Refractive medium anomaly detection method, storage medium and electronic equipment

Abstract

The invention provides a refractive medium anomaly detection method, a storage medium and electronic equipment, and the method comprises the steps: carrying out the image quality enhancement processing of an eyeball shooting image after the eyeball shooting image of a to-be-detected target is obtained, obtaining an enhanced eyeball image, and cutting the enhanced eyeball image according to the boundary of a pupil region, thereby obtaining a pupil image of the to-be-detected target. A dioptric medium detection model is obtained in advance in a learning training mode, a pupil image is input into the dioptric medium detection model, and then a region segmentation result of the pupil image can be obtained, and the region segmentation result comprises a pupil normal region, a pupil abnormal region and an eyelash region. And determining a refractive medium anomaly detection result of the target to be detected according to the position and the area ratio of the pupil anomaly region in the pupil image. According to the method, the segmentation precision and accuracy of the abnormal region in the pupil image can be improved, false detection of eyelashes is avoided, the accuracy of refractive medium anomaly detection is improved, and then the accuracy of judging whether cataract exists or not can be improved.

Description

Technical Field

[0001] The present application relates to the technical field of eye refractive medium screening, and in particular to a refractive medium abnormality detection method, storage medium and electronic device. Background Art

[0002] Refractive media, including the cornea, aqueous humor, lens, and vitreous body, are the pathways for light to enter the eye and reach the retina. The refractive media are transparent and avascular as a whole to achieve the best refractive effect. Abnormalities in refractive media can lead to vision problems, especially cataracts. The early symptoms of cataracts are usually not obvious, but as the disease progresses, symptoms such as blurred vision, diplopia, myopia, and glare will appear, and may cause complete blindness in patients. Therefore, it is very important to determine whether cataracts are present by screening for abnormalities in refractive media.

[0003] At present, the diagnosis of cataracts is mainly done by taking a slit lamp to photograph the eyeball to obtain a pupil image, and doctors make judgments based on the pupil image. When taking pictures with a slit lamp, the light source is visible light, and the light spots reflected from the fundus during the shooting are likely to affect the judgment results. In addition, the eyelashes may overlap with the pupil during the shooting process and are also considered to be part of the pupil. Therefore, the pupil image cannot truly represent the state of the eyeball, and the judgment results obtained based on it will also be biased. Summary of the invention

[0004] The technical problem to be solved by the present application is that the refractive medium abnormality detection scheme in the prior art is prone to deviation, and thus a refractive medium abnormality detection method, storage medium and electronic device are provided.

[0005] In a first aspect, the technical solution of the present application provides a method for detecting abnormality of a refractive medium, comprising:

[0006] Acquire an eyeball image of the target to be tested;

[0007] Performing image quality enhancement processing on the eyeball image to obtain an enhanced eyeball image;

[0008] Determine a pupil area in the enhanced eyeball image, and clip the enhanced eyeball image according to a boundary of the pupil area to obtain a pupil image of the target to be measured;

[0009] Inputting the pupil image into a refractive medium detection model that has been trained in advance, obtaining a regional segmentation result of the pupil image output by the refractive medium detection model, wherein the regional segmentation result includes: a normal pupil area, an abnormal pupil area, and an eyelash area contained in the pupil image; wherein the refractive medium detection model is trained using pupil image samples that have been processed with enhanced image quality;

[0010] The refractive medium abnormality detection result of the target to be measured is determined according to the position and area ratio of the abnormal pupil area in the pupil image.

[0011] In some embodiments of the refractive medium abnormality detection method, the image quality enhancement process of the eyeball image is performed to obtain an enhanced eyeball image, including:

[0012] Acquire spatial information of the image captured by the eyeball, and express the spatial information as the sum of an illumination component and a reflection component;

[0013] Performing transformation processing on the spatial information to obtain frequency domain information corresponding to the spatial information;

[0014] Performing filtering on the frequency domain information to obtain the sum of the high frequency component and the suppressed low frequency component in the frequency domain information as filtered frequency domain information;

[0015] Transforming the filtered frequency domain information to obtain filtered spatial information corresponding to the filtered frequency domain information, wherein the filtered spatial information is represented as the sum of a filtered illumination component and a filtered reflection component;

[0016] The enhanced eyeball image is obtained according to the filtered illumination component and the filtered reflection component.

