Imaging method of fundus camera and portable fundus camera

Abstract

The invention discloses an imaging method of a fundus camera and a portable fundus camera, and relates to the field of ophthalmology detection, and the method comprises the steps: collecting an eye image of a target object, and determining the identity information of the target object according to the eye image; acquiring historical detection data of the target object according to the identity information; when it is determined that the eyes of the target object are in a preset detection action according to the eye image, obtaining an infrared preview image; determining a first focal length of the target object under the infrared light irradiation condition according to the infrared preview image; determining a second focal length of the target object under the condition of white light irradiation according to the first focal length and historical detection data; and acquiring an eye fundus image of the target object by taking the second focal length as focusing setting information. By implementing the method, the focusing and image acquisition efficiency of the fundus camera can be improved.

Description

Technical Field

[0001] The present application relates to the field of ophthalmic detection, and in particular to an imaging method of a fundus camera and a portable fundus camera. Background Art

[0002] Fundus examination is an indispensable and important means of ophthalmic diagnosis. It can directly observe the fundus structures such as the retina, optic disc, macula and retinal blood vessels, and provide key information for the diagnosis of various eye diseases. With the continuous advancement of medical technology, fundus cameras have been widely used in ophthalmic clinical practice as a non-invasive examination tool. High-quality and efficient fundus images not only help doctors make accurate diagnoses, but also provide important basis for tracking and managing patients' conditions.

[0003] At present, the commonly used fundus camera imaging method mainly adopts digital imaging technology. During the imaging process, the operator first adjusts the position and focus of the camera to align it with the patient's pupil. Then, the fundus is illuminated by the camera's light source, and the reflected light is captured by a digital sensor. The captured image is then digitally processed to form a fundus image that can be used by doctors for diagnosis.

[0004] However, the related technology requires a long time for focusing and image acquisition, and has low detection efficiency, which causes eye fatigue in patients and affects the image quality. Summary of the invention

[0005] The present application provides an imaging method of a fundus camera and a portable fundus camera, which are used to improve the focusing and image acquisition efficiency of the fundus camera.

[0006] In a first aspect, the present application provides an imaging method for a fundus camera, which is applied to the fundus camera, the method comprising: acquiring an eye image of a target object, and determining identity information of the target object based on the eye image; acquiring historical detection data of the target object based on the identity information; acquiring an infrared preview image when it is determined based on the eye image that the eye of the target object is in a preset detection action; determining a first focal length of the target object under infrared light irradiation based on the infrared preview image; determining a second focal length of the target object under white light irradiation based on the first focal length and the historical detection data; and acquiring a fundus image of the target object using the second focal length as focusing setting information.

[0007] In the above embodiment, the fundus camera can obtain historical detection data according to the identity information of the target object, and predict the second focal length under white light irradiation in combination with the first focal length determined by the infrared preview image. The focal length prediction is performed based on the historical data and the infrared preview, which shortens the focusing time during white light imaging and improves the acquisition efficiency of the fundus image. At the same time, since the white light exposure time is reduced, the stimulation to the subject's eyes is also reduced, and the subject's comfort is improved.

[0008] In combination with some embodiments of the first aspect, in some embodiments, there are multiple infrared preview images, each infrared preview image corresponds to a focal length position; according to the infrared preview image, determining the first focal length of the target object under infrared light irradiation, specifically including: extracting edge features of the infrared preview image, and determining a corresponding clarity evaluation index according to the edge features; when the clarity evaluation index reaches a preset clarity threshold, recording the corresponding target focal length position; and calculating the first focal length according to the optical system parameters of the fundus camera and the target focal length position.

[0009] In the above embodiment, when acquiring the infrared preview image, the fundus camera selects the optimal target focal length position by analyzing the edge features and clarity of the image at multiple focal length positions, and calculates the first focal length in combination with the optical system parameters. Through multiple sampling analyses, the accuracy of focal length determination in the infrared preview stage is improved.

[0010] In combination with some embodiments of the first aspect, in some embodiments, the second focal length of the target object under white light irradiation is determined based on the first focal length and historical detection data, specifically including: extracting multiple detection records of the target object from the historical detection data; the detection records contain the infrared focal length of the target object under infrared light, the white light focal length under white light, and the historical refractive power during detection; calculating the focus deviation value between the infrared focal length and the white light focal length; establishing a deviation prediction model based on the focus deviation value and the corresponding historical refractive power; inputting the first focal length and the current refractive power of the target object into the deviation prediction model to obtain a predicted deviation value; and calculating the second focal length based on the first focal length and the predicted deviation value.

[0011] In the above embodiment, the fundus camera uses the infrared focal length, white light focal length and refractive power information in the historical detection data to establish a deviation prediction model, which can accurately predict the deviation value of the white light focal length based on the current infrared focal length and the refractive power of the target object. It utilizes individualized historical information to improve the accuracy of white light focal length prediction.

[0012] In combination with some embodiments of the first aspect, in some embodiments, after the step of acquiring a fundus image of the target object using the second focal length as focus setting information, the method further includes: identifying overexposed areas and underexposed areas in the fundus image; using a tone mapping algorithm to process the overexposed areas and underexposed areas respectively to obtain a tone optimized area; and restoring the tone optimized area to the fundus image to obtain a fundus optimized image.

[0013] In the above embodiment, after acquiring the fundus image, the fundus camera effectively improves the overall quality of the fundus image by identifying over-exposed and under-exposed areas and optimizing the processing using a tone mapping algorithm. By balancing the brightness and contrast of different areas, the fundus structure is clearer, which facilitates the doctor's accurate diagnosis.

[0014] In combination with some embodiments of the first aspect, in some embodiments, identifying overexposed areas and underexposed areas in a fundus image specifically includes: determining the HSV color space corresponding to the fundus image; calculating the distribution histogram of the V channel pixel values ​​in the HSV color space; determining a high brightness threshold and a low brightness threshold based on the distribution histogram; identifying an area where the V channel pixel value is higher than the high brightness threshold as an overexposed area; and identifying an area where the V channel pixel value is lower than the low brightness threshold as an underexposed area.

[0015] In the above embodiment, the fundus camera uses the V channel pixel value distribution of the HSV color space to determine the high and low brightness thresholds, thereby achieving accurate identification of abnormal exposure areas. The analysis is based on the color space and is more sensitive and reliable than the traditional grayscale threshold method, and can adapt to fundus images under different lighting conditions.

