1], thus finding the optimal projection ellipse. In this embodiment, the method and formula described in Embodiment 1 above are used to randomly perturb x1 and y1 within a certain range, and the iris center coordinates after the perturbation with the largest E value are selected as the optimal iris center coordinates.

[0092] Iris region determination and post-processing module: This module determines the iris region based on the optimal projection ellipse and performs post-processing on it. In this embodiment, the method and formula described in Embodiment 1 above are used to remove occlusions from the eye image and convert the iris region into polar coordinates.

[0093] Example 3

[0094] As shown in Figures 2-4, this embodiment utilizes the pyramid algorithm to improve the speed of the algorithm when the center of the eyeball and the radius of the iris are unknown. The specific steps are as follows:

[0095] 1. Given an eye image I(x,y), generate a series of images I obtained by progressively downsampling from the image at maximum resolution. p (x p ,y p Each layer's image is the original image after Gaussian smoothing. For example, the minimum resolution is 30 pixels high and 40 pixels wide, or 60 pixels high and 80 pixels wide, which requires less computation.

[0096] 2. Starting with the image with the lowest resolution, set the center of the eyeball C. e = (x0, y0, 0), iris radius r and iris center C i = (x1, y1, z1) perturbation range, adjusted according to image resolution; within the parameter perturbation range, for each possible eye center C e iris radius r and iris center C i The parametric equation of the circle in three-dimensional space is calculated and projected onto the two-dimensional image plane to obtain the ellipse equation. For each ellipse equation, the integral and differential values ​​of the absolute values ​​of the cross-boundary differences of the corresponding points are calculated, and the ellipse equation with the largest integral and differential value is selected as the inner and outer boundaries of the iris.

[0097] 3. The acquired eye center, iris radius, and iris center parameters are passed to the next image, and the process is repeated until the image with the highest resolution is output, at which point the iris region is output. This reduces the perturbation range of the eye center, iris center, and iris radius, improving the processing speed.

[0098] Example 4

[0099] As shown in Figures 5-8, this embodiment uses the case where the center of the eyeball and the radius of the iris are known, and utilizes the pyramid algorithm to improve the speed of the algorithm. The specific steps are as follows:

[0100] 1. If the distance from the center of the iris to the center of the eyeball is unknown, based on the known center of the eyeball C... e Given (x0, y0, 0) and iris radius r, calculate the distance from the center of the iris to the center of the eyeball.

[0101] 2. Given an eye image I(x,y), generate a series of images I obtained by progressively downsampling from the image at maximum resolution. p (x p ,y p Each layer's image is the original image after Gaussian smoothing.

[0102] 3. Starting with the lowest resolution image, set the iris center C. i = (x1, y1, z1) perturbation range, adjusted according to image resolution; within the parameter perturbation range, for each possible iris center C i The parametric equation of the circle in three-dimensional space is calculated and projected onto the two-dimensional image plane to obtain the ellipse equation. For each ellipse equation, the integral of its corresponding boundary point is calculated, and the ellipse equation with the largest integral is selected as the inner and outer boundaries of the iris.

[0103] 3. Pass the acquired iris center parameters to the next image, and repeat the iris center C process. i The perturbation process continues until the image reaches maximum resolution, thus reducing the perturbation range at the center of the iris and improving the overall speed.

[0104] Example 5:

[0105] This embodiment improves step 5 of the technical solution. Step 5 is changed to calculate the integral value of the absolute value of the cross-boundary difference of the corresponding point for each ellipse equation.

[0106] The corresponding point is the point corresponding to a radius of r at a specified angle α in the iris ellipse equation. The corresponding point boundary difference is the difference in pixel value between a point with a radius of r-Δr at a specified angle α in the iris ellipse equation and a point with a radius of r+Δr.

[0107] This algorithm can be expressed as:

[0108] The above are merely preferred embodiments of the present invention. The scope of protection of the present invention is not limited to the above embodiments. All technical solutions falling within the scope of the present invention's concept are within the scope of protection of the present invention. It should be noted that for those skilled in the art, any improvements and modifications made without departing from the principle of the present invention should also be considered within the scope of protection of the present invention.

Claims

1. A method for iris segmentation constrained by three-dimensional eyeball model parameters, characterized in that, Acquire eye images, establish a three-dimensional spatial coordinate system based on the eyeball, with the X-axis pointing to the right as positive, the Y-axis pointing upwards as positive, and the Z-axis pointing outwards as positive, and obtain the parameters of the eyeball center and iris radius; Establish the parametric equations of a circle in three-dimensional space; Based on the three-dimensional circular vector equation of the iris, the iris elliptic function is derived; Iris segmentation is performed based on the iris elliptic function.

2. The iris segmentation method constrained by the parameters of a three-dimensional eyeball model according to claim 1, characterized in that, The data for the center of the eyeball and the radius of the iris were obtained through the following process: Based on the iris segmentation algorithm, the iris region in the eye image is detected, the iris center and the iris region are obtained, and the coordinates of the iris center at the position where the difference between the minor axis and the major axis of the iris ellipse in the eye movement video is the smallest or the smallest is used as the coordinates of the eyeball center. The major axis of the ellipse is taken as the iris radius.

3. The iris segmentation method constrained by the parameters of a three-dimensional eyeball model according to claim 1, characterized in that, The data for the center of the eyeball and the radius of the iris were obtained through the following process: Based on the iris segmentation algorithm, the iris region in the eye image is detected, the iris center and iris region are obtained, the optical axis is calculated according to the iris ellipse parameters, the point with the smallest distance from each optical axis is taken as the eyeball center, and then the iris radius is calculated.