[0017] In some embodiments of the refractive medium abnormality detection method, the refractive medium detection model is obtained by:

[0018] Obtain pupil image samples after image quality enhancement processing;

[0019] Annotating a normal pupil area, an abnormal pupil area, and an eyelash area in the pupil image sample to obtain a pupil image sample with annotated labels;

[0020] An initial deep learning model is selected, and the pupil image samples with annotated labels are used as input. The initial deep learning model is trained with the sample area segmentation results and confidence obtained by segmenting the pupil image samples according to the annotated labels as outputs. When the confidence reaches the set confidence, the training is stopped to obtain the refractive medium detection model; the sample area segmentation results include the normal pupil area, abnormal pupil area and eyelash area contained in the pupil image sample.

[0021] In some embodiments, the method for detecting abnormal refractive medium, wherein the step of acquiring spatial information of the image captured by the eyeball and expressing the spatial information as the sum of an illumination component and a reflection component comprises:

[0022] Obtain the spatial data of the eyeball image: f(x,y)=f i (x,y)fr (x, y); where x and y represent the positions of pixels in the eye-captured image; f i (x, y) represents the illumination function of the image captured by the eye, and the illumination function varies with position; f r (x, y) represents the reflection function of the image captured by the eyeball, and the reflection function varies with position;

[0023] Logarithmic transformation is performed on the spatial data of the eyeball image to obtain the spatial information: lnf(x, y) = lnf i (x,y)+lnf r (x,y); where lnf i (x,y) represents the illumination component and lnf r (x,y) represents the reflection component.

[0024] In some embodiments, the method for detecting abnormal refractive media, filtering the frequency domain information to obtain the sum of the high-frequency component and the suppressed low-frequency component in the frequency domain information as the filtered frequency domain information, comprises:

[0025] The frequency domain filter H is implemented using the following function mo (u, v) performs the filtering process on the frequency domain information:

[0026] H mo (u,v)=(γ H -γ L )H hp (u,v)+γ L ;

[0027] H hp (u,v)=1-exp(-c(D 2 (u,v) / D0 2 ));

[0028] Where: u, v represent the horizontal and vertical coordinates in the frequency domain coordinate system; H hp (u, v) is the high-pass filter function; D(u, v) represents the input frequency domain signal of the high-pass filter function, D0 represents the cutoff frequency of the high-pass filter function; c is the high-pass filter morphology control parameter; γ H represents the high frequency component weight, γ L Represents the weight of low-frequency components.

[0029] In some schemes of the refractive medium abnormality detection method, the eyeball image and the pupil image samples are both images captured under an infrared light source.

[0030] In some embodiments, the refractive medium abnormality detection method, determining the refractive medium abnormality detection result of the target to be detected according to the position and area ratio of the pupil abnormality area in the pupil image, comprises:

[0031] Obtain the center point position (x0, y0) and area A0 of the pupil image;

[0032] Obtain the center point position (xi, yi) and area Ai of the pupil abnormality area;

[0033] The area ratio is calculated as follows: R = Ai / A0;

[0034] The distance between the center point of the pupil abnormality area and the center point of the pupil image is calculated:

[0035]

[0036] The abnormal grade of the refractive medium is calculated as follows: G=k1×R+k2 / S; wherein K1 and K2 are preset weight coefficients.

[0037] In a second aspect, the technical solution of the present application provides a computer-readable storage medium, in which program information is stored. After a computer reads the program information, the computer executes the steps of the refractive medium abnormality detection method described in any one of the technical solutions of the first aspect.

[0038] In a third aspect, the technical solution of the present application provides a computer program product, including a computer program / instruction, characterized in that when the computer program / instruction is executed by a processor, the steps of the refractive medium abnormality detection method described in any technical solution of the first aspect are implemented.

[0039] In a fourth aspect, the technical solution of the present application provides an electronic device, comprising a memory, a processor, and a computer program stored in the memory, characterized in that the processor executes the computer program to implement the steps of the refractive medium abnormality detection method described in any technical solution of the first aspect.

