COMPUTER-IMPLEMENTED METHOD FOR DETERMINING CENTERING PARAMETERS

Description

[0001] The invention relates to a computer-implemented method for determining centering parameters for adapting spectacle lenses to a predetermined spectacle frame and to the head of a test subject according to the preamble of claim 1.

[0002] Centration parameters are used to correctly position or center spectacle lenses within a frame, ensuring they are in the correct position relative to the wearer's eyes. These parameters include anatomical parameters of the individual, such as pupillary distance; frame-specific parameters, such as lens width or height; and combinations of anatomical and frame-specific parameters, such as vertex distance and optical center height. DIN EN ISO 13666, dated October 2013, provides an overview of common centration parameters.

[0003] Known methods for determining centering parameters involve capturing uncalibrated images, requiring the attachment of a measuring device to the eyeglasses or eyeglass frame. This is cumbersome and disruptive. Furthermore, US Patent 2003 / 0123026 A1 discloses a device that automatically calculates centering parameters from simultaneously captured calibrated images. DE 10 2011 115 239 A1 discloses a method of the aforementioned type, in which edge profiles are used to represent the spectacle lens edge.

[0004] The object of the invention is to further develop a computer-implemented method of the type mentioned above in such a way that it is easier to carry out.

[0005] This problem is solved according to the invention by a computer-implemented method with the features of claim 1. Advantageous embodiments of the invention are the subject of the dependent claims.

[0006] The invention is based on the idea of ​​providing a simplified model for the spectacle lenses to be accommodated in the spectacle frame by determining parameters describing the geometry of the spectacle frame and in particular parameters describing the nasal and / or temporal frame edges, and adapting the model for the spectacle lenses to these parameters.

[0007] Calibrated images are provided. Their calibration encompasses the extrinsic properties of the cameras capturing the images, or the cameras capturing the images sequentially, such as the relative orientation of their optical axes and their relative arrangement in space, as well as their intrinsic properties—that is, the properties of the cameras themselves that define how a point in space, located in the internal coordinate system of the respective camera, is mapped to the coordinates of the pixels in the captured image. A detailed description of camera calibration can be found in the textbook "Multiple View Geometry in Computer Vision" by Richard Hartley and Andrew Zisserman, 2nd edition, Cambridge University Press 2004, and in particular on page 8.

[0008] A further development of the method involves adapting planes to the parameters describing the nasal and / or temporal frame edges as approximations for the surfaces of the lenses to be accommodated in the spectacle frame. This simplifies the model and makes it easier to calculate. Advantageously, the parameters describing the nasal and temporal frame edges are determined using epipolar geometry and / or triangulation. It is preferred that at least three calibrated images of the head, taken simultaneously from different directions, are provided, with a frontal image showing the head from the front and two lateral images showing the head from the left and right. Advantageously, the directions of capture of the lateral images should each form an angle of at least 60 degrees with the direction of capture of the frontal image, e.g., 90° ± 10°.

[0009] According to the invention, the geometric parameters describing the position of the eyes include the position of the corneal vertex in space.

[0010] According to the invention, the position of the corneal apex in space, determined in a first approximation, is subjected to a correction calculation. The type of correction calculation then depends on how the position of the corneal apex in space is determined in the first approximation.

[0011] According to the invention, in a so-called pupil-based evaluation, the position of the corneal vertex in space is determined to a first approximation as the intersection of a ray of vision tangent to the cornea from a side camera capturing the lateral image with a ray of vision directed towards the pupil from a front camera capturing the anterior image. Furthermore, according to the invention, the position of the corneal vertex is calculated using a correction calculation according to a = q + µ * v + µ * w, where a is the position vector of the corneal vertex in space after the correction calculation, q is the position of the corneal vertex to a first approximation, µ is an empirical value for the distance between the pupil center and the corneal vertex, v is a unit vector of the spatial direction from the pupil center to the front camera, and w is a unit vector of the imaging direction passing through the center of the cornea.