[0016] In combination with some embodiments of the first aspect, in some embodiments, when it is determined based on the eye image that the eye of the target object is in a preset detection action, before the step of obtaining the infrared preview image, the method also includes: obtaining the left-mirror eye image and the right-mirror eye image captured by the binocular camera module of the fundus camera; calculating the initial distance between the target object's eye and the binocular camera module based on the image deviation position of the preset reference object in the left-mirror eye image and the right-mirror eye image; and determining the initial focal length of the target object when performing infrared preview based on the initial distance.

[0017] In the above embodiment, the fundus camera uses a binocular camera module to obtain left and right eye images, calculates the initial distance between the target object's eyes and the camera by analyzing the image deviation position of the reference object, and determines the initial focal length setting for the infrared preview stage through stereo vision ranging, thereby shortening the focusing time.

[0018] In combination with some embodiments of the first aspect, in some embodiments, after the step of acquiring a fundus image of the target object using the second focal length as the focusing setting information, the method further includes: performing feature analysis on the fundus structure in the fundus image to obtain a lesion detection result; determining a pathological evolution indication of the lesion detection result in combination with historical detection data; and generating a detection report based on the pathological evolution indication.

[0019] In the above embodiment, after acquiring the fundus image, the fundus camera performs feature analysis on the fundus structure and generates a test report containing indications of pathological evolution in combination with historical test data, thereby improving the doctor's diagnostic efficiency and facilitating long-term tracking and management of the disease.

[0020] In a second aspect, an embodiment of the present application provides a fundus camera, comprising: one or more processors and a memory; the memory is coupled to the one or more processors, the memory is used to store computer program code, the computer program code includes computer instructions, and the one or more processors call the computer instructions to enable the fundus camera to perform the method described in the first aspect and any possible implementation of the first aspect.

[0021] In a third aspect, an embodiment of the present application provides a computer program product comprising instructions, which, when executed on a fundus camera, enables the fundus camera to execute the method described in the first aspect and any possible implementation of the first aspect.

[0022] In a fourth aspect, an embodiment of the present application provides a computer-readable storage medium, comprising instructions, which, when executed on a fundus camera, enable the fundus camera to execute the method described in the first aspect and any possible implementation of the first aspect.

[0023] It can be understood that the fundus camera provided in the second aspect, the computer program product provided in the third aspect, and the computer storage medium provided in the fourth aspect are all used to execute the method provided in the embodiment of the present application. Therefore, the beneficial effects that can be achieved can refer to the beneficial effects in the corresponding method, which will not be repeated here.

[0024] One or more technical solutions provided in the embodiments of the present application have at least the following technical effects or advantages: 1. Due to the adoption of a focal length prediction method based on historical data and infrared preview, the optimal focal length under white light irradiation can be determined quickly and accurately, effectively solving the problem of long focusing time and low efficiency of white light imaging in the prior art. By obtaining the identity information and historical detection data of the target object, the fundus camera establishes an individualized focal length prediction basis, and uses the correspondence between the infrared focal length and the white light focal length in the historical data to calculate the focal length, thereby improving the detection efficiency and accuracy.

[0025] 2. Due to the use of the abnormal exposure area recognition method based on the HSV color space and the tone mapping optimization algorithm, it is possible to accurately locate and improve the uneven exposure problem in the fundus image, effectively solving the problems of unstable fundus image quality and loss of local details in the existing technology, thereby achieving high-quality and information-rich fundus image output.

[0026] 3. Due to the use of the initial distance measurement and initial focal length setting method based on the binocular camera, the appropriate starting focal length can be quickly determined in the infrared preview stage, which effectively solves the problem of blind and time-consuming initial focusing in the prior art, thereby realizing a faster and more accurate fundus image acquisition process. The eye image is captured by the left and right lenses at the same time, and the parallax principle is used to calculate the precise distance between the eye and the camera. It is suitable for various inspection scenarios and realizes the improvement of the adaptability of application scenarios. BRIEF DESCRIPTION OF THE DRAWINGS

[0027] Figure 1 is a flow chart of an imaging method of a fundus camera in an embodiment of the present application; Figure 2 is another flowchart of the imaging method of the fundus camera in an embodiment of the present application; Figure 3 It is a schematic diagram of the structure of a physical device of a fundus camera in an embodiment of the present application. DETAILED DESCRIPTION

[0028] The terms used in the following embodiments of the present application are only for the purpose of describing specific embodiments, and are not intended to be used as limitations to the present application. As used in the specification of the present application, the singular expressions "one", "a kind of", "above", "the" and "this" are intended to also include plural expressions, unless there is a clear indication to the contrary in the context. It should also be understood that the term "and / or" used in the present application refers to any or all possible combinations comprising one or more of the listed items.

[0029] In the following, the terms "first" and "second" are used for descriptive purposes only and are not to be understood as suggesting or implying relative importance or implicitly indicating the number of the indicated technical features. Thus, a feature defined as "first" or "second" may explicitly or implicitly include one or more of the features, and in the description of the embodiments of the present application, unless otherwise specified, "plurality" means two or more.

[0030] See also Figure 1 , which is a flow chart of the imaging method of the fundus camera in the embodiment of the present application.

[0031] S101, collecting an eye image of a target object, and determining identity information of the target object according to the eye image.

[0032] The target object refers to a patient or examinee who needs to undergo fundus examination. The eye image refers to a digital image containing the eye area of ​​the target object acquired by the front camera of the fundus camera or other image acquisition device. The identity information is used to represent the personal identification information of the target object, such as name, ID number or other unique identifier.

[0033] Specifically, when the target object sits in front of the fundus camera and is ready to be examined, the fundus camera will start the image acquisition module. The fundus camera first acquires the eye image of the target object, including the entire face or a partial eye area. The fundus camera then uses image processing and face recognition technology to analyze the acquired eye image and extract key features such as iris texture, corneal shape, etc. The fundus camera compares these features with the pre-stored patient database to determine the identity information of the target object.

[0034] S102: Obtain historical detection data of the target object according to the identity information.

