Eyeball positioning method, system, device and storage medium in ophthalmic surgery

Abstract

The present invention provides an eyeball positioning method, system, device and storage medium in ophthalmic surgery, wherein the method comprises: obtaining two eye images at adjacent moments and performing image enhancement operations to obtain a first eye image and a second eye image, comparing the two images, and determining whether eyeball rotation occurs; when eyeball rotation occurs, obtaining the first pupil center position corresponding to the first eye image, extracting iris feature pairs of the two images, inputting them into an eyeball rotation recognition model, outputting the eyeball rotation angle, and adjusting the surgical pupil center position according to the eyeball rotation angle and the first pupil center position; when eyeball rotation does not occur, graying the second eye image, and using a grayscale analysis method based on a sliding window to obtain the second pupil center position to adjust the surgical pupil center position. The present invention can realize eyeball positioning under the conditions of eyeball rotation and pupil dilation, and improves the accuracy and reliability of eyeball positioning.

Description

Technical Field

[0001] The present invention relates to the technical field of ophthalmic surgery, and in particular to an eye positioning method, system, electronic device and computer-readable storage medium in ophthalmic surgery. Background Art

[0002] During ophthalmic surgery, there is a need for eye positioning and rotation alignment. In this process, the center of the pupil is usually used as the alignment point for inspection and treatment. However, human eyes rotate, which causes the center of the pupil to shift, and a large shift in the center of the pupil may affect the final effect of the surgery.

[0003] In order to avoid the above situation, the displacement of the pupil center is usually determined according to the rotation of the eyeball to achieve eyeball positioning and rotation alignment. However, due to the existence of pupil reflex, the size of the pupil changes with the light intensity, which causes the pupil center to shift. The existing technology does not consider the pupil positioning method in this case, resulting in doubts about the reliability of the final eyeball positioning.

[0004] Therefore, there is an urgent need for an eye positioning method that can simultaneously realize eye positioning for both eye rotation and pupil dilation, and improve the accuracy and reliability of eye positioning. Summary of the invention

[0005] Based on this, it is necessary to provide an eye positioning method, system, electronic device and computer-readable storage medium for ophthalmic surgery to address the above technical problems.

[0006] A method for eye positioning in ophthalmic surgery comprises the following steps: acquiring two eye images at adjacent moments, and performing image enhancement operations on the two eye images to obtain a first eye image and a second eye image arranged in chronological order; comparing the first eye image and the second eye image to determine whether eye rotation occurs; if eye rotation occurs, acquiring a first pupil center position corresponding to the first eye image, extracting an iris feature pair of the first eye image and the second eye image, inputting the iris feature pair into a constructed eye rotation recognition model, outputting an eye rotation angle, and adjusting the surgical pupil center position according to the eye rotation angle and the first pupil center position; if eye rotation does not occur, graying the second eye image, and using a sliding window-based gray analysis method to obtain a second pupil center position, and adjusting the surgical pupil center position according to the second pupil center position.

[0007] In one embodiment, the image enhancement operation adopts a histogram correction operation.

[0008] In one embodiment, the comparing the first eye image and the second eye image to determine whether eye rotation has occurred includes: locating the iris area in the first eye image and the second eye image, and graying the iris area; mapping the grayed iris area onto a unit circle, and calculating the Zernike moment value; determining whether eye rotation has occurred by the difference between the Zernike moment values ​​of the two images; wherein, when the difference between the Zernike moment values ​​is 0, it indicates that eye rotation has not occurred; and when the difference between the Zernike moment values ​​is not 0, it indicates that eye rotation has occurred.

[0009] In one embodiment, the method of obtaining the first pupil center position corresponding to the first eye image, extracting the iris feature pair of the first eye image and the second eye image, inputting the iris feature pair into a constructed eye rotation recognition model, outputting the eye rotation angle, and adjusting the surgical pupil center position according to the eye rotation angle and the first pupil center position includes: obtaining the first pupil center position corresponding to the first eye image; using semantic segmentation technology to perform semantic segmentation on the first eye image and the second eye image, outputting two iris features, the iris features including iris texture, iris color and iris size, and mapping the two iris features to the same polar coordinates to obtain an iris feature pair; obtaining multiple pairs of sample irises at the same polar coordinates; and obtaining a plurality of sample irises at the same polar coordinates. Feature pairs, a twin neural network is constructed based on a convolutional neural network, the twin neural network includes two identical sub-networks, multiple pairs of the sample iris feature pairs are used to train the twin neural network to obtain an eye rotation recognition model; the iris feature pairs are input into the eye rotation recognition model, and the eye rotation angle is output; the phase angle of the corresponding image is obtained according to the Zernike moment, the difference between the phase angles of the two images is calculated, and the iris rotation angle is estimated according to the difference in phase angles; the eye rotation angle is verified by the iris rotation angle, and if the verification is passed, the current pupil center position is calculated according to the eye rotation angle and the first pupil center position, and the surgical pupil center position corresponding to the surgical equipment is adjusted according to the current pupil center position.