[0012] Alternatively, it is also possible to perform a reflex-based evaluation by generating a flash of light, preferably using an LED, when taking the images, whereby the position of the corneal vertex in space is determined to a first approximation as the position of the reflection point of the flash of light on the cornea.Starting from this first approximation, if the light flash is generated by means of a light source positioned centrally in front of the subject's face, the position of the corneal vertex in the horizontal direction from the reflection point can advantageously be calculated using the correction calculation by adding Δx = + / -r * sin (½ * (arccos z / a + arctan x / (zv))) to the x-coordinate, where r is an empirical value for the corneal radius, a is the distance of the optical center of a front camera capturing the front image to the reflection point, x and z are the x- and z-coordinates of the reflection point in a coordinate system with the zero point at the optical center of the front camera, where the recording direction of the front camera corresponds to the z-direction and the x-direction is horizontally orthogonal to the z-direction and points to the right when viewed in the z-direction, and v is the distance of the light source generating the light flash from the optical center of the front camera in the z-direction.In addition or alternatively, the position of the corneal vertex in the y-direction can be advantageously calculated using the correction calculation starting from the reflection point by adding Δy = + / -r * sin ½ * (arctan I / (dv)), where r is an empirical value for the corneal radius, d is the distance of the optical center of the front camera to the reflection point, v is the distance of the light source generating the light flash from the optical center of the front camera in the z-direction, and I is the distance of the light source from the optical center of the front camera in the y-direction, where the y-direction is orthogonal to the x-direction and the z-direction and points upwards in space.

[0013] The plus sign in the x-direction is used when the corneal vertex of the left eye (from the subject's perspective) is detected; the minus sign is used for the right eye (from the subject's perspective). In the y-direction, the plus sign is used when the light source emitting the flash is positioned at a lower height than the front camera; the minus sign is used when it is mounted at a greater height.

[0014] Advantageously, the pupil or reflex point is detected using feature extraction and / or feature matching (feature comparison) and / or machine learning by comparison with a large amount of previously known data. This step can be preceded by face detection and / or the detection of facial features such as the eyes as a preprocessing step, in which it is determined which image data belong to the subject's face, so that only this data is used in the detection process.

[0015] In order to determine the corneal vertex in space for both eyes, it is preferred that at least one calibrated third image, taken laterally with respect to the head at the same time as the first and second images, is provided.

[0016] Preferably, the computer-implemented method according to the invention is carried out with a device as described in principle in claim 15 and in detail in the following description of the figures.

[0017] The invention will now be explained in more detail with reference to an embodiment schematically illustrated in the drawing. The drawing shows: Figure 1a, a device for determining centering parameters in perspective view and in a front view; Figure 2, an illustration of the correction calculation for pupil-based determination of the corneal vertex position; Figure 3a, an illustration of the correction calculation for reflex-based determination of the corneal vertex position; Figures 4a, b, views of a head wearing glasses with approximate frame edges projected onto it from the front and side; and Figure 5, a schematic representation of approximately determined lens planes.

[0018] The device 10 shown in the drawing serves to determine centering parameters for spectacle fitting. It has a column 12 which supports a height-adjustable camera carrier 14, which in turn supports a number of cameras 16a, 16b. In plan view, the camera carrier 14 is approximately circularly curved and extends between two free ends 18, which are spaced apart from each other. An inner surface 20 of the camera carrier 14 encloses an interior space 22 to the front, i.e., towards the column 12, and to the sides. During image capture by the cameras 16a, 16b, the head of a test subject is located within this interior space. The inner surface 20 is concavely curved in a direction running between the free ends 18 and has, for example, the shape of a section of a cylindrical surface, where the cylinder can have a circular or oval base.In order to position the camera carrier 14 at the correct height relative to the subject's head, a lifting device (not shown in detail) is arranged in the column 12, with which the camera carrier 14 can be moved up and down by motor.