[0035] Among them, historical test data refers to various relevant data records generated when the target subject underwent fundus examination in the past, including previous fundus images, diagnosis results, medication records, vision changes, etc.

[0036] Specifically, after determining the identity information of the target object, the fundus camera will be connected to the patient data management system or electronic medical record system. The fundus camera uses the obtained identity information as a search keyword to query the relevant database. The fundus camera will not only obtain the most recent examination record, but also retrieve all available historical data for comprehensive analysis. The acquired data may include past fundus images, refractive power change records, intraocular pressure measurement results, vision test data, etc. The fundus camera will preliminarily process and organize these data in preparation for subsequent analysis and use.

[0037] In some embodiments, the acquisition and processing of historical detection data can be achieved in a variety of ways: Optionally, the fundus camera can use a distributed database query method. First, the fundus camera searches for relevant records in the local database based on the identity information; if the local data is incomplete, it sends an encrypted query request to the database of other medical institutions; finally, the fundus camera integrates and deduplicates the data obtained from different sources to form a complete historical detection data set. Optionally, the fundus camera can use an intelligent data analysis method. First, the fundus camera extracts the original historical data from the database; then, the data is cleaned and standardized using a machine learning algorithm to remove outliers and redundant information; finally, the fundus camera performs feature extraction and trend analysis on the processed data to generate a summary report containing key information. It is understandable that other methods can also be used to achieve the acquisition and processing of historical detection data, such as using blockchain technology to ensure the security and integrity of the data, or using natural language processing technology to extract useful information from unstructured medical records, which is not limited here.

[0038] S103: When it is determined according to the eye image that the eye of the target object is in a preset detection action, an infrared preview image is acquired.

[0039] The preset detection action refers to the target object's eyes being in a specific state or position suitable for fundus examination, including eyes opened to a suitable size, sight aligned in a specific direction, etc. The infrared preview image refers to the image collected by irradiating the target object's fundus with an infrared light source, which is usually used for preliminary evaluation and focus adjustment.

[0040] Specifically, the fundus camera continuously monitors the eye images collected in real time by the fundus camera. When the fundus camera detects that the eye state of the target object meets the preset detection conditions, such as the eyes are fully open and the pupil is in the proper position, the infrared light source will be triggered. The fundus camera then begins to collect a series of infrared preview images. Although the infrared preview image is not as clear as the white light image, it can provide sufficient information for preliminary assessment of the fundus condition and adjustment of the focus, and it is less irritating to the patient's eyes. The fundus camera continuously collects multiple infrared preview images to ensure the best quality images for subsequent analysis.

[0041] In some embodiments, the confirmation of the preset detection action and the acquisition of the infrared preview image can be achieved in a variety of ways: Optionally, the fundus camera can use real-time eye tracking technology. First, the fundus camera captures the eye movement of the target object through a high-speed camera; then, the eye position and pupil size are analyzed in real time using a computer vision algorithm; finally, when it is detected that the eye is stable and the pupil is at a suitable size, the infrared light source is automatically triggered and the preview image is collected. Optionally, the fundus camera can use a method combining voice guidance and image feedback. First, the fundus camera guides the target object to adjust the eye position through voice instructions; then, the eye image is displayed in real time and the ideal position is indicated using visual markers; finally, when the image analysis algorithm confirms that the eye position is correct, the infrared preview image is automatically collected. It is understandable that other methods can also be used to achieve the confirmation of the preset detection action and the acquisition of the infrared preview image, such as using artificial intelligence to assist in determining the best shooting time, or combining virtual reality technology to provide more intuitive visual guidance, which is not limited here.

[0042] S104: Determine a first focal length of the target object under infrared light irradiation according to the infrared preview image.

[0043] The first focal length indicates the distance from the lens to the fundus that can obtain a clear fundus image under infrared light irradiation conditions.

[0044] Specifically, the fundus camera will quickly analyze a series of acquired infrared preview images. First, the fundus camera will use image processing algorithms to evaluate the clarity of each image, including edge detection, contrast analysis and other methods. Then, the fundus camera will select the best infrared preview image based on the clarity score. For the selected image, the fundus camera will further analyze the fundus structural features, such as the clarity of the optic disc, blood vessels, etc. Combining these analysis results with the optical parameters of the fundus camera, the fundus camera uses a specific algorithm to calculate the focal length value that can obtain the clearest infrared fundus image, that is, the first focal length.

[0045] In some embodiments, the determination of the first focal length can be achieved in a variety of ways: Optionally, the fundus camera can use an autofocus algorithm. First, the fundus camera quickly collects multiple infrared preview images at different focal lengths; then, the clarity score is calculated for each image, such as using the Laplace operator to evaluate image sharpness; finally, the focal length corresponding to the highest point of the clarity score is found through interpolation or curve fitting methods, and determined as the first focal length. Optionally, the fundus camera can use a deep learning method. First, a convolutional neural network model is trained using a large number of annotated fundus images, which can evaluate image clarity and predict the optimal focal length; then, the acquired infrared preview image is input into the trained model; finally, the model directly outputs the predicted optimal focal length value as the first focal length. It is understandable that other methods can also be used to determine the first focal length, such as optimizing focal length calculation in combination with wavefront aberration analysis technology, or using multispectral imaging technology to improve the accuracy of focal length estimation, which is not limited here.

[0046] S105 . Determine a second focal length of the target object under white light illumination according to the first focal length and historical detection data.

[0047] The second focal length refers to the distance from the lens to the fundus that can obtain a clear fundus image under white light illumination. The historical detection data refers to the target object's past fundus examination records, including previous focal length settings, diopter changes and other information.

[0048] Specifically, the fundus camera will first analyze the focal length records in the historical detection data, especially the difference in focal length under infrared light and white light. The fundus camera will establish a personalized focal length conversion model, taking into account factors such as the target object's eye characteristics, age, and refractive power changes. The fundus camera then inputs the first focal length just determined (the focal length under infrared light) into this model. The model predicts the possible best focal length under white light based on historical data and the current first focal length. The fundus camera will also take into account the current state of the target object, such as pupil size, intraocular pressure, etc., fine-tune the prediction results, and finally determine the second focal length.