[0040] Compared with the prior art, the above technical solution provided by this application has the following technical effects:

[0041] The refractive medium abnormality detection method, storage medium and electronic device provided by the present application, after acquiring the eyeball captured image of the target to be measured, first perform image quality enhancement processing on the eyeball captured image to obtain the enhanced eyeball image, so as to improve the contrast and clarity of the pupil area in the eyeball captured image. On this basis, the pupil area in the enhanced eyeball image can be determined more accurately. Accordingly, the enhanced eyeball image can be cropped according to the boundary of the pupil area to obtain the pupil image of the target to be measured. A refractive medium detection model is obtained in advance by learning and training, and the pupil image with higher contrast and clarity obtained above is input into the refractive medium detection model to obtain the regional segmentation result of the pupil image, and the regional segmentation result includes a normal pupil area, an abnormal pupil area and an eyelash area. According to the position and area proportion of the abnormal pupil area in the pupil image, the refractive medium abnormality detection result of the target to be measured can be determined, excluding the influence of the eyelash area. Through the above-mentioned scheme of the present application, the segmentation accuracy and precision of abnormal areas in the pupil image can be improved, the false detection of eyelashes can be avoided, the accuracy of abnormal refractive medium detection can be improved, and the accuracy of judging whether cataracts are present can be improved. BRIEF DESCRIPTION OF THE DRAWINGS

[0042] Figure 1 This is a flow chart of a method for detecting abnormality of a refractive medium according to an embodiment of the present application;

[0043] Figure 2 A schematic diagram of cropping a pupil image in an eyeball image according to an embodiment of the present application;

[0044] Figure 3 A comparison diagram of the original pupil image and the pupil image obtained in the embodiment of the present application;

[0045] Figure 4a This is a schematic diagram of the normal pupil area obtained by pupil image segmentation; Figure 4b A schematic diagram of the abnormal pupil area obtained by pupil image segmentation; Figure 4c Schematic diagram of the eyelash region obtained by pupil image segmentation;

[0046] Figure 5 A curve diagram of a frequency domain filter function according to an embodiment of the present application;

[0047] Figure 6 This is a schematic diagram of the implementation process of the inverted residual module of the Mobile-Unet network model;

[0048] Figure 7 This is a logical diagram of the Mobile-Unet network model used in the initial deep learning model;

[0049] Figure 8A schematic diagram of the hardware connection relationship of an electronic device for executing a refractive medium abnormality detection method according to an embodiment of the present application. DETAILED DESCRIPTION

[0050] The specific implementation of the present application is further described below with reference to the accompanying drawings.

[0051] It is easy to understand that according to the technical solution of the present application, without changing the essential spirit of the present application, a variety of structural modes and implementation modes that can be replaced by those skilled in the art can be replaced with each other. Therefore, the following specific implementation modes and drawings are only exemplary descriptions of the technical solution of the present application, and should not be regarded as the entirety of the present application or as a limitation or restriction on the technical solution of the application.

[0052] The directional terms such as up, down, left, right, front, back, front, back, top, bottom, etc. mentioned or may be mentioned in this specification are defined relative to the structures shown in the drawings. They are relative concepts and may change accordingly according to different positions and different usage conditions. Therefore, these or other directional terms should not be interpreted as restrictive terms.

[0053] This embodiment provides a method for detecting abnormal refractive medium, which is applied to a control system of an instrument for eye examination, such as Figure 1 As shown, the method comprises the following steps:

[0054] S10: Acquire an eyeball image of the target to be measured.

[0055] Specifically, the target to be tested is the inspected object, the face is photographed by an image acquisition device, and a clear image containing the eyeball is taken as the eyeball photographed image, such as Figure 2 To ensure the integrity of the eyeball, the eyeball image includes the forehead area and the face area, which can be removed by cropping later.

[0056] S20: performing image quality enhancement processing on the eyeball image to obtain an enhanced eyeball image.

[0057] Specifically, image quality enhancement processing is used to improve the quality of images captured by the eye, including adjusting image parameters such as clarity, contrast, and saturation.

[0058] S30: Determine a pupil area in the enhanced eyeball image, and crop the enhanced eyeball image according to a boundary of the pupil area to obtain a pupil image of the target to be measured.

[0059] Specifically, Figure 3As shown, the left side is the original pupil image, and the right side is the pupil image obtained by cropping after image quality enhancement. By comparison, the pupil image in the present application scheme improves the display quality, especially the light spot of the light source is clearer, the boundary of the abnormal area is clearer, the eyelash display is clearer, and the pupil boundary is clearer.

[0060] S40: Input the pupil image into a pre-trained refractive medium detection model, and obtain a regional segmentation result of the pupil image output by the refractive medium detection model, wherein the regional segmentation result includes: a normal pupil area, an abnormal pupil area, and an eyelash area contained in the pupil image; wherein the refractive medium detection model is trained using pupil image samples after image quality enhancement processing.

