Visual function detection method and system based on image analysis

Abstract

The invention relates to the technical field of visual function detection, in particular to a visual function detection method and system based on image analysis, and the method comprises the following steps: collecting a multi-wavelength reflection image of fundus, adjusting a light source to set wavelength parameters, activating camera equipment to receive an image, and carrying out the denoising and contrast adjustment of the collected image. A multi-wavelength retinal image dataset is generated. According to the method, the definition and the contrast ratio of the image are greatly improved by collecting the multi-wavelength image and carrying out detailed denoising processing, higher precision is brought to damage identification, misdiagnosis opportunities are reduced, the retinal damage is mapped by calculating the light intensity gradient, the damage range is accurately captured, positioning of the damage area is refined, and the accuracy of damage identification is improved. Through high-speed capture and data analysis of pupil dynamic response, deep insight of an eye movement mode is provided, visual function evaluation is more visual and comprehensive, and comprehensiveness and accuracy of evaluation are enhanced through cross validation of visual field defect and pupil dynamic data.

Description

Technical Field

[0001] The present invention relates to the technical field of visual function detection, and in particular to a visual function detection method and system based on image analysis. Background Art

[0002] The field of visual function testing technology includes methods and systems for evaluating and analyzing human visual ability, which are usually used in ophthalmic medicine, visual science research, and vision-related human-computer interface fields. Visual function testing is a key means of evaluating eye health and visual performance, ranging from basic vision tests to complex visual function analysis, which can help doctors diagnose visual disorders, evaluate the effectiveness of visual correction measures, and make early diagnosis of eye diseases.

[0003] Among them, the visual function detection method based on image analysis refers to the use of image capture equipment and analysis software to evaluate visual function, mainly by capturing image data of the eye, and then analyzing the retinal response or eye movement pattern through image processing technology. It can provide quantitative information about the visual state of the subject, and use image analysis and processing algorithms to determine the various parameters of visual function. The whole process does not involve complex computational model adjustments, but evaluates visual function by directly analyzing image data.

[0004] Existing technologies have obvious deficiencies in image detail resolution and comprehensive analysis, which limits their application in ophthalmic diagnosis and visual science research. Standardized visual tests lack adaptability to individual visual differences and cannot fully reveal complex retinal lesions or early functional loss. When processing dynamic visual information, existing methods generally fail to accurately reflect subtle changes in the pupil, resulting in insufficient assessment of the dynamic characteristics of the pupil, making it difficult to effectively correlate retinal function and pupillary response. Limitations may not only lead to incomplete diagnostic information, but may also affect treatment decisions and the evaluation of the effectiveness of visual correction measures, especially in areas that require high precision operations. Summary of the invention

[0005] The purpose of the present invention is to solve the shortcomings in the prior art and to propose a visual function detection method based on image analysis.

[0006] In order to achieve the above object, the present invention adopts the following technical solution: a visual function detection method based on image analysis, comprising the following steps: S1: Collect multi-wavelength reflectance images of the fundus, adjust the light source to set wavelength parameters, activate the camera to receive images, denoise and adjust the contrast of the collected images, and generate a multi-wavelength retinal image dataset; S2: Based on the multi-wavelength retinal image data set, the difference between the light intensity of each pixel and the light intensity of surrounding pixels is calculated, the light intensity gradient is collected, and the retinal damage and functional defect areas are revealed according to the difference calculation results, the retinal damage range is mapped, and the visual field defect area image is obtained; S3: Capture the dynamic process of pupil changes in response to light through high-speed video, adjust the shooting frequency to match the speed of pupil contraction and expansion, measure the change of pupil diameter for each frame of image, analyze the reaction speed of pupil to instantaneous light intensity change based on the measurement results, determine the sensitivity of pupil adjustment, and generate a pupil detection report; S4: Combining the visual field defect area image with the pupil detection report, cross-validating the two data, analyzing the state of visual function, and generating a comprehensive visual function evaluation result.

[0007] As a further solution of the present invention, the steps of acquiring the multi-wavelength retinal image dataset are: S111: Setting the wavelength of the light source to match the absorption characteristics of the retina, adjusting the light source to cover the entire spectrum from ultraviolet to infrared, and obtaining the adjusted wavelength parameters of the light source; S112: activating the camera device and calibrating the photosensitivity settings using the adjusted light source wavelength parameters, initializing the capture efficiency and quality of the multi-wavelength image, and generating a multi-wavelength original image set; S113: Perform signal processing on the multi-wavelength original image set, perform denoising and contrast enhancement, using the formula: ; Calculate the pixel value of the enhanced image , we get a multi-wavelength retinal image dataset, where represents a single image pixel value in the multi-wavelength raw image set, represent The average pixel value, represent The pixel standard deviation.