[0010] In one of the embodiments, the second eye image is grayed, and a grayscale analysis method based on a sliding window is used to obtain a second pupil center position, and the surgical pupil center position is adjusted according to the second pupil center position, including: mapping the RGB values ​​of the second eye image to a new coordinate space according to a preset mapping rule to obtain a corresponding grayscale image; obtaining a circular sliding window and a sliding rule of a preset size, and based on the sliding rule, traversing the grayscale image through the circular sliding window to calculate the grayscale density of the eye area covered by the circular sliding window; comparing the grayscale density of the eye area to locate the pupil area, and determining the second pupil center position according to the pupil area; and adjusting the surgical pupil center position corresponding to the surgical equipment according to the second pupil center position.

[0011] In one of the embodiments, after mapping the RGB values ​​of the second eye image to a new coordinate space according to a preset mapping rule to obtain a corresponding grayscale image, the method further includes: using a Canny edge detection algorithm to filter the grayscale image, and performing threshold segmentation based on a maximum between-class variance method or a maximum entropy method to obtain a feature grayscale image; and using a grayscale analysis method based on a sliding window to analyze the feature grayscale image to obtain the second pupil center position.

[0012] In one of the embodiments, the method further includes: verifying the second pupil center position using edge extraction technology or color information.

[0013] An eyeball positioning system in ophthalmic surgery is used to implement the eyeball positioning method in ophthalmic surgery as described above, comprising: an eye image acquisition module, used to acquire two eye images at adjacent moments, and perform image enhancement operations on the two eye images to obtain a first eye image and a second eye image arranged in chronological order; an eyeball rotation judgment module, used to compare the first eye image and the second eye image to determine whether eyeball rotation occurs; an eyeball rotation angle output module, used to acquire a first pupil center position corresponding to the first eye image when eyeball rotation occurs, extract an iris feature pair of the first eye image and the second eye image, input the iris feature pair into a constructed eyeball rotation recognition model, output an eyeball rotation angle, and adjust the surgical pupil center position according to the eyeball rotation angle and the first pupil center position; a pupil center position acquisition module, used to grayscale the second eye image when eyeball rotation does not occur, and obtain the second pupil center position using a grayscale analysis method based on a sliding window, and adjust the surgical pupil center position according to the second pupil center position.

[0014] An electronic device comprises a memory, a processor and a computer program stored in the memory and executable on the processor, wherein when the processor executes the program, the steps of the eye positioning method in ophthalmic surgery described in the above-mentioned embodiments are implemented.

[0015] A computer-readable storage medium stores a computer program, which, when executed by a processor, implements the steps of the eye positioning method in ophthalmic surgery described in the above-mentioned embodiments.

[0016] Compared with the prior art, the advantages and beneficial effects of the present invention are as follows: by acquiring two eye images at adjacent moments and performing image enhancement operations on the two eye images, a first eye image and a second eye image arranged in chronological order are obtained, and the two images are compared to determine whether eye rotation occurs; when eye rotation occurs, the first pupil center position corresponding to the first eye image is acquired, the iris feature pair of the first eye image and the second eye image is extracted, and input into an eye rotation recognition model, the eye rotation angle is output, and the surgical pupil center position is adjusted according to the eye rotation angle and the first pupil center position, so as to realize eye positioning under eye rotation; when eye rotation does not occur, the second eye image is grayed, and the second pupil center position is obtained by using a gray analysis method based on a sliding window to adjust the surgical pupil center position, so as to realize eye positioning under pupil dilation, which can ensure positioning accuracy while improving positioning efficiency; and separate processing is performed according to whether the eye rotates, so as to realize eye positioning in two cases, which can improve eye positioning accuracy and reliability. BRIEF DESCRIPTION OF THE DRAWINGS