[0049] In some embodiments, the second focal length can be determined in a variety of ways: Optionally, the fundus camera can use a statistical regression method. First, the corresponding relationship between the infrared focal length and the white light focal length is extracted from the historical data; then, a mapping function from the infrared focal length to the white light focal length is established using a multivariate linear regression or nonlinear regression model; finally, the current first focal length is input into the function to calculate the predicted second focal length. Optionally, the fundus camera can use a machine learning method. First, a neural network model is trained using historical data, and the input includes features such as infrared focal length, patient age, and refractive power, and the output is the white light focal length; then, the current first focal length and the relevant information of the target object are input into the trained model; finally, the predicted value output by the model is the second focal length. It is understandable that other methods can also be used to determine the second focal length, such as using a fuzzy logic control algorithm to dynamically adjust the focal length prediction, or combining optical coherence tomography (OCT) technology to improve the accuracy of focal length conversion, which is not limited here.

[0050] S106 , acquiring a fundus image of the target object using the second focal length as focus setting information.

[0051] The focus setting information refers to the parameters used to control the position of the fundus camera lens. The fundus image refers to a high-resolution digital image that clearly records the fundus structure of the target object under white light.

[0052] Specifically, the fundus camera will first convert the calculated second focal length value into a focus control parameter that can be recognized by the fundus camera optical system. Then, the fundus camera will accurately control the lens moving mechanism to adjust the lens to the position corresponding to the second focal length. After the adjustment is completed, the fundus camera will start the white light source to illuminate the fundus of the target object. After ensuring that the light intensity and uniformity meet the requirements, the fundus camera will trigger the image sensor to capture the fundus image. To ensure image quality, the fundus camera will take a fundus image, or take multiple images continuously, and select the clearest and best contrast one as the final fundus image. The whole process will be carried out quickly to reduce irritation to the target object's eyes.

[0053] In some embodiments, fundus images can be acquired in a variety of ways: Optionally, the fundus camera can use multi-exposure synthesis technology. First, the fundus camera takes the second focal length as the reference and quickly and continuously shoots multiple fundus images with different exposure times; then, these images are synthesized using a high dynamic range (HDR) algorithm; finally, the synthesized image is color balanced and detail enhanced to obtain the final high-quality fundus image. Optionally, the fundus camera can use real-time image quality assessment and adaptive adjustment methods. First, the fundus camera starts shooting at the second focal length and analyzes the image quality in real time; then, if it is detected that some areas of the image are not clear, the fundus camera will fine-tune the focal length and lighting parameters; finally, when the preset quality standard is reached, the final image is automatically captured and saved. It is understandable that other methods can also be used to acquire fundus images, such as using light field camera technology to achieve post-digital refocusing, or combining artificial intelligence to assist the fundus camera in real-time optimization of shooting parameters, which is not limited here.

[0054] In the above embodiment, the smart fundus camera can automatically adapt to the eye characteristics of different patients and optimize the imaging process. In practical applications, this technology can not only be used for routine fundus examinations, but can also be extended to various scenarios such as long-term disease monitoring. Figure 2 , is another flow chart of the imaging method of the fundus camera in an embodiment of the present application.

[0055] S201, collecting an eye image of a target object, and determining identity information of the target object according to the eye image.

[0056] Referring to step S101 , the fundus camera determines the identity information of the target object.

[0057] S202: Obtain historical detection data of the target object according to the identity information.

[0058] Referring to step S102 , the fundus camera acquires historical detection data of the target object.

[0059] S203: When it is determined according to the eye image that the eye of the target object is in a preset detection action, an infrared preview image is acquired.

[0060] Referring to step S103 , the fundus camera acquires an infrared preview image.

[0061] In some embodiments, the fundus camera obtains the left-mirror eye image and the right-mirror eye image captured by the binocular camera module of the fundus camera; based on the image deviation position of a preset reference object in the left-mirror eye image and the right-mirror eye image, the initial distance between the target object's eye and the binocular camera module is calculated; based on the initial distance, the initial focal length of the target object during infrared preview is determined.

[0062] Among them, the binocular camera module refers to an image acquisition system composed of two camera lenses placed side by side. The left-lens eye image and the right-lens eye image represent the eye images of the target object taken simultaneously by the left and right lenses respectively. The preset reference object refers to a specific marking point or structure in the image used for distance measurement, such as the corner of the eye or the center of the pupil. The image deviation position is used to indicate the relative position difference of the same reference object in the left and right images. The initial distance refers to the straight-line distance from the eye of the target object to the binocular camera module. The initial focal length represents the infrared preview starting focal length value preset based on the initial distance.

[0063] Specifically, before the fundus camera starts the fundus examination, it will first start the binocular camera module. The binocular camera module will capture two left and right eye images at the same time. The fundus camera then analyzes the two images and identifies predefined reference objects, such as the corners of the eyes or the edges of the pupils. The fundus camera then accurately measures the difference in pixel coordinates of these reference objects in the left and right images. Using this difference value, combined with the optical parameters of the binocular camera (such as baseline distance and focal length), the fundus camera calculates the exact distance from the target object's eye to the camera through the principle of triangulation. Finally, the fundus camera substitutes this distance value into the preset optical formula to calculate a suitable initial focal length value as the starting setting for the infrared preview stage.

[0064] S204: extract edge features of the infrared preview image, and determine corresponding clarity evaluation indicators according to the edge features.

[0065] Among them, edge features represent the sharpness of the outline and details of objects in the image. The clarity evaluation index refers to the numerical standard used to quantify the clarity of the image.

[0066] Specifically, the fundus camera starts processing after acquiring the infrared preview image. First, the fundus camera preprocesses the image, such as denoising and contrast enhancement. Then, the fundus camera applies edge detection algorithms, such as Sobel or Canny operators, to extract edge information in the image. The fundus camera focuses on the edge features of key structures such as fundus blood vessels and optic disc edges. Next, the fundus camera calculates the intensity and distribution of these edges to derive a numerical indicator that reflects the overall image clarity. This indicator can be the average gradient of edge pixels, the energy ratio of high-frequency components, etc.