[0061] Specifically, the pupil image samples used in training the refractive medium detection model and the pupil images in this solution are processed in the same image quality enhancement manner to reduce the environmental noise of the analysis results of the refractive medium detection model. Figure 3 The pupil image shown on the right can be segmented into Figure 4a The normal pupil area shown, Figure 4b The abnormal pupil area and Figure 4c Eyelash area shown.

[0062] S50: Determine the refractive medium abnormality detection result of the target to be measured according to the position and area ratio of the abnormal pupil area in the pupil image.

[0063] Specifically, the closer the abnormal pupil area is to the pupil center, the greater the impact on vision; the larger the abnormal pupil area is, the greater the impact on vision. Therefore, the abnormal refractive medium detection result of the target to be measured can be determined based on the position and area proportion of the abnormal pupil area in the pupil image. The upper limit threshold of the distance between the abnormal pupil area and the pupil center can be set. Once the distance between the abnormal pupil area and the pupil center exceeds the upper limit threshold, the refractive medium can be determined to be abnormal. The upper limit threshold of the area ratio of the abnormal pupil area to the pupil image area can also be set. Once the ratio of the abnormal pupil area to the pupil image area exceeds the upper limit threshold of the area ratio, the refractive medium can be determined to be abnormal.

[0064] The above technical solution provided by the present application, after obtaining the eyeball image of the target to be measured, firstly performs image quality enhancement processing on the eyeball image to obtain the enhanced eyeball image, so as to improve the contrast and clarity of the pupil area in the eyeball image. On this basis, the pupil area in the enhanced eyeball image can be determined more accurately. Accordingly, the enhanced eyeball image can be clipped according to the boundary of the pupil area to obtain the pupil image of the target to be measured. The refractive medium detection model is obtained in advance by learning and training, and the pupil image with higher contrast and clarity obtained above is input into the refractive medium detection model to obtain the regional segmentation result of the pupil image, and the regional segmentation result includes the normal pupil area, the abnormal pupil area and the eyelash area. According to the position and area proportion of the abnormal pupil area in the pupil image, the abnormal refractive medium detection result of the target to be measured can be determined, eliminating the influence of the eyelash area. Through the above scheme of the present application, the segmentation accuracy and accuracy of the abnormal area in the pupil image can be improved, the false detection of eyelashes can be avoided, the accuracy of the abnormal refractive medium detection can be improved, and the accuracy of the judgment of whether it is cataract can be improved.

[0065] Furthermore, in the refractive medium abnormality detection method in the above scheme, the step S20 of performing image quality enhancement processing on the eyeball image to obtain an enhanced eyeball image includes:

[0066] S201: Acquire spatial information of the image captured by the eyeball, and express the spatial information as the sum of an illumination component and a reflection component.

[0067] Specifically, spatial information is time domain information. In the time domain space, the image captured by the eye can be seen as consisting of two parts: the light intensity at different spatial positions, which is characterized by slow changes and concentrated in the low-frequency part of the image captured by the eye; the image reflected from the scene to the eye, which contains various information about the scene and is rich in high-frequency components. In order to facilitate the analysis of the image captured by the eye, the spatial information can be represented as the sum of the illumination component and the reflection component. That is, based on the illumination / reflectivity relationship data of the image captured by the eye as the basis for frequency domain processing, the quality of the image captured by the eye is improved by compressing the brightness range and enhancing the contrast. As an implementation method, the present application obtains the spatial information in the following way:

[0068] S2011: Acquire the spatial data of the eyeball image: f(x, y) = f i (x,y)f r (x, y); where x and y represent the positions of pixels in the eye-captured image; f i (x, y) represents the illumination function of the image captured by the eye, and the illumination function varies with position; f r (x, y) represents the reflection function of the image captured by the eyeball, and the reflection function varies with position.

[0069] The eyeball image includes multiple rows and columns of pixels, and the values ​​of x and y can be determined according to the number of rows and columns where the pixels are located. The eyeball image can also be defined in a rectangular coordinate system, and the position of the eyeball image corresponding to the origin of the coordinate system can be selected according to the needs. For example, the pupil center pixel can be selected to overlap with the origin of the coordinate system, and on this basis, the position of each pixel can be determined according to the distance between each pixel and the pupil center pixel. In addition, the illumination function and the reflection function can be determined according to the existing method, which will not be described in detail here.