[0017] Figure 1 A schematic diagram of a flow chart of an eyeball positioning method in an ophthalmic surgery according to an embodiment;

[0018] Figure 2 is a schematic structural diagram of an eye positioning system in ophthalmic surgery in one embodiment;

[0019] Figure 3 FIG. 1 is a schematic diagram of the internal structure of an electronic device in an embodiment. DETAILED DESCRIPTION

[0020] Before describing the specific embodiments of the present invention, the overall concept of the present invention is described as follows:

[0021] The present invention is mainly developed based on the process of eyeball positioning in ophthalmic surgery. Currently, ophthalmic surgery does not fully consider the pupil center positioning method under the condition of pupil dilation, resulting in doubts about the reliability of the final positioning.

[0022] Therefore, the present invention proposes an eye positioning method in ophthalmic surgery, which obtains two eye images at adjacent moments and performs image enhancement operation on the two eye images to obtain a first eye image and a second eye image arranged in chronological order, and compares the two images to determine whether eye rotation occurs; when eye rotation occurs, the first pupil center position corresponding to the first eye image is obtained, the iris feature pair of the first eye image and the second eye image is extracted, and input into an eye rotation recognition model, the eye rotation angle is output, and the surgical pupil center position is adjusted according to the eye rotation angle and the first pupil center position, so as to realize eye positioning under eye rotation; when eye rotation does not occur, the second eye image is grayed, and the second pupil center position is obtained by using a gray analysis method based on a sliding window to adjust the surgical pupil center position, so as to realize eye positioning under pupil dilation, which can ensure positioning accuracy while improving positioning efficiency; and separate processing is performed according to whether the eye rotates, so as to realize eye positioning in two cases, which can improve eye positioning accuracy and reliability.

[0023] After introducing the overall concept of the present invention, in order to make the purpose, technical solution and advantages of the present invention more clear, the present invention is further described in detail by specific implementation methods in combination with the accompanying drawings. It should be understood that the specific embodiments described herein are only used to explain the present invention and are not used to limit the present invention.

[0024] In one embodiment, Figure 1 As shown, a method for eye positioning in ophthalmic surgery is provided, comprising the following steps:

[0025] Step S110, acquiring two eye images at adjacent moments, and performing image enhancement operation on the two eye images to obtain a first eye image and a second eye image arranged in chronological order.

[0026] Specifically, after the patient arrives at the operating table and the surgical equipment is in place, relevant surgical equipment is used to acquire two eye images at adjacent moments, and image enhancement operations are performed on the two eye images respectively to obtain a first eye image and a second eye image arranged in chronological order, wherein the first eye image is acquired earlier than the second eye image.

[0027] Image enhancement operations can be performed using Fourier transform, wavelet transform, grayscale transform, histogram correction, smoothing filter, and sharpening filter. Image enhancement operations can effectively highlight features of interest, improve the clarity of eye images, and facilitate subsequent accurate eye positioning.

[0028] Among them, the image enhancement operation adopts the histogram correction operation.

[0029] Since in the eye image, the sclera of the eye is brighter, the iris is darker, and the pupil is the darkest, therefore, in this embodiment, a histogram correction operation is used to perform an image enhancement operation on the two eye images to enhance the contrast of the sample images so that the grayscale value of the sclera part after processing becomes higher, while the grayscale value of the iris part becomes lower.

[0030] Step S120 , comparing the first eye image and the second eye image to determine whether eye rotation occurs.

[0031] Specifically, the first eye image and the second eye image are compared to determine whether the eyeball rotates at the current moment, so that different methods can be used to locate the eyeball according to whether the eyeball rotates or not, thereby improving the reliability of eyeball positioning.

[0032] Among them, step S120 includes: locating the iris area in the first eye image and the second eye image, and graying the iris area; mapping the grayed iris area onto the unit circle respectively, and calculating the Zernike moment value; judging whether eye rotation occurs by the difference between the Zernike moment values ​​of the two images; wherein, when the difference between the Zernike moment values ​​is 0, it indicates that eye rotation does not occur; when the difference between the Zernike moment values ​​is not 0, it indicates that eye rotation occurs.