[0067] In some embodiments, edge feature extraction and clarity evaluation can be achieved in a variety of ways: Optionally, the fundus camera can use a multi-scale edge analysis method. First, the fundus camera performs a multi-layer pyramid decomposition on the image; then, the edge detection operator is applied at each scale; finally, the edge information of each scale is integrated, and the weighted average is calculated to obtain the final clarity index. Optionally, the fundus camera can adopt a clarity evaluation method based on machine learning. First, a convolutional neural network model is trained using a large number of annotated fundus images; then, the current infrared preview image is input into the model; finally, the model directly outputs a score representing the image clarity. It is understandable that other methods can also be used to achieve edge feature extraction and clarity evaluation, such as combining wavelet transform to analyze the frequency domain characteristics of the image, or using phase consistency analysis to evaluate the structural clarity of the image, which is not limited here.

[0068] S205. When the clarity evaluation index reaches a preset clarity threshold, the corresponding target focal length position is recorded.

[0069] The preset clarity threshold refers to the minimum acceptable value of the clarity evaluation index predefined by the fundus camera. The target focal length position indicates the distance setting of the current lens relative to the fundus.

[0070] Specifically, the fundus camera will continuously monitor the clarity evaluation index calculated in the previous step. When it is detected that the index value reaches or exceeds the preset clarity threshold for the first time, the fundus camera will trigger the recording operation. This preset threshold is usually determined based on a large amount of experimental data and expert experience. The fundus camera will accurately record the position parameters of the lens at this time, which may include the step value of the focus motor, the zoom ratio of the optical fundus camera, etc. These data will be temporarily stored in the cache of the fundus camera for subsequent quick access and processing.

[0071] In some embodiments, the clear threshold judgment and target focal position recording can be achieved in a variety of ways: Optionally, the fundus camera can use a dynamic threshold adjustment method. First, the fundus camera starts the evaluation according to the initial preset threshold; then, as the evaluation process proceeds, the fundus camera dynamically adjusts the threshold according to the changing trend of the image quality; finally, when the clarity index stably exceeds the adjusted threshold, the corresponding focal position is recorded. Optionally, the fundus camera can adopt a multi-point sampling comparison method. First, the fundus camera quickly collects images of multiple focal positions within a certain range; then, calculates the clarity evaluation index of each position; finally, selects the position with the highest clarity and exceeds the preset threshold as the target focal position and records it. It can be understood that other methods can also be used to achieve clear threshold judgment and target focal position recording, such as using fuzzy logic control algorithm to optimize the threshold judgment process, or combining eye tracking technology to adjust the focal recording strategy in real time, which is not limited here.

[0072] S206 : Calculate a first focal length according to the optical system parameters of the fundus camera and the target focal length position.

[0073] Among them, optical system parameters refer to a series of values ​​that describe the optical performance of the fundus camera, including lens focal length, aperture size, sensor size, etc. The first focal length represents the ideal distance from the lens to the fundus that can obtain a clear fundus image under infrared light irradiation conditions.

[0074] Specifically, after recording the target focal position, the fundus camera will start the focal length calculation process. First, the fundus camera will call the pre-stored fundus camera optical system parameters, which are usually determined during camera calibration. Then, the fundus camera combines the data of the target focal position with these optical parameters and substitutes them into a specific optical calculation formula. This formula takes into account factors such as the magnification of the lens and the length of the optical path. Through complex calculations, the fundus camera converts the mechanical position parameters into actual physical focal length values, namely the first focal length, which represents the ideal distance from the lens to the fundus when obtaining a clear fundus image under infrared light.

[0075] S207 . Determine a second focal length of the target object under white light illumination according to the first focal length and historical detection data.

[0076] Referring to step S105 , the fundus camera determines a second focal length of the target object.

[0077] In some embodiments, the fundus camera extracts multiple detection records of the target object from historical detection data; the detection records include the infrared focal length of the target object under infrared light, the white light focal length under white light, and the historical refractive power during detection; the focus deviation value between the infrared focal length and the white light focal length is calculated; a deviation prediction model is established based on the focus deviation value and the corresponding historical refractive power; the first focal length and the current refractive power of the target object are input into the deviation prediction model to obtain a predicted deviation value; and the second focal length is calculated based on the first focal length and the predicted deviation value.

[0078] Among them, infrared focal length and white light focal length refer to the focal length of the lens when obtaining a clear image under different light sources. Diopter is a measure of the refractive power of the eye. The focus deviation value refers to the difference between the infrared focal length and the white light focal length.

[0079] Specifically, after determining the first focal length, the fundus camera will access the database to retrieve the historical detection data of the target object. The fundus camera will screen out multiple records containing complete focal length and refractive power information, usually selecting the most recent several inspection results. Then, the fundus camera will calculate the difference between the infrared focal length and the white light focal length in each record to obtain a series of historical focus deviation values. Next, the fundus camera uses these deviation values ​​and the corresponding historical refractive power data to build a deviation prediction model through regression analysis or machine learning algorithms. After the model is established, the fundus camera will input the currently obtained first focal length (i.e., infrared focal length) and the latest refractive power data of the target object, and the model will output a predicted deviation value. Finally, the fundus camera adds the first focal length to this predicted deviation value to obtain the predicted white light focal length, i.e., the second focal length.

[0080] In some embodiments, the deviation prediction and the calculation of the second focal length can be achieved in a variety of ways: Optionally, the fundus camera can use a multivariate linear regression method. First, the fundus camera uses the diopter and infrared focal length in the historical data as independent variables and the focus deviation value as the dependent variable; then, the least squares method is used to fit the linear regression equation; finally, the current first focal length and diopter are substituted into the equation to calculate the predicted deviation value, and then the second focal length is determined. Optionally, the fundus camera can use a support vector regression (SVR) algorithm. First, the fundus camera performs feature engineering on the historical data to extract relevant features; then, the hyperparameters of the SVR are optimized using methods such as grid search, and the deviation prediction model is trained; finally, the current data is input into the trained SVR model to obtain the predicted deviation value and calculate the second focal length. It can be understood that other methods can also be used to achieve deviation prediction and the calculation of the second focal length, such as using a neural network to construct a nonlinear prediction model, or combining the Bayesian optimization method to dynamically adjust the prediction strategy, which is not limited here.

[0081] S208 , acquiring a fundus image of the target object using the second focal length as focus setting information.

[0082] Referring to step S106 , the fundus camera acquires a fundus image of the target object.

[0083] In some embodiments, the fundus camera performs feature analysis on the fundus structure in the fundus image to obtain lesion detection results; combines historical detection data to determine the pathological evolution indication of the lesion detection results; and generates a detection report based on the pathological evolution indication.