[0070] S2012: Perform logarithmic transformation on the spatial data of the eyeball image to obtain the spatial information: lnf(x, y) = lnf i (x,y)+lnf r (x,y); where lnf i (x,y) represents the illumination component and lnf r (x,y) represents the reflection component.

[0071] S202: Perform transformation processing on the spatial information to obtain frequency domain information corresponding to the spatial information.

[0072] Specifically, the spatial information is mapped from the time domain to the frequency domain through Fourier transform, that is:

[0073] DFT[lnf(x,y)]=DFT[lnf i (x,y)]+DFT[lnf r (x,y)].

[0074] S203: Filter the frequency domain information to obtain the sum of the high frequency component and the suppressed low frequency component in the frequency domain information as filtered frequency domain information.

[0075] Specifically, a frequency domain filter H is selected mo (u,v), filter DFT[lnf(x,y)].

[0076] The image captured by the eyeball is an image with uneven illumination, and the frequency domain filter H mo (u,v) needs to achieve brightness adjustment and contrast enhancement at the same time, so as to improve the image quality of the image captured by the eyeball. As mentioned above, the image reflected by the scene to the eyeball contains various information about the scene and is rich in high-frequency components, which corresponds to the pupil abnormality area of ​​focus. Therefore, the frequency domain filter H mo (u,v) should suppress the low-frequency brightness component and enhance the high-frequency reflection component. The frequency domain filter H mo (u,v) should be a high-pass filter, but it cannot completely delete the low-frequency components. Its form should be as follows Figure 5shown.

[0077] As a preferred solution, this embodiment uses the frequency domain filter H implemented by the following function: mo (u, v) performs the filtering process on the frequency domain information:

[0078] H mo (u,v)=(γ H -γ L )H hp (u,v)+γ L ;

[0079] H hp (u,v)=1-exp(-c(D 2 (u,v) / D0 2 ));

[0080] Where: u, v represent the horizontal and vertical coordinates in the frequency domain coordinate system; H hp (u, v) is the high-pass filter function; D(u, v) represents the input frequency domain signal of the high-pass filter function, D0 represents the cutoff frequency of the high-pass filter function; c is the high-pass filter morphology control parameter; γ H represents the high frequency component weight, γ L Represents the weight of low-frequency components.

[0081] In the above scheme, the high-pass filter function selects the Gaussian high-pass filter function, and the morphology control parameter c can control the steepness (slope) of the transition section from low frequency to high frequency. The larger the value, the steeper the steepness (slope). In addition, the high-pass filter can also select the Butterworth high-pass filter, Laplacian filter, etc.

[0082] S204: transforming the filtered frequency domain information to obtain filtered spatial information corresponding to the filtered frequency domain information, and the filtered spatial information is represented as the sum of a filtered illumination component and a filtered reflection component.

[0083] Specifically, this step still uses Fourier transform to map the filtered frequency domain information back to the time domain.

[0084] S205: Obtain the enhanced eyeball image according to the filtered illumination component and the filtered reflection component.

[0085] According to step S201, this step takes the index of the filtered illumination component and the filtered reflection component to obtain a time domain filtering result, that is, the enhanced eyeball image can be obtained.

[0086] Through the filtering process in the above scheme, the contrast and standardized brightness of the eye-captured image can be increased at the same time, thereby achieving the purpose of enhancing the image quality of the eye-captured image. The eye-captured image is represented as the product of its illumination function and reflectance function. Although the two are inseparable in the time domain, they can be linearly separated in the frequency domain through Fourier transform. Since the illumination can be regarded as the lighting in the environment, the relative change is very small, and it can be regarded as the low-frequency component of the eye-captured image; while the reflectivity of the eye-captured image has a relatively large change, it can be regarded as a high-frequency component. By separately processing the effects of illumination and reflectivity on the grayscale value of the pixel point, by means of the frequency domain filter provided by the present application, the illumination of the eye-captured image is made more uniform, and the reflective detail features such as light spots and eyelashes are clearer.

[0087] Further preferably, the refractive medium detection model in the above solution is obtained by:

[0088] S1: Obtain pupil image samples after image quality enhancement processing.

[0089] Specifically, this step obtains pupil image samples for training, and the pupil image samples are images processed according to the aforementioned image quality enhancement processing method.

[0090] S2: Annotate the normal pupil area, abnormal pupil area and eyelash area in the pupil image sample to obtain a pupil image sample with annotated labels.

[0091] Specifically, it can be done in the same way as setting labels for training samples.