[0033] Specifically, according to the obtained first eye image and second eye image, the iris area in the two is located, and the iris area is grayed, and the grayed iris area is respectively mapped to the unit circle, and its Zernike moment value is calculated. By comparing the difference between the Zernike moment values ​​of the second eye image and the first eye image, it is judged whether the eyeball rotates. Since the phase angle of the corresponding image can be obtained by the Zernike moment value, and the rotation angle of the iris relative to the two images can be estimated according to the difference between the phase angles of the two images, it is possible to directly judge whether the eyeball rotates according to the difference in the Zernike moment values ​​of the two images; wherein, when the Zernike moment value difference is 0, it indicates that the eyeball rotation has not occurred, and when the Zernike moment value difference is not 0, it indicates that the eyeball rotation has occurred. The iris positioning method based on the Zernike moment is used to judge whether the eyeball rotates, and the rotation angle of the iris can be estimated more accurately, so as to improve the robustness and accuracy of eye positioning, and facilitate subsequent eye positioning.

[0034] Of course, the rotation angle of the iris can also be estimated by using a method based on image moments, by calculating the geometric moments of the segmented iris region, especially the rotation angle about the horizontal axis, and calculating the minimum moment of inertia to estimate the rotation angle of the iris.

[0035] Step S130: If eye rotation occurs, the first pupil center position corresponding to the first eye image is obtained, the iris feature pair of the first eye image and the second eye image is extracted, the iris feature pair is input into the constructed eye rotation recognition model, the eye rotation angle is output, and the surgical pupil center position is adjusted according to the eye rotation angle and the first pupil center position.

[0036] Specifically, an eye rotation recognition model is constructed in advance, and when the judgment result is that eye rotation has occurred, the first pupil center position corresponding to the first eye image is obtained, and the iris feature pair of the first eye image and the second eye image is extracted. The iris feature pair is input into the constructed eye rotation recognition model, and the eye rotation angle is output. Based on the eye rotation angle and the first pupil center position, the current pupil center position is calculated and used to adjust the surgical pupil center position, thereby realizing eye positioning in the case of eye rotation with high positioning accuracy.

[0037] In addition, when eye rotation and pupil dilation occur simultaneously, the eye rotation angle can also be obtained through the output of the eye rotation recognition model to improve the accuracy of eye positioning.

[0038] Among them, step S130 includes: obtaining a first pupil center position corresponding to the first eye image; using semantic segmentation technology to perform semantic segmentation on the first eye image and the second eye image, outputting two iris features, the iris features including iris texture, iris color and iris size, and mapping the two iris features to the same polar coordinates to obtain an iris feature pair; obtaining multiple pairs of sample iris feature pairs in the same polar coordinates, constructing a twin neural network based on a convolutional neural network, the twin neural network including two identical sub-networks, using multiple sample iris feature pairs to train the twin neural network to obtain an eye rotation recognition model; inputting the iris feature pair into the eye rotation recognition model, and outputting the eye rotation angle; obtaining the phase angle of the corresponding image according to the Zernike moment, calculating the difference in the phase angles of the two images, and estimating the iris rotation angle according to the difference in the phase angles; using the iris rotation angle to verify the eye rotation angle, if the verification passes, calculating the current pupil center position according to the eye rotation angle and the first pupil center position, and adjusting the surgical pupil center position corresponding to the surgical equipment according to the current pupil center position.

[0039] Specifically, the first pupil center position corresponding to the first eye image is obtained to facilitate the positioning of the current eyeball position; in order to segment the iris features in the first eye image and the second eye image, semantic segmentation technology can be used to segment and obtain two iris features, and the iris features include iris texture, iris color and iris size. When performing iris recognition, the pupil center position corresponding to the first eye image can be used for measurement to improve the accuracy of iris recognition, thereby segmenting and obtaining the corresponding iris features.

[0040] Multiple pairs of sample iris feature pairs in the same polar coordinates are obtained, and a twin neural network is constructed based on a convolutional neural network. The twin neural network includes two identical sub-networks to respectively identify the iris features of the two images, and finally output the eye rotation angle. The twin neural network is trained by multiple pairs of sample iris feature pairs to obtain an eye rotation recognition model; the iris feature pairs are input into the eye rotation recognition model, and the eye rotation angle is output.