[0084] Among them, fundus structures refer to various anatomical features on the retina, such as the optic disc, macula, blood vessels, etc. Feature analysis refers to the quantitative or qualitative evaluation of the morphology, color, distribution, etc. of these structures. Lesion detection results refer to the areas or phenomena with pathological changes found through analysis. Pathological evolution indicators are used to indicate the trend of lesion status changes over time.

[0085] Specifically, after acquiring high-quality fundus images, the fundus camera will start the automatic analysis process. First, the fundus camera will use image segmentation algorithms to identify and locate the main fundus structures. Then, the fundus camera will perform detailed feature extraction on each structure, such as measuring the size of the optic disc, analyzing the direction of blood vessels, and evaluating the integrity of the macular area. The fundus camera will compare these features with normal reference values ​​to identify possible abnormal areas. Next, the fundus camera will access historical detection data, compare the current lesion detection results with past records, and analyze the development trend of the lesions. Based on this longitudinal comparison, the fundus camera will generate pathological evolution indications to describe the stability, progression, or improvement of the lesions. Finally, the fundus camera will integrate information such as current test results, historical comparative analysis, and pathological evolution predictions to automatically generate a structured test report.

[0086] In some embodiments, fundus lesion detection and report generation can be achieved in a variety of ways: Optionally, the fundus camera can use a lesion detection method based on deep learning. First, the fundus camera uses a pre-trained convolutional neural network to extract features from the fundus image; then, the potential lesion area is located through the target detection algorithm; finally, the classification model is used to make a specific lesion type judgment for each detected area to generate detailed detection results. Optionally, the fundus camera can adopt a rule-based expert fundus camera method. First, the fundus camera builds a detailed diagnostic rule library based on the professional knowledge text file of the ophthalmologist; then, the extracted fundus features are input into this rule fundus camera for reasoning; finally, the fundus camera gives a lesion diagnosis opinion based on the reasoning result, and generates a personalized detection report based on a preset report template. It is understandable that other methods can also be used to achieve fundus lesion detection and report generation, such as combining multimodal image fusion technology to improve diagnostic accuracy, or using natural language processing technology to generate a more understandable report description, which is not limited here.

[0087] S209: Identify overexposed areas and underexposed areas in the fundus image.

[0088] Among them, the overexposed area refers to the part of the image where the brightness is too high and the details are lost. The underexposed area refers to the image area where the brightness is too low and the details are not clear.

[0089] Specifically, the fundus camera starts the analysis process after acquiring the fundus image. First, the fundus camera converts the color fundus image into a grayscale image for brightness analysis. Then, the fundus camera calculates the brightness histogram of the entire image to understand the overall distribution of pixel brightness. Based on this distribution, the fundus camera sets thresholds for high brightness and low brightness. Next, the fundus camera scans the entire image pixel by pixel, marking areas where the brightness exceeds the high threshold as overexposed and areas where the brightness is below the low threshold as underexposed. The fundus camera can also take into account local contrast to more accurately identify problem areas.

[0090] In some embodiments, the fundus camera determines the HSV color space corresponding to the fundus image; calculates the distribution histogram of the V channel pixel values ​​in the HSV color space; determines the high brightness threshold and the low brightness threshold based on the distribution histogram; identifies the area where the V channel pixel value is higher than the high brightness threshold as an overexposed area; and identifies the area where the V channel pixel value is lower than the low brightness threshold as an underexposed area.

[0091] Among them, HSV color space refers to a three-dimensional color model with hue, saturation and value as coordinates. The V channel represents the component representing brightness in the HSV model. The distribution histogram refers to a statistical chart that describes the distribution of pixel values ​​at different brightness levels. The high brightness threshold and low brightness threshold represent the critical values ​​for judging overexposure and underexposure, respectively. The overexposed area refers to the part of the image with abnormally high brightness, and the underexposed area refers to the part with abnormally low brightness.

[0092] Specifically, after acquiring the fundus image, the fundus camera will first convert the image from the RGB color space to the HSV color space. This conversion helps to better separate the brightness information. Then, the fundus camera will focus on the V channel of the HSV space and calculate the value distribution of all pixels on the V channel. The fundus camera will generate a histogram showing the number of pixels at different brightness levels. Based on this histogram, the fundus camera will use statistical methods, such as percentiles or standard deviations, to determine the thresholds for high brightness and low brightness. After determining the threshold, the fundus camera will scan the V channel of the entire image pixel by pixel. For pixels whose V values ​​exceed the high brightness threshold, the fundus camera will mark them as overexposed; for pixels whose V values ​​are lower than the low brightness threshold, the fundus camera will mark them as underexposed. Finally, the fundus camera will generate a mask map of the exposure problem area for subsequent image optimization processing.

[0093] In some embodiments, the identification of abnormal exposure areas can be achieved in a variety of ways: Optionally, the fundus camera can use an adaptive threshold method. First, the fundus camera calculates the global mean and standard deviation of the V channel; then, the mean plus or minus several standard deviations is used as the initial high and low thresholds; finally, the fundus camera dynamically adjusts these thresholds in the local area to adapt to the brightness characteristics of different areas, so as to more accurately identify the abnormal exposure areas. Optionally, the fundus camera can use a clustering-based method. First, the fundus camera performs K-means clustering on the pixel values ​​of the V channel, usually selecting 3 to 5 categories; then, the center value of each category is analyzed, and the highest and lowest categories are regarded as potential over-exposure and under-exposure areas, respectively; finally, the fundus camera combines the spatial continuity information to determine the boundaries of these areas to obtain the final abnormal exposure area. It can be understood that other methods can also be used to achieve the identification of abnormal exposure areas, such as using local contrast analysis methods to improve the robustness of recognition, or combining deep learning models to directly learn the characteristics of abnormal exposure from images, which are not limited here.

[0094] S210 , using a tone mapping algorithm to process the overexposed area and the underexposed area respectively to obtain a tone optimized area.

[0095] The tone mapping algorithm is a mathematical method used to adjust the brightness and contrast of an image in order to expand the dynamic range and enhance details. The tone optimized area refers to the area of ​​the image where the visual effect is improved after the algorithm adjustment.