[0092] S3: Select an initial deep learning model, take the pupil image samples with annotated labels as input, and train the initial deep learning model with the sample area segmentation results and confidence obtained by segmenting the pupil image samples according to the annotated labels as outputs, and stop training when the confidence reaches the set confidence to obtain the refractive medium detection model; the sample area segmentation results include the normal pupil area, abnormal pupil area and eyelash area contained in the pupil image sample.

[0093] The selected initial deep learning model can be selected from existing deep learning algorithms in the prior art.

[0094] In order to simplify the operation process and improve the operation efficiency, the present application scheme selects the Mobile-Unet network model and pupil image samples with annotated labels for training. Specifically, if the lightweight requirements are not met, the backbone network can select the MobileNetV3 network model of a three-layer neural network. The MobileNetV3 network model is a network that combines deep separable convolution, linear bottleneck, inverse residual structure and lightweight attention mechanism. The network has the advantages of small number of parameters, fast speed, moderate depth, etc., and has great advantages as a backbone network for medical image segmentation tasks. Among them: the deep separable convolution uses a deep separable convolution with a convolution kernel of 3x3, which can reduce the amount of calculation by 8 to 9 times compared to ordinary convolution. The linear bottleneck structure is a bottleneck structure in which the last layer of convolution uses linear activation (replacing the ReLU function with a linear function). When the output dimension is low, the ReLU function is used, which can easily cause information loss, so linear activation is chosen at the last layer. The logic of the inverse residual structure is as follows Figure 6 As shown in the figure, the essence of the whole process is "expansion-convolution-compression", which is shuttle-shaped; the convolution operation is: use 1×1PW Conv to perform convolution to increase the dimension, use 3×3DWConv to perform channel-by-channel convolution to extract features, and then use 1×1PW Conv to perform convolution to reduce the dimension again; as shown in the figure, the propagation between two adjacent layers is implemented using BN activation function and ReLU6 activation function. The purpose of the first dimensional convolution operation in the inverted residual is to expand the number of channels in order to extract more information. DW Conv cannot change the number of channels, so the number of channels is expanded first, and 3×3DW Conv is performed in the high-dimensional space to extract more information.

[0095] The Unet network is a fully convolutional neural network and is currently a popular model for medical image segmentation. The network consists of two parts: an encoder (upsampling) and a decoder (downsampling). The encoding part reduces the dimension and extracts features of the input image through convolution and pooling; the decoding part increases the dimension and amplifies the features of low-pixel (shallow features) images through upsampling. The different feature maps obtained by encoding will be connected to the corresponding layers of the decoding part in a channel fusion manner, where shallow features are used for segmentation and deep features are used for positioning. The effective combination of the two helps to increase the diversity of features. Therefore, as a specific task-oriented network, it can be better applied to the segmentation of medical images. This solution adopts Figure 7 The Mobile-Unet network model shown in the figure uses the MobileNetV3 lightweight network as mentioned above as the backbone network, and the task network is the commonly used Unet network. First, the bneck module in MobileNetV3 is transplanted to the encoding part of the Unet network; secondly, bneck modules of different depths are used to extract features according to the position of the Unet encoding part. Generally, the deeper the position, the more bneck modules are required, and the more advanced the abstract features extracted.

[0096] The Mobile-Unet network selected in this application extracts the features of MobileNetV3 and combines them with the feature connections of Unet, which helps to reduce the problem of too many parameters and overfitting in the network layers, thus having the advantages of high accuracy, simple calculation and fewer parameters.

[0097] Further preferably, in the above solution, the eyeball image and the pupil image samples are both images taken under an infrared light source. In this solution, shooting under an infrared light source can prevent the pupil of the target to be measured from contracting after being stimulated by light, so that the pupil is as large as possible during the shooting process, which is more conducive to the screening of refractive medium abnormalities.

[0098] As mentioned above, the closer the abnormal pupil area is to the pupil center, the greater the impact on vision; the larger the abnormal pupil area, the greater the impact on vision. In this solution, in order to quantify the test results, the refractive medium abnormality level is provided as a parameter of the test result. On this basis, the refractive medium abnormality detection result of the target to be tested is determined according to the position and area proportion of the abnormal pupil area in the pupil image in step S50, including:

[0099] S501: Obtain the center point position (x0, y0) and area A0 of the pupil image.

[0100] Specifically, x0 and y0 are determined according to the row and column numbers of the center point of the pupil image, and the area calculation formula is determined according to the pupil shape. Usually, the pupil is a circle, and the area A0 can be solved according to the circular area calculation formula.