[0041] In order to improve the accuracy of the eye rotation angle, the phase angle of the corresponding image can also be obtained according to the Zernike moment, and the iris rotation angle between the two images can be estimated according to the difference in the phase angles of the two images, and the output eye rotation angle is verified by using the iris rotation angle; the verification result is judged according to the set angle difference threshold, and when the angle difference between the eye rotation angle and the iris rotation angle exceeds the angle difference threshold, the verification passes, and the current pupil center position is calculated according to the eye rotation angle and the first pupil center position, and the surgical center position corresponding to the surgical equipment is adjusted according to the current pupil center position to improve the eye positioning accuracy in ophthalmic surgery and ensure the safety of the surgery; otherwise, if the verification fails, the eye rotation angle is obtained again, thereby improving the accuracy and reliability of eye positioning.

[0042] Step S140: If the eyeball does not rotate, the second eye image is grayed, and a gray analysis method based on a sliding window is used to obtain the second pupil center position, and the surgical pupil center position is adjusted according to the second pupil center position.

[0043] Specifically, when the judgment result is that no eye rotation has occurred, the second eye image is grayed, and a grayscale analysis method based on a sliding window is used to obtain the second pupil center position. The surgical pupil center position is adjusted according to the second pupil center position, which can improve the eye positioning efficiency when the pupil is dilated while ensuring the eye positioning accuracy and reliability.

[0044] Among them, step S140 includes: according to a preset mapping rule, mapping the RGB image of the second eye image to a new coordinate space to obtain a corresponding grayscale image; obtaining a circular sliding window and sliding rules of a preset size, traversing the grayscale image through the circular sliding window based on the sliding rules, and calculating the grayscale density of the eye area covered by the circular sliding window; comparing the grayscale density of the eye area, locating the pupil area, and determining the second pupil center position according to the pupil area; adjusting the surgical pupil center position corresponding to the surgical equipment according to the second pupil center position.

[0045] Specifically, a pre-set mapping rule is obtained, and the RGB values ​​of the second eye image are mapped to a new coordinate space to obtain a corresponding grayscale image, so as to obtain a better grayscale effect, so that the grayscale effect of the pupil part can be more prominent. The mapping rule can be obtained by calculating the maximum and minimum values ​​of the RGB space, dividing each color component by the maximum value, obtaining a normalized value, and converting the normalized value to other color spaces, such as HSV or HLS.

[0046] A circular sliding window with a preset sliding rule and a preset size is obtained. Based on the sliding rule, the grayscale image is traversed using the circular sliding window. For example, the circular sliding window is traversed in an order from bottom to top and from left to right throughout the grayscale image. The grayscale density of the eye area covered by the sliding window is calculated at each position. Since the grayscale value of the pupil part is the smallest, the grayscale density of the eye area can be compared using the minimum grayscale principle to locate the pupil area. The center point of the pupil area is determined based on the pupil area to obtain a second pupil center position. The surgical pupil center position corresponding to the surgical device is adjusted based on the second pupil center position, thereby achieving efficient positioning and adjustment of the eyeball.

[0047] Among them, after obtaining the grayscale image, it also includes: using the Canny edge detection algorithm to filter the grayscale image, and performing threshold segmentation based on the maximum inter-class variance method or the maximum entropy method to obtain a feature grayscale image; using a grayscale analysis method based on a sliding window to analyze the feature grayscale image to obtain the second pupil center position.

[0048] Specifically, since the collected images may contain certain noise, in order to improve the image recognition rate, the Canny edge detection algorithm can be used to filter the grayscale image, and threshold segmentation can be performed based on the maximum inter-class variance method or the maximum entropy method to obtain a feature grayscale image and reduce the impact of image noise; the feature grayscale image is analyzed using a grayscale analysis method based on a sliding window to obtain the second pupil center position, thereby improving the accuracy of eye positioning.

[0049] After obtaining the second pupil center position and before adjusting the surgical pupil center position according to the second pupil center position, the method further includes: verifying the second pupil center position by using edge extraction technology or color information.

[0050] Specifically, since eye images may contain factors such as eyelid shadows and moles, after obtaining the second pupil center position, edge extraction technology or color information can be used to verify it, thereby effectively eliminating adverse factors affecting eye recognition and improving the reliability and accuracy of eye positioning.