[0008] As a further solution of the present invention, the step of acquiring the light intensity gradient data is: S211: Based on the multi-wavelength retinal image dataset, data capture of the light intensity of each pixel in the image data is performed to obtain an initial light intensity dataset; S212: Based on the initial light intensity data set, using the formula: ; Calculate the light intensity difference between each pixel and its neighboring pixels , and obtain the light intensity difference data set, where Represents pixel The light intensity, Representative Points The surrounding pixels are is the number of pixels in the set; S213: Analyze the light intensity difference data set and set a threshold , identifying light intensity changes exceeding a threshold The corresponding area is marked as the area with significant changes in light intensity gradient, and light intensity gradient data is generated.

[0009] As a further solution of the present invention, the step of acquiring the image of the visual field defect area is: S221: Based on the light intensity gradient data, determine the area where the light intensity changes significantly, compare it with the critical threshold, and mark the part exceeding the critical threshold as a potential retinal damage area to generate damage identification data; S222: Based on the damage identification data, spatially map the potential retinal damage area, depict the boundary of the damage area and superimpose it on the standard model of the retina to generate a retinal damage map; S223: Using the retinal damage map, converting the mapping of the damaged area into a visual field defect map, emphasizing the damaged area through differential color marking, intuitively displaying the location and range of retinal damage, and obtaining a visual field defect area image.

[0010] As a further solution of the present invention, the step of measuring the change in pupil diameter is: S311: Record the dynamic process of the pupil in the continuous light change by a high-speed camera, adjust the shooting frequency of the camera, match the contraction and expansion of the pupil each time, and generate a pupil dynamic video; S312: extracting a pupil image of each frame from the pupil dynamic video, analyzing the geometric features of the pupil, and accurately measuring the diameter of the pupil of each frame to obtain pupil diameter sequence data; S313: Perform frame-by-frame differential processing on the pupil diameter sequence data, using the formula: ; Calculate the change in pupil diameter between the i-th frame and the previous frame , generate pupil diameter change data, where, and Represent the pupil diameters in two consecutive frames respectively.

[0011] As a further solution of the present invention, the steps of obtaining the pupil detection report are: S321: Based on the pupil diameter change data, the formula is used: ; Calculate the pupil's reaction speed to each change in light intensity , generate pupil response speed data, where represents the change in pupil diameter, Indicates a time interval; S322: Analyze the pupil response speed data, perform statistical processing on the data, evaluate the sensitivity of pupil adjustment, determine the pupil's adaptability to light intensity changes and the adjustment speed, and obtain a pupil adjustment sensitivity evaluation result; S323: Integrate the pupil response speed data and the pupil accommodation sensitivity evaluation result, compare the data with the health standard, evaluate the pupil function, and generate a pupil detection report.

[0012] As a further solution of the present invention, the steps of obtaining the comprehensive visual function evaluation result are: S411: Integrate the visual field defect area image and the pupil detection report, count the coordinates of the visual field defect and the time series of pupil dynamic changes, and obtain a retinal and pupil data set; S412: performing data cross-validation based on the retina and pupil data set to confirm the possibility and scope of visual function abnormality and generate a visual function abnormality correlation analysis result; S413: Based on the visual function abnormality correlation analysis results, the formula is used: ; Calculation of comprehensive visual function assessment score , generate comprehensive visual function assessment results, where represents the retinal damage score obtained from the correlation analysis of visual function abnormalities, is the mean of all reaction times in the pupil reaction dataset, is the total number of data items.