[0019] All cameras 16a, 16b are arranged in a camera assembly 26 extending between the free ends 18. In the present embodiment, the camera assembly 26 is configured as a camera row 26, the cameras 16a, 16b of which are all located at the same height, with their optical axes directed towards the interior 22. In the present embodiment, the camera row 26 comprises a front camera 16a arranged in the center of the camera carrier 14, the optical axis of which is directed frontally towards the subject's face, and eight side cameras 16b arranged in pairs symmetrically with respect to a perpendicular plane of symmetry passing through the optical axis of the front camera 16a, four of which are directed towards the subject's face from the left and four from the right. The cameras 16a, 16b are also calibrated so that they can simultaneously capture calibrated images of the subject.Calibration encompasses extrinsic properties such as the relative alignment of their optical axes and their relative arrangement in space, as well as intrinsic properties, i.e., the properties of the cameras themselves that define how a point in space, located in the internal coordinate system of the respective camera, is mapped to the coordinates of the pixels of the captured image.

[0020] The camera carrier 14 encloses the interior space 22 only at the front, towards the column 12, and at the sides, i.e., to the left and right of the subject's head. It is open at the top, bottom, and rear 30, with the free ends 18 being at least 25 cm apart, allowing the subject to comfortably approach from behind. In the illustrated embodiment, this distance is 70 to 80 cm.

[0021] For illuminating the interior 22, a lighting device is provided with an upper light bar 32 extending above the camera array 26 and a lower light bar 34 extending below the camera array 26, each comprising a plurality of LEDs as light sources. The upper light bar 32 and the lower light bar 34 each extend continuously or with interruptions over a length at least as long as the circumferential length of the camera array 26 measured between its free ends 18. This corresponds to a circumferential angle of at least 160 degrees. Near the free ends 18, the upper light bar 32 and the lower light bar 34 are each connected to each other by means of a further vertically extending light bar 36. The camera array 26 is thus completely framed by at least one row of LEDs.The device 10 also includes a control or regulating device, not shown in detail in the drawing, with which the light intensity emitted by the LEDs can be controlled or regulated depending on the light intensity detected by the cameras 16a, 16b. The LEDs of the light strips 32, 34, 36 are grouped into sectors whose emitted light intensities can be controlled or regulated separately. Furthermore, it is provided that the light intensities emitted by the individual LEDs can also be controlled or regulated separately by means of the control or regulating device.

[0022] To correctly position the subject within the interior 22, the two side cameras 16b closest to the front camera 16a are configured to measure the distance of the subject's head from the center 38 of the camera carrier 14. A display unit (not shown) indicates to the subject whether they are correctly positioned. This display unit features several differently colored light sources arranged in a row. The central light source glows green when the subject is correctly positioned. Extending from the central light source, there are yellow, orange, and red light sources in each direction, in that order. The corresponding color indicates whether the subject is slightly, significantly, or much too far from the center 38 of the camera carrier 14, or slightly, significantly, or much too close to it.To ensure that the subject's field of view is directed towards infinity when determining the centering parameters, a fixation device 42 is provided on the camera carrier 14, which generates a fixation pattern for the subject in the form of a speckle pattern. The fixation pattern is positioned slightly higher than the front camera 16a, so that the subject looks over it. This allows their face to be captured to the greatest extent possible.

[0023] The device 10 is particularly suitable for creating an avatar of the subject's head, which can be used to determine the centering parameters. For this purpose, calibrated images of the subject's head without glasses or frames are captured by cameras 16a and 16b. Using a suitable process for geometric position determination, such as triangulation, a depth profile of the head is created that approximates it very accurately. The head is represented by a multitude of points, which can be connected to each other using a grid pattern or stored as a point cloud. In the subsequent determination of the centering parameters, the avatar thus generated can be used to determine centering parameters that cannot be determined, or can only be determined approximately, due to the geometric properties of the glasses or frames worn by the subject.For example, a wide temple can obscure the eye in a side view to such an extent that the vertex distance cannot be determined, or can only be determined very inaccurately. Furthermore, tinted or highly reflective lenses can make it difficult or impossible to see the eyes clearly. To address this, the depth profile of the avatar is projected onto the images of the subject wearing the glasses or frames, which are captured by cameras 16a and 16b. The centration parameters, which can only be determined inadequately due to the limited vision caused by the glasses or frames, are then determined using the avatar's image data. The avatar can be adjusted to the images of the subject wearing the glasses or frames to minimize deviations.