[0051] In this embodiment, by acquiring two eye images at adjacent moments and performing image enhancement operations on the two eye images, a first eye image and a second eye image arranged in chronological order are obtained, and the two images are compared to determine whether eye rotation occurs; when eye rotation occurs, the first pupil center position corresponding to the first eye image is acquired, and the iris feature pair of the first eye image and the second eye image is extracted and input into an eye rotation recognition model, and the eye rotation angle is output, and the surgical pupil center position is adjusted according to the eye rotation angle and the first pupil center position, so as to realize eye positioning under eye rotation; when eye rotation does not occur, the second eye image is grayed, and the second pupil center position is obtained by a gray analysis method based on a sliding window, so as to adjust the surgical pupil center position, thereby realizing eye positioning under pupil dilation, which can ensure positioning accuracy while improving positioning efficiency; and separate processing is performed according to whether the eye rotates, so as to realize eye positioning in two cases, which can improve eye positioning accuracy and reliability.

[0052] like Figure 2 As shown, an eye positioning system 20 in ophthalmic surgery is provided, which is used to implement the above-mentioned method of eye positioning in ophthalmic surgery, including: an eye image acquisition module 21, an eye rotation judgment module 22, an eye rotation angle output module 23 and a pupil center position acquisition module 24, wherein:

[0053] The eye image acquisition module 21 is used to acquire two eye images at adjacent moments and perform image enhancement operation on the two eye images to obtain a first eye image and a second eye image arranged in chronological order;

[0054] An eyeball rotation determination module 22, used for comparing the first eye image with the second eye image to determine whether eyeball rotation occurs;

[0055] The eyeball rotation angle output module 23 is used to obtain the first pupil center position corresponding to the first eye image when eyeball rotation occurs, extract the iris feature pair of the first eye image and the second eye image, input the iris feature pair to the constructed eyeball rotation recognition model, output the eyeball rotation angle, and adjust the surgical pupil center position according to the eyeball rotation angle and the first pupil center position;

[0056] The pupil center position acquisition module 24 is used to grayscale the second eye image when no eyeball rotation occurs, and obtain the second pupil center position using a grayscale analysis method based on a sliding window, and adjust the surgical pupil center position according to the second pupil center position.

[0057] In one embodiment, the eye rotation judgment module 22 is specifically used to: locate the iris area in the first eye image and the second eye image, and grayscale the iris area; map the grayscaled iris area to the unit circle, and calculate the Zernike moment value; determine whether eye rotation occurs by the difference between the Zernike moment values ​​of the two images; wherein, when the difference between the Zernike moment values ​​is 0, it indicates that eye rotation does not occur; when the difference between the Zernike moment values ​​is not 0, it indicates that eye rotation occurs.

[0058] In one embodiment, the eye rotation angle output module 23 is specifically used to: obtain the first pupil center position corresponding to the first eye image; use semantic segmentation technology to perform semantic segmentation on the first eye image and the second eye image, output two iris features, the iris features include iris texture, iris color and iris size, and map the two iris features to the same polar coordinates to obtain an iris feature pair; obtain multiple pairs of sample iris feature pairs in the same polar coordinates, build a twin neural network based on a convolutional neural network, the twin neural network includes two identical sub-networks, and use multiple sample iris feature pairs to train the twin neural network to obtain an eye rotation recognition model; input the iris feature pair into the eye rotation recognition model, and output the eye rotation angle; obtain the phase angle of the corresponding image according to the Zernike moment, calculate the difference in the phase angles of the two images, and estimate the iris rotation angle according to the difference in the phase angles; use the iris rotation angle to verify the eye rotation angle, if the verification is passed, calculate the current pupil center position according to the eye rotation angle and the first pupil center position, and adjust the surgical pupil center position corresponding to the surgical equipment according to the current pupil center position.

[0059] In one embodiment, the pupil center position acquisition module 24 is specifically used to: map the RGB image of the second eye image to a new coordinate space according to a preset mapping rule to obtain a corresponding grayscale image; obtain a circular sliding window and sliding rule of a preset size, and based on the sliding rule, traverse the grayscale image through the circular sliding window to calculate the grayscale density of the eye area covered by the circular sliding window; compare the grayscale density of the eye area to locate the pupil area, and determine the second pupil center position according to the pupil area; adjust the surgical pupil center position corresponding to the surgical equipment according to the second pupil center position.

[0060] In one embodiment, the pupil center position acquisition module 24 is also used to: use the Canny edge detection algorithm to filter the grayscale image, and perform threshold segmentation based on the maximum inter-class variance method or the maximum entropy method to obtain a characteristic grayscale image; use a grayscale analysis method based on a sliding window to analyze the characteristic grayscale image to obtain a second pupil center position.