4. The iris segmentation method constrained by the parameters of a three-dimensional eyeball model according to claim 1, characterized in that, The data for the center of the eyeball and the radius of the iris were obtained through the following process: Establish functions with independent variables of eye center, iris radius, and distance from iris center to eye center. Search for the iris center point using the three-dimensional iris circle formula. The return value is the maximum gradient value of the iris ellipse boundary. Solve for eye center, iris radius, and distance from iris center to eye center by minimizing the objective function.

5. The iris segmentation method constrained by the parameters of a three-dimensional eyeball model according to claim 1, characterized in that, The method for establishing the parametric equations of a circle in three-dimensional space includes the following steps: Let the diameter of the three-dimensional circle be r, and let r = a. Let the initial unit normal vector of the circle's plane be V0 = [0,0,1]. After the circle rotates θ around the X-axis for the first time, its projection onto the XY plane is an ellipse. Let its major axis be a and its minor axis be b. The initial unit normal vector of the rotated three-dimensional circular plane is V1 = [0, -sinθ, cosθ]. A three-dimensional circle is rotated a second time around the Z-axis by an angle δ. Its projection onto the XY plane is an ellipse with major axis a, minor axis b, and rotation angle δ. The normal vector of the rotated three-dimensional circle is o. o=[sinθsinδ,-sinθcosδ,cosθ] The rotated three-dimensional circle is expressed as: circle=(vcosβ+usinβ)r Where v=[cosδ,sinδ,0], u=o*v=[-cosθsinδ,cosθcosδ,sinθ]; Therefore, a circle in three-dimensional space can be expressed as follows: Circles=((cosδcosβ-cosθsinδsinβ)i+(sinδcosβ+cosθcosδsinβ)j +(sinθsinβ)k)r The parametric equation of a three-dimensional circle centered at the origin is expressed as: X=(cosδcosβ-cosθsinδsinβ)r Y=(sinδcosβ+cosθcosδsinβ)r Z=(sinθsinβ)r Where a is the major axis of the elliptical shape projected onto the XY plane by the three-dimensional spatial circle, b is the minor axis of the elliptical shape projected onto the XY plane by the three-dimensional spatial circle, θ is the angle of the first rotation of the three-dimensional spatial circle around the X-axis, and δ is the angle of the second rotation of the three-dimensional spatial circle around the Z-axis.

6. The iris segmentation method constrained by the parameters of a three-dimensional eyeball model according to claim 5, characterized in that, Based on the three-dimensional circular vector equation of the iris, the elliptic function of the iris is derived through the following process: Input one or more eye images I(x,y); Let the center of the eyeball be C. e = (x0, y0, 0), iris radius is r, iris center C i = (x1, y1, z1), where d is the distance from the center of the iris to the center of the eyeball; After the eyeball rotates, the projection of the iris circle is an ellipse. The major axis of the ellipse is a, the minor axis is b, the rotation angle is δ, the unit optical axis is o, and the angle between the optical axis and the Z-axis is θ. Where θ is the angle of the first rotation of the three-dimensional circle around the X-axis, and δ is the angle of the second rotation of the three-dimensional circle around the Z-axis; Setting the parameters C of the 3D eye model e ,r,d,C i The perturbation range, for each possible C e ,r,d,C i Calculate the three-dimensional circular vector of the iris. equation: V iris =C i +(vcosβ+usinβ)r Where v = [cosδ, sinδ, 0] is the unit vector in the iris plane, and u = o * v = [-cosθsinδ, cosθcosδ, sinθ] is the unit vector in the iris plane perpendicular to v; When V iris = [X,Y,Z], then X=x1+(cosδcosβ-cosθsinδsinβ)r Y=y1+(sinδcosβ+cosθcosδsinβ)r Z=(sinδsinβ)r Based on the three-dimensional circular vector equation of the iris, the iris elliptic function G(r,β) is derived as follows: G(r,β)=I(x1+(cos(δ)cos(β)-cos(θ)sin(δ)sin(β))r,y1+(sinδcosβ+cosθcosδsinβ)r) For each ellipse equation, calculate the integral at its corresponding boundary point. Select the ellipse equation with the largest integral value as the inner and outer boundaries of the iris, and output the corresponding center of the eyeball, center of the iris, iris radius, and distance from the center of the iris to the center of the eyeball. Where Δβ=2π / n, Δr=1, and t is the threshold value for the cross-boundary point.

7. The iris segmentation method constrained by the parameters of a three-dimensional eyeball model according to claim 6, characterized in that, Preprocessing operations such as grayscale conversion, histogram equalization, and filtering are performed on the eye image I(x,y).

8. The iris segmentation method constrained by the parameters of a three-dimensional eyeball model according to claim 6, characterized in that, The default distance from the center of the iris to the center of the eyeball is set to 9. The iris segmentation method constrained by the parameters of a three-dimensional eyeball model according to claim 6, characterized in that, Iris segmentation based on the iris elliptic function includes the following process: For each ellipse equation, calculate the integral and differential value of the absolute value of the cross-boundary difference at the corresponding point: Among them, G σ is a Gaussian smoothing function, I(x,y) is the eye image, r,d are the radius of the iris circle and the distance from the center of the iris to the center of the eyeball, x0,y0 are the coordinates of the iris center, and ds is a infinitesimal element on the circumference; Select the ellipse equation with the largest integral and differential value as the inner and outer boundaries of the iris, and output the corresponding iris center coordinates and iris region. The iris image is cropped and scaled to 80*60 to initially segment the inner and outer boundaries of the iris. Gradually zoom in on the image to accurately locate the center of the iris and fit the inner and outer boundaries of the iris with an ellipse; The iris region is output from the original image to obtain the intrairis boundary segmentation result that is not constrained by eye parameters.

Images