[0101] S502: Obtain the center point position (xi, yi) and area Ai of the abnormal pupil area.

[0102] Specifically, xi and yi are determined according to the row and column numbers where the center point of the abnormal pupil area is located, and the area calculation formula is determined according to the shape of the abnormal pupil area.

[0103] S503: Calculate the area ratio: R=Ai / A0.

[0104] S504: Calculate the distance between the center point of the pupil abnormality area and the center point of the pupil image:

[0105]

[0106] S505: Calculate the abnormal level of the refractive medium: G=k1×R+k2 / S; wherein k1 and k2 are preset weight coefficients.

[0107] k1 and k2 can be adjusted according to actual conditions, and are generally set to 0.5 and 0.5. Through this application scheme, the abnormal level of the refractive medium of the target to be measured can be obtained and the detection result can be quantified.

[0108] An embodiment of the present application provides a computer-readable storage medium, in which program information is stored. After a computer reads the program information, the computer executes the steps of the refractive medium abnormality detection method described in any one of the schemes in the above method embodiments.

[0109] An embodiment of the present application provides a computer program product, including a computer program / instruction, characterized in that when the computer program / instruction is executed by a processor, the steps of the refractive medium abnormality detection method described in any one of the schemes in the above method embodiments are implemented.

[0110] The present application also provides an electronic device, such as Figure 8As shown, the electronic device includes at least one processor 81 and at least one memory 82, at least one of the memories 82 stores program information, and at least one of the processors 81 reads the program information and executes the refractive medium abnormality detection method described in any one of the above method embodiments. The device may also include: an input device 83 and an output device 84. The processor 81, the memory 82, the input device 83 and the output device 84 can be connected in communication. The memory 82, as a non-volatile computer-readable storage medium, can be used to store non-volatile software programs, non-volatile computer executable programs and modules. The processor 81 executes various functional applications and data processing by running the non-volatile software programs, instructions and modules stored in the memory 82, that is, the refractive medium abnormality detection method provided by any one of the above schemes is implemented. The memory 82 may include a program storage area and a data storage area, wherein the program storage area may store an operating system and an application required by at least one function; the data storage area may store data created according to the use of the refractive medium abnormality detection method, etc. In addition, the memory 82 may include a high-speed random access memory, and may also include a non-volatile memory, such as at least one disk storage device, a flash memory device, or other non-volatile solid-state storage devices. In some embodiments, the memory 82 may optionally include a memory remotely arranged relative to the processor 81, and these remote memories may be connected to the device for executing the refractive medium abnormality detection method through a network. Examples of the above-mentioned network include but are not limited to the Internet, an intranet, a local area network, a mobile communication network, and a combination thereof. The input device 83 may receive input user clicks, and generate signal inputs related to user settings and function controls of the refractive medium abnormality detection method. The output device 84 may include a display device such as a display screen. When the one or more modules are stored in the memory 82, when they are run by the one or more processors 81, the refractive medium abnormality detection method in any of the above-mentioned method embodiments is executed.

[0111] As needed, the above technical solutions can be combined to achieve the best technical effect.

[0112] The above are only the principles and preferred embodiments of the present application. It should be noted that, for ordinary technicians in this field, on the basis of the principles of the present application, several other modifications can be made, which should also be considered as the protection scope of the present application.

Claims

1. A method for detecting abnormality of a refractive medium, characterized in that: include: Acquire an eyeball image of the target to be tested; Performing image quality enhancement processing on the eyeball image to obtain an enhanced eyeball image; Determine a pupil area in the enhanced eyeball image, and clip the enhanced eyeball image according to a boundary of the pupil area to obtain a pupil image of the target to be measured; Inputting the pupil image into a refractive medium detection model that has been trained in advance, obtaining a regional segmentation result of the pupil image output by the refractive medium detection model, wherein the regional segmentation result includes: a normal pupil area, an abnormal pupil area, and an eyelash area contained in the pupil image; wherein the refractive medium detection model is trained using pupil image samples that have been processed with enhanced image quality; The refractive medium abnormality detection result of the target to be measured is determined according to the position and area ratio of the abnormal pupil area in the pupil image.