[0024] The device 10 described above can be used for pupil-based detection of a corneal vertex as well as for reflex-based detection of a corneal vertex in both eyes of the subject as follows.

[0025] In the pupil-based method according to Figure 2The position of the corneal vertex in space is first determined as an approximation of the intersection point q of a first visual ray 52 tangent to the cornea 50 from one of the side cameras 16b capturing a lateral image of the subject with a second visual ray 56 directed towards the pupil 54 from the front camera 16a capturing an anterior image of the subject. A corrected position of the corneal vertex in space is then calculated using the equation a = q + µ * v + µ * w. Here, µ is an empirical value for the distance between the pupil center and the corneal vertex, which regularly takes on values ​​between 2.5 mm and 4 mm. v is a unit vector of the spatial direction from the pupil center p to the front camera 16a, whose coordinates are given by the variable c1, and is calculated as v = (p - c1) / |p - c1|.w is a unit vector of the imaging direction passing through the center m of the corneal sphere, directed towards the fixation pattern of the fixation device 42 at the spatial point t, and is calculated as w = (t - m) / | t - m |. All values ​​a, q, p, c1, t and m are three-dimensional vectors.

[0026] In the reflex-based determination of the position of the corneal vertex according to Figure 3a , 3b Two correction calculations are to be made, the first of which is ( Figure 3a ) one correction in the x-direction, the second correction ( Figure 3b) a correction in the y-direction. These spatial directions are defined by an internal coordinate system of the front camera 16a, which has its zero point at the optical center of the front camera 16a. The z-direction is defined by the recording direction of the front camera 16a, the x-direction is a direction that runs horizontally and orthogonally to the z-direction and points to the right when viewed in its direction, and the y-direction runs orthogonally to both the x-direction and the z-direction and points upwards in space. In the reflection-based measurement, a flash of light is emitted by means of a light source, in this case an LED 58, the reflection of which on the cornea is detected by the front camera 16a and at least one of the side cameras 16b and forms the first approximation for the position of the corneal apex in space. The reflection point is in Figure 3a , 3bdenoted by "approx". In the x-direction, a correction is made by adding Δx = + / -r * sin (½ * (arccos z / a + arctan x / (zv))) to the x-coordinate of the reflection point approx, where the plus sign is used when applying to the left eye, the minus sign when applying to the right eye (cf. Figure 3a ) to be used. r is an empirical value for the corneal radius, which is typically approximately 8 mm. a is the distance from the optical center of the front camera 16a to the reflection point approx, and v is the distance of the LED 58 to the optical center of the front camera 16a in the z-direction. x and z, in turn, are the coordinates in the x- and z-directions.

[0027] In the y-direction, a correction is made starting from the reflection point approx by adding Δy = + / -r * sin ½ * (arctan I / (dv)). r is the empirical value for the corneal radius, d is the distance from the optical center of the front camera 16a to the reflection point approx in the z-direction, v is the distance of the LED 58 to the optical center of the front camera 16a in the z-direction, and I is the distance of the LED 58 from the optical center of the front camera 16a in the y-direction. The plus sign is used when the LED 58 is located below the front camera 16a, i.e., its y-coordinate is smaller than the y-coordinate of the front camera 16a or its optical center. If the LED is located above the front camera 16a, the minus sign is used.

[0028] In the described methods, the pupil or the approximate reflex point can be detected, for example, by feature extraction, feature matching, and / or machine learning through comparison with a large amount of previously known data. This detection step can be preceded by a step in which a face detector identifies which pixels belong to the subject's face or eye area, thus narrowing down the search for the pupil or approximate reflex point.