[0061] In one embodiment, the pupil center position acquisition module 24 is further used to verify the second pupil center position using edge extraction technology or color information.

[0062] In one embodiment, an electronic device is provided. The device may be a server, and its internal structure diagram may be as follows: Figure 3 As shown. The device includes a processor, a memory, a network interface and a database connected through a system bus. Among them, the processor of the device is used to provide computing and control capabilities. The memory of the device includes a non-volatile storage medium and an internal memory. The non-volatile storage medium stores an operating system, a computer program and a database. The internal memory provides an environment for the operation of the operating system and the computer program in the non-volatile storage medium. The database of the device is used to store configuration templates and can also be used to store target web page data. The network interface of the device is used to communicate with an external terminal through a network connection. When the computer program is executed by the processor, the eye positioning method in ophthalmic surgery is implemented.

[0063] Those skilled in the art will understand that Figure 3 The structure shown in the figure is merely a block diagram of a partial structure related to the scheme of the present application, and does not constitute a limitation on the device to which the scheme of the present application is applied. The specific device may include more or fewer components than shown in the figure, or combine certain components, or have a different arrangement of components.

[0064] In one embodiment, a computer-readable storage medium may be provided, wherein the storage medium stores a computer program, wherein the computer program includes program instructions, and when the program instructions are executed by a computer, the computer executes the method described in the aforementioned embodiment. The computer may be part of the eye positioning system in the above-mentioned ophthalmic surgery.

[0065] A person skilled in the art can understand that all or part of the processes in the above-mentioned embodiments can be implemented by instructing the relevant hardware through a computer program, and the program can be stored in a computer-readable storage medium, and when the program is executed, it can include the processes of the embodiments of the above-mentioned methods. The storage medium can be a disk, an optical disk, a read-only memory (ROM) or a random access memory (RAM), etc.

[0066] Obviously, those skilled in the art should understand that the modules or steps of the present invention described above can be implemented by a general-purpose computing device, they can be concentrated on a single computing device, or distributed on a network composed of multiple computing devices, and optionally, they can be implemented by a program code executable by a computing device, so that they can be stored in a computer storage medium (ROM / RAM, magnetic disk, optical disk) and executed by the computing device, and in some cases, the steps shown or described can be executed in a different order than that here, or they can be made into individual integrated circuit modules, or multiple modules or steps therein can be made into a single integrated circuit module for implementation. Therefore, the present invention is not limited to any specific combination of hardware and software.

[0067] The above contents are further detailed descriptions of the present invention in combination with specific implementation methods, and it cannot be determined that the specific implementation of the present invention is limited to these descriptions. For ordinary technicians in the technical field to which the present invention belongs, several simple deductions or substitutions can be made without departing from the concept of the present invention, which should be regarded as falling within the scope of protection of the present invention.

Claims

1. A method for eyeball positioning in ophthalmic surgery, characterized in that: The following steps are involved: Acquire two eye images at adjacent moments, and perform image enhancement operations on the two eye images to obtain a first eye image and a second eye image arranged in chronological order; comparing the first eye image and the second eye image to determine whether eye rotation occurs; If eye rotation occurs, the first pupil center position corresponding to the first eye image is obtained, the iris feature pair of the first eye image and the second eye image is extracted, the iris feature pair is input into the constructed eye rotation recognition model, the eye rotation angle is output, and the surgical pupil center position is adjusted according to the eye rotation angle and the first pupil center position; If the eyeball does not rotate, the second eye image is grayed, and a grayscale analysis method based on a sliding window is used to obtain the second pupil center position, and the surgical pupil center position is adjusted according to the second pupil center position.

2. The eye positioning method in ophthalmic surgery according to claim 1, characterized in that: The image enhancement operation adopts a histogram correction operation.

3. The eye positioning method in ophthalmic surgery according to claim 1, characterized in that: The comparing the first eye image and the second eye image to determine whether eye rotation occurs includes: Locating iris regions in the first eye image and the second eye image, and performing grayscale processing on the iris regions; Map the grayscaled iris area onto the unit circle and calculate the Zernike moment value; The difference between the Zernike moments of the two images is used to determine whether eye rotation occurs; When the difference between the Zernike moment values ​​is 0, it indicates that no eye rotation occurs; when the difference between the Zernike moment values ​​is not 0, it indicates that eye rotation occurs.