[0096] Specifically, after identifying areas with abnormal exposure, the fundus camera will adopt corresponding tone mapping strategies for different types of areas. For overexposed areas, the fundus camera will use an algorithm to compress highlights, reduce brightness and try to restore lost details. For underexposed areas, the fundus camera will apply an algorithm to enhance shadows, increase brightness and enhance contrast. These algorithms usually take into account the perceptual characteristics of the human eye to ensure that the processed images look natural and harmonious. The fundus camera will use a local adaptive method to dynamically adjust the mapping parameters according to the characteristics of the surrounding area to achieve the best optimization effect.

[0097] In some embodiments, tone mapping and optimization can be achieved in a variety of ways: Optionally, the fundus camera can use a multi-scale fusion method. First, the fundus camera decomposes the original image at multiple scales; then, the tone mapping algorithm is applied at different scales; finally, the processed results of each scale are weighted fused to obtain the final tone optimized area. Optionally, the fundus camera can adopt a tone optimization method based on deep learning. First, a generative adversarial network (GAN) is trained using a large number of high-quality fundus images; then, the identified abnormal exposure area is input into the trained GAN; finally, the GAN generates an optimized area image with better visual effect. It can be understood that other methods can also be used to achieve tone mapping and optimization, such as using an adaptive tone mapping algorithm based on the anatomical structure of the retina, or combining the human eye visual perception model to optimize the processing effect, which is not limited here.

[0098] S211 , restoring the tone optimized area to the fundus image to obtain a fundus optimized image.

[0099] Among them, the fundus optimized image represents the final high-quality fundus image after tone adjustment and detail enhancement.

[0100] Specifically, after the fundus camera completes the tone optimization of the local area, it will start the image fusion process. First, the fundus camera will create a blank layer of the same size as the original fundus image. Then, the fundus camera will accurately locate the optimized area to the corresponding position on this layer. In order to avoid obvious boundaries between the processed and unprocessed areas, the fundus camera will apply a gradient blending algorithm to the edge area. This includes the use of feathering or alpha channel blending technology. The fundus camera will also perform a global color balance adjustment to ensure that the tone of the overall image is coordinated and consistent. Finally, the fundus camera merges the fused layer with the original image to generate the final fundus optimized image.

[0101] In some embodiments, the restoration of the tone optimized area and the generation of the final image can be achieved in a variety of ways: Optionally, the fundus camera can use Poisson image editing technology. First, the fundus camera calculates the gradient field of the optimized area and the original image; then, the Poisson equation is used to solve the best fusion result; finally, the solution result is combined with the original image to obtain a seamlessly fused fundus optimized image. Optionally, the fundus camera can use a multi-resolution spline interpolation method. First, the fundus camera decomposes the original image and the optimized area at multiple scales; then, the spline interpolation algorithm is used at each scale for fusion; finally, the fusion results of each scale are reconstructed to obtain the final fundus optimized image. It can be understood that other methods can also be used to achieve the restoration of the tone optimized area and the generation of the final image, such as using deep learning-based image restoration technology to achieve a more natural fusion effect, or combining retinal vascular structure information to guide the fusion process to maintain the continuity of the anatomical structure, which is not limited here.

[0102] In the embodiment of the present application, due to the use of an intelligent adaptive focus fundus camera based on historical data and infrared preview, combined with real-time image processing and lesion analysis technology, it is possible to quickly and accurately obtain high-quality fundus images and provide in-depth pathological analysis, effectively solving the problems of complex operation, time-consuming inspection, unstable image quality, and low analysis efficiency of traditional fundus cameras, thereby achieving efficient and high-quality fundus examination and diagnosis. It not only improves the work efficiency of ophthalmologists and reduces the discomfort of patients, but also provides effective support for the early detection and long-term monitoring of fundus diseases.

[0103] The fundus camera in the embodiment of the present invention is described below from the perspective of hardware processing. Figure 3 , is a schematic diagram of a physical device structure of a fundus camera in an embodiment of the present application.

[0104] It should be noted that Figure 3 The structure of the fundus camera shown is only an example and should not bring any limitation to the functions and scope of use of the embodiments of the present invention.

[0105] like Figure 3As shown, the fundus camera includes a central processing unit (CPU) 301, which can perform various appropriate actions and processes according to the program stored in the read-only memory (ROM) 302 or the program loaded from the storage part 308 to the random access memory (RAM) 303, such as executing the method described in the above embodiment. In the RAM 303, various programs and data required for system operation are also stored. The CPU 301, the ROM 302, and the RAM 303 are connected to each other through the bus 304. The input / output (I / O) interface 305 is also connected to the bus 304.

[0106] The following components are connected to the I / O interface 305: an input section 306 including an audio input device, a button switch, etc.; an output section 307 including a liquid crystal display (LCD) and an audio output device, an indicator light, etc.; a storage section 308 including a hard disk, etc.; and a communication section 309 including a network interface card such as a LAN (Local Area Network) card, a modem, etc. The communication section 309 performs communication processing via a network such as the Internet. A drive 310 is also connected to the I / O interface 305 as needed. A removable medium 311, such as a magnetic disk, an optical disk, a magneto-optical disk, a semiconductor memory, etc., is installed on the drive 310 as needed so that a computer program read therefrom is installed into the storage section 308 as needed.

[0107] In particular, according to an embodiment of the present invention, the process described above with reference to the flowchart can be implemented as a computer software program. For example, an embodiment of the present invention includes a computer program product, which includes a computer program carried on a computer-readable medium, and the computer program includes a computer program for executing the method shown in the flowchart. In such an embodiment, the computer program can be downloaded and installed from a network through the communication part 309, and / or installed from a removable medium 311. When the computer program is executed by the central processing unit (CPU) 301, various functions defined in the present invention are performed.

[0108] It should be noted that specific examples of computer-readable storage media may include, but are not limited to: an electrical connection with one or more wires, a portable computer disk, a hard disk, a random access memory (RAM), a read-only memory (ROM), an erasable programmable read-only memory (EPROM), a flash memory, an optical fiber, a portable compact disk read-only memory (CD-ROM), an optical storage device, a magnetic storage device, or any suitable combination thereof. In the present invention, a computer-readable storage medium may be any tangible medium containing or storing a program that may be used by or in combination with an instruction execution system, apparatus, or device.