[0029] The position of the corneal vertex in space is used to determine the centration parameters during spectacle fitting. Using calibrated images captured by cameras 16a and 16b, geometric parameters describing the geometry of the spectacle frame are determined by geometric position determination, in particular by triangulation or epipolar geometry. These parameters include the nasal and temporal frame rims 60 and 62, as shown in Figure 4a, 4b As indicated by example. The parameters 60, 62 describing the nasal and temporal frame edges are used to approximate the surfaces of the lenses to be accommodated in the spectacle frame. Figure 5 The figure schematically shows the planes 64 in front of the respective, approximately represented cornea 66 with schematically represented sight rays 68 of the cameras 16a, 16b. Finally, the centration parameters are calculated from the obtained data.

Claims

1. Computer-implemented method for determining centring parameters for fitting spectacle lenses to a predetermined spectacle frame and to the head of a test subject, wherein at least two calibrated images, which are recorded from different recording directions in relation to one another, of the head wearing the spectacle frame are provided, wherein geometric parameters, which describe the position of the eyes and the geometry of the spectacle frame, are ascertained from the images by means of geometric position determination, wherein a three-dimensional model for the spectacle lenses to be accommodated in the spectacle frame is fitted to the geometric parameters describing the geometry of the spectacle frame, wherein centring parameters are calculated from the geometric parameters describing the position of the eyes and the geometry of the spectacle frame, and wherein planes are fitted to the parameters describing the geometry of the spectacle frame as an approximation for the surfaces of the spectacle lenses to be accommodated in the spectacle frame or the spectacle frame edge accommodating the spectacle lenses, characterized in that the geometric parameters describing the position of the eyes comprise the position of the corneal vertex in space, the position of the corneal vertex is subjected to a correction calculation, the position of the corneal vertex in space is determined in a first approximation as the intersection q of a first line of sight (52), which is tangent to the cornea (50), of a side camera (16b) recording the side image with a second line of sight (56), directed onto the pupil (54), a front camera (16a) recording the front image, and the position of the corneal vertex is calculated by means of the correction calculation according to a = q + µ * v + µ * w, wherein a is the position vector of the corneal vertex and space after carrying out the correction calculation, q is the position of the corneal vertex in a first approximation, µ is an experiential value or an actual value for the distance between the pupil centre and the corneal vertex, v is a unit vector of the spatial direction from the pupil centre to the front camera, and w is a unit vector of the recording direction extending through the centre of the corneal sphere.

2. Computer-implemented method according to Claim 1, characterized in that at least two images of the head wearing the spectacle frame, which are calibrated in relation to one another and are recorded simultaneously from different recording directions, are provided.

3. Computer-implemented method according to Claim 1 or 2, characterized in that the geometric parameters describing the geometry of the spectacle frame comprised parameters describing the nasal and / or temporal frame edges, and the three-dimensional model is fitted to the parameters describing the nasal and / or temporal frame edges.

4. Computer-implemented method according to any one of Claims 1 to 3, characterized in that the parameters describing the geometry of the spectacle frame are ascertained by means of epipolar geometry and / or by means of triangulation.

5. Computer-implemented method according to any one of the preceding claims, characterized in that at least three images of the head, which are recorded simultaneously from different recording directions and are calibrated in relation to one another, are provided, wherein a front image depicts the head from the front and one side image depicts the head from the left and from the right in each case.

6. Computer-implemented method according to Claim 5, characterized in that the recording directions of the side images each enclosed an angle of at least 60° with the recording direction of the front image.

7. Computer program having program code for carrying out all method steps according to any one of Claims 1 to 6 when the computer program is loaded in a computer and executed in a computer.

8. Use of a device (10) for carrying out a computer-implemented method according to any one of Claims 1 to 6, wherein the device (10) has a camera carrier (14), which partially encloses an interior (22) open on top, on the bottom, and toward a rear side (30) and carries at least three cameras (16a, 16b), which are arranged between two free ends (18) of the camera carrier (14) and are directed toward the interior (22), wherein the camera carrier (14) comprises a lighting unit (32, 34, 36) for illuminating the interior (22).