4. The eye positioning method in ophthalmic surgery according to claim 3, characterized in that: The step of obtaining a first pupil center position corresponding to the first eye image, extracting an iris feature pair of the first eye image and the second eye image, inputting the iris feature pair into a constructed eye rotation recognition model, outputting an eye rotation angle, and adjusting a surgical pupil center position according to the eye rotation angle and the first pupil center position includes: Acquire a first pupil center position corresponding to the first eye image; Using semantic segmentation technology, semantic segmentation is performed on the first eye image and the second eye image, two iris features are output, the iris features include iris texture, iris color and iris size, and the two iris features are mapped to the same polar coordinates to obtain an iris feature pair; Acquire multiple pairs of sample iris feature pairs in the same polar coordinates, construct a twin neural network based on a convolutional neural network, the twin neural network includes two identical sub-networks, and use multiple pairs of the sample iris feature pairs to train the twin neural network to obtain an eye rotation recognition model; Inputting the iris feature pair into the eye rotation recognition model, and outputting the eye rotation angle; Obtaining the phase angle of the corresponding image according to the Zernike moment, calculating the difference between the phase angles of the two images, and estimating the iris rotation angle according to the difference in the phase angles; The eyeball rotation angle is verified using the iris rotation angle. If the verification is successful, the current pupil center position is calculated based on the eyeball rotation angle and the first pupil center position, and the surgical pupil center position corresponding to the surgical equipment is adjusted based on the current pupil center position.

5. The eye positioning method in ophthalmic surgery according to claim 1, characterized in that: The grayscale processing of the second eye image and obtaining the second pupil center position by using a grayscale analysis method based on a sliding window, and adjusting the surgical pupil center position according to the second pupil center position include: According to a preset mapping rule, the RGB values ​​of the second eye image are mapped to a new coordinate space to obtain a corresponding grayscale image; Obtaining a circular sliding window of a preset size and a sliding rule, and based on the sliding rule, traversing the grayscale image through the circular sliding window to calculate the grayscale density of the eye area covered by the circular sliding window; Comparing the grayscale density of the eye area, locating the pupil area, and determining the second pupil center position according to the pupil area; Adjust the surgical pupil center position corresponding to the surgical equipment according to the second pupil center position.

6. The eye positioning method in ophthalmic surgery according to claim 5, characterized in that: After mapping the RGB values ​​of the second eye image to a new coordinate space according to a preset mapping rule to obtain a corresponding grayscale image, the method further includes: The grayscale image is filtered using a Canny edge detection algorithm, and a threshold segmentation is performed based on a maximum inter-class variance method or a maximum entropy method to obtain a feature grayscale image; The characteristic grayscale image is analyzed using a grayscale analysis method based on a sliding window to obtain the second pupil center position.

7. The eye positioning method in ophthalmic surgery according to claim 1, characterized in that: Also includes: The second pupil center position is verified using edge extraction technology or color information.

8. An eye positioning system for ophthalmic surgery, characterized in that: A method for eye positioning in ophthalmic surgery according to any one of claims 1 to 7, comprising: An eye image acquisition module, used to acquire two eye images at adjacent moments, and perform image enhancement operation on the two eye images to obtain a first eye image and a second eye image arranged in chronological order; An eyeball rotation determination module, used for comparing the first eye image with the second eye image to determine whether eyeball rotation occurs; an eyeball rotation angle output module, used for obtaining the first pupil center position corresponding to the first eye image when eyeball rotation occurs, extracting the iris feature pair of the first eye image and the second eye image, inputting the iris feature pair into the constructed eyeball rotation recognition model, outputting the eyeball rotation angle, and adjusting the surgical pupil center position according to the eyeball rotation angle and the first pupil center position; The pupil center position acquisition module is used to grayscale the second eye image when no eye rotation occurs, and obtain the second pupil center position by using a grayscale analysis method based on a sliding window, and adjust the surgical pupil center position according to the second pupil center position.

9. An electronic device comprising a memory, a processor, and a computer program stored in the memory and executable on the processor, characterized in that: When the processor executes the computer program, the steps of the method according to any one of claims 1 to 7 are implemented.

10. A computer-readable storage medium having a computer program stored thereon, characterized in that: When the computer program is executed by a processor, the steps of the method according to any one of claims 1 to 7 are implemented.

Images