2. The method for detecting abnormality of refractive medium according to claim 1, characterized in that: The step of performing image quality enhancement processing on the eyeball image to obtain an enhanced eyeball image includes: Acquire spatial information of the image captured by the eyeball, and express the spatial information as the sum of an illumination component and a reflection component; Performing transformation processing on the spatial information to obtain frequency domain information corresponding to the spatial information; Performing filtering on the frequency domain information to obtain the sum of the high frequency component and the suppressed low frequency component in the frequency domain information as filtered frequency domain information; Transforming the filtered frequency domain information to obtain filtered spatial information corresponding to the filtered frequency domain information, wherein the filtered spatial information is represented as the sum of a filtered illumination component and a filtered reflection component; The enhanced eyeball image is obtained according to the filtered illumination component and the filtered reflection component.

3. The method for detecting refractive medium abnormality according to claim 2, characterized in that: The refractive medium detection model is obtained in the following manner: Obtain pupil image samples after image quality enhancement processing; Annotating a normal pupil area, an abnormal pupil area, and an eyelash area in the pupil image sample to obtain a pupil image sample with annotated labels; An initial deep learning model is selected, and the pupil image samples with annotated labels are used as input. The initial deep learning model is trained with the sample area segmentation results and confidence obtained by segmenting the pupil image samples according to the annotated labels as outputs. When the confidence reaches the set confidence, the training is stopped to obtain the refractive medium detection model; the sample area segmentation results include the normal pupil area, abnormal pupil area and eyelash area contained in the pupil image sample.

4. The method for detecting refractive medium abnormality according to claim 2, characterized in that: The step of acquiring the spatial information of the image captured by the eyeball and expressing the spatial information as the sum of an illumination component and a reflection component includes: Obtain the spatial data of the eyeball image: f(x,y)=f i (x,y)f r (x, y); where x and y represent the positions of pixels in the eye-captured image; f i (x, y) represents the illumination function of the image captured by the eye, and the illumination function varies with position; f r (x, y) represents the reflection function of the image captured by the eyeball, and the reflection function varies with position; Logarithmic transformation is performed on the spatial data of the eyeball image to obtain the spatial information: lnf(x,y)=lnf i (x,y)+lnf r (x,y); where lnf i (x,y) represents the illumination component and lnf r (x,y) represents the reflection component.

5. The method for detecting refractive medium abnormality according to claim 2, characterized in that: The filtering process on the frequency domain information to obtain the sum of the high frequency component and the suppressed low frequency component in the frequency domain information as the filtered frequency domain information includes: The frequency domain filter H is implemented using the following function mo (u, v) performs the filtering process on the frequency domain information: H mo (u,v)=(γ H -c L )H hp (u,v)+γ l ; H hp (u,v)=1-exp(-c(D 2 (u,v) / D0 2 )); Where: u, v represent the horizontal and vertical coordinates in the frequency domain coordinate system; H hp (u, v) is the high-pass filter function; D(u, v) represents the input frequency domain signal of the high-pass filter function, D0 represents the cutoff frequency of the high-pass filter function; c is the high-pass filter morphology control parameter; γ H represents the high frequency component weight, γ L Represents the weight of low-frequency components.

6. The method for detecting abnormality of a refractive medium according to any one of claims 1 to 5, characterized in that: The eyeball image and the pupil image samples are both images captured under an infrared light source.

7. The method for detecting refractive medium abnormality according to any one of claims 1 to 5, characterized in that: The determining, according to the position and area ratio of the abnormal pupil area in the pupil image, a refractive medium abnormality detection result of the target to be measured comprises: Obtain the center point position (x0, y0) and area A0 of the pupil image; Obtaining the center point position (xi, yi) and area Ai of the pupil abnormality area; The area ratio is calculated as follows: R = Ai / A0; The distance between the center point of the pupil abnormality area and the center point of the pupil image is calculated: The abnormal grade of the refractive medium is calculated as follows: G=k1×R+k2 / S; wherein K1 and K2 are preset weight coefficients.

8. A computer-readable storage medium, characterized in that: The storage medium stores program information, and the computer reads the program information and executes the steps of the refractive medium abnormality detection method according to any one of claims 1 to 7.

9. A computer program product, characterized in that The method comprises a computer program / instruction, wherein when the computer program / instruction is executed by a processor, the steps of the refractive medium abnormality detection method according to any one of claims 1 to 7 are implemented.

10. An electronic device comprising a memory, a processor and a computer program stored in the memory, characterized in that: The processor executes the computer program to implement the steps of the refractive medium abnormality detection method according to any one of claims 1 to 7.

Images