[0109] The flowcharts and block diagrams in the accompanying drawings illustrate the possible architecture, functions and operations of the systems, methods and computer program products according to various embodiments of the present invention. Each box in the flowchart or block diagram may represent a module, a program segment, or a part of a code, and the above-mentioned module, program segment, or a part of a code contains one or more executable instructions for implementing the specified logical functions. It should also be noted that in some alternative implementations, the functions marked in the box may also occur in an order different from that marked in the accompanying drawings.

[0110] Specifically, the fundus camera of this embodiment includes a processor and a memory. The memory stores a computer program. When the computer program is executed by the processor, the imaging method of the fundus camera provided in the above embodiment is implemented.

[0111] As another aspect, the present invention further provides a computer-readable storage medium, which may be included in the fundus camera described in the above embodiment; or may exist independently without being assembled into the fundus camera. The above storage medium carries one or more computer programs, and when the above one or more computer programs are executed by a processor of the fundus camera, the fundus camera implements the imaging method of the fundus camera provided in the above embodiment.

[0112] As described above, the above embodiments are only used to illustrate the technical solutions of the present application, rather than to limit them. Although the present application has been described in detail with reference to the aforementioned embodiments, those skilled in the art should understand that they can still modify the technical solutions described in the aforementioned embodiments, or make equivalent replacements for some of the technical features therein. However, these modifications or replacements do not cause the essence of the corresponding technical solutions to deviate from the scope of the technical solutions of the embodiments of the present application.

[0113] As used in the above embodiments, the term "when..." may be interpreted to mean "if..." or "after..." or "in response to determining..." or "in response to detecting...", depending on the context. Similarly, the phrases "upon determining..." or "if (the stated condition or event) is detected" may be interpreted to mean "if determining..." or "in response to determining..." or "upon detecting (the stated condition or event)" or "in response to detecting (the stated condition or event)", depending on the context.

[0114] Those skilled in the art can understand that to implement all or part of the processes in the above-mentioned embodiments, the processes can be completed by computer programs to instruct related hardware, and the programs can be stored in computer-readable storage media. When the programs are executed, they can include the processes of the above-mentioned method embodiments. The aforementioned storage media include: ROM or random access memory RAM, magnetic disk or optical disk and other media that can store program codes.

Claims

1. An imaging method of a fundus camera, characterized in that: Applied to a fundus camera, the method comprises: Acquire an eye image of a target object, and determine identity information of the target object based on the eye image; Acquire historical detection data of the target object according to the identity information; When it is determined according to the eye image that the eye of the target object is in a preset detection action, acquiring an infrared preview image; Determining a first focal length of the target object under infrared light irradiation according to the infrared preview image; Determining a second focal length of the target object under white light illumination according to the first focal length and the historical detection data; The fundus image of the target object is acquired by taking the second focal length as focus setting information.

2. The method according to claim 1, characterized in that There are multiple infrared preview images, each of which corresponds to a focal length position; determining a first focal length of the target object under infrared light irradiation according to the infrared preview images specifically includes: Extracting edge features of the infrared preview image, and determining corresponding clarity evaluation indicators according to the edge features; When the clarity evaluation index reaches a preset clarity threshold, recording the corresponding target focal length position; A first focal length is calculated according to the optical system parameters of the fundus camera and the target focal length position.

3. The method according to claim 1, characterized in that The determining, according to the first focal length and the historical detection data, a second focal length of the target object under white light irradiation specifically includes: Extracting multiple detection records of the target object from the historical detection data; the detection records contain the infrared focal length of the target object under infrared light, the white light focal length under white light, and the historical refractive power during detection; Calculating a focus deviation value between the infrared focal length and the white light focal length; Establishing a deviation prediction model according to the focus deviation value and the corresponding historical diopter; Inputting the first focal length and the current diopter of the target object into a deviation prediction model to obtain a predicted deviation value; The second focal length is calculated based on the first focal length and the predicted deviation value.

4. The method according to claim 1, characterized in that After the step of acquiring the fundus image of the target object by using the second focal length as the focus setting information, the method further includes: identifying overexposed areas and underexposed areas in the fundus image; Using a tone mapping algorithm to process the overexposed area and the underexposed area respectively to obtain a tone optimized area; The tone optimized area is restored to the fundus image to obtain a fundus optimized image.

5. The method according to claim 4, characterized in that The identifying of overexposed areas and underexposed areas in the fundus image specifically includes: Determine the HSV color space corresponding to the fundus image; Calculate the distribution histogram of the V channel pixel values ​​in the HSV color space; Determine a high brightness threshold and a low brightness threshold according to the distribution histogram; Identify an area where the V channel pixel value is higher than the high brightness threshold as an overexposed area; Regions where the V channel pixel values ​​are lower than the low brightness threshold are identified as underexposed regions.

6. The method according to claim 1, characterized in that When it is determined according to the eye image that the eye of the target object is in a preset detection action, before the step of acquiring the infrared preview image, the method further includes: Acquire a left-lens eye image and a right-lens eye image captured by a binocular camera module of the fundus camera; Calculate the initial distance between the eye of the target object and the binocular camera module according to the image deviation position of the preset reference object in the left mirror eye image and the right mirror eye image; An initial focal length of the target object during infrared preview is determined according to the initial distance.

7. The method according to claim 1, characterized in that After the step of acquiring the fundus image of the target object by using the second focal length as the focus setting information, the method further includes: Performing feature analysis on the fundus structure in the fundus image to obtain a lesion detection result; Determining the pathological evolution indication of the lesion detection result in combination with the historical detection data; A test report is generated based on the pathological evolution indication.

8. A fundus camera, characterized in that: The fundus camera comprises: one or more processors and a memory; the memory is coupled to the one or more processors, the memory is used to store computer program code, the computer program code comprises computer instructions, and the one or more processors call the computer instructions to enable the fundus camera to execute the method as described in any one of claims 1-7.

9. A computer-readable storage medium comprising instructions, characterized in that: When the instructions are executed on a fundus camera, the fundus camera is caused to perform the method according to any one of claims 1 to 7.

10. A computer program product, characterized in that When the computer program product runs on a fundus camera, the fundus camera is caused to perform the method according to any one of claims 1 to 7.

Images