[0013] Visual function detection system based on image analysis, including: The light source setting module sets the wavelength of the light source to match the absorption characteristics of the retina, adjusts the light source to cover all spectra, activates the camera device and calibrates the photosensitivity settings, performs denoising and contrast enhancement, calculates the pixel values ​​of the enhanced image, and obtains a multi-wavelength retinal image dataset; The retinal analysis module captures the light intensity of each pixel in the image data based on the multi-wavelength retinal image data set, calculates the light intensity difference between each pixel and the adjacent pixel, identifies the area where the light intensity change exceeds the threshold, compares it with the critical threshold, and marks the part exceeding the critical threshold as a potential retinal damage area. The potential retinal damage area is spatially mapped, and the damaged area is emphasized by differential color marking to obtain an image of the visual field defect area; The pupil dynamic monitoring module records the dynamic process of the pupil in continuous light changes through a high-speed camera, extracts the pupil image of each frame, analyzes the geometric characteristics of the pupil, and accurately measures the pupil diameter of each frame, calculates the change in pupil diameter between the previous frame and the pupil's reaction speed to each light intensity change, determines the pupil's adaptability and adjustment speed to light intensity changes, compares it with the health standard, evaluates pupil function, and generates a pupil detection report; The comprehensive visual function evaluation module integrates the visual field defect area image and pupil detection report, counts the coordinates of the visual field defect and the time series of pupil dynamic changes, performs data cross-validation, calculates the comprehensive visual function evaluation score, and generates a comprehensive visual function evaluation result.

[0014] Compared with the prior art, the advantages and positive effects of the present invention are: In the present invention, by collecting multi-wavelength images and performing meticulous denoising processing, the clarity and contrast of the images are greatly improved, which brings higher accuracy to injury identification and reduces the chance of misdiagnosis. By calculating the light intensity gradient to map retinal damage, not only the damage range is accurately captured, but also the positioning of the damaged area is refined. The high-speed capture of pupil dynamic response and its data analysis provide in-depth insights into eye movement patterns, making the evaluation of visual function more intuitive and comprehensive. By cross-validating the visual field defect and pupil dynamic data, the comprehensiveness and accuracy of the evaluation are enhanced. BRIEF DESCRIPTION OF THE DRAWINGS

[0015] Figure 1 It is a main step flow chart of the present invention; Figure 2 is a flowchart of obtaining a multi-wavelength retinal image data set of the present invention; Figure 3 This is a flow chart for obtaining light intensity gradient data of the present invention; Figure 4 The flowchart of obtaining the image of the visual field defect area of ​​the present invention is as follows; Figure 5 It is a flow chart of measuring pupil diameter change of the present invention; Figure 6 A flowchart for obtaining a pupil detection report of the present invention; Figure 7 The figure is a flow chart for obtaining the comprehensive visual function evaluation results of the present invention. DETAILED DESCRIPTION

[0016] In order to make the purpose, technical solution and advantages of the present invention more clearly understood, the present invention is further described in detail below in conjunction with the accompanying drawings and embodiments. It should be understood that the specific embodiments described herein are only used to explain the present invention and are not intended to limit the present invention.

[0017] In the description of the present invention, it should be understood that the terms "length", "width", "up", "down", "front", "back", "left", "right", "vertical", "horizontal", "top", "bottom", "inside", "outside" and the like indicate positions or positional relationships based on the positions or positional relationships shown in the drawings, and are only for the convenience of describing the present invention and simplifying the description, rather than indicating or implying that the device or element referred to must have a specific orientation, be constructed and operated in a specific orientation, and therefore cannot be understood as limiting the present invention. In addition, in the description of the present invention, "multiple" means two or more, unless otherwise clearly and specifically defined.

[0018] See also Figure 1 , a visual function detection method based on image analysis, comprising the following steps: S1: Collect multi-wavelength reflectance images of the fundus, adjust the light source to set wavelength parameters, activate the camera to receive images, denoise and adjust the contrast of the collected images, and generate a multi-wavelength retinal image dataset; S2: Based on the multi-wavelength retinal image dataset, the difference between the light intensity of each pixel and the light intensity of the surrounding pixels is calculated, and the light intensity gradient is collected. According to the difference calculation results, the retinal damage and functional defect areas are revealed, the damage range of the retina is mapped, and the image of the visual field defect area is obtained; S3: Capture the dynamic process of pupil changes in response to light through high-speed video, adjust the shooting frequency to match the speed of pupil contraction and expansion, measure the change of pupil diameter for each frame of image, analyze the reaction speed of pupil to instantaneous light intensity change based on the measurement results, determine the sensitivity of pupil adjustment, and generate a pupil detection report; S4: Combine the visual field defect area image and pupil detection report, cross-validate the two data, analyze the status of visual function, and generate a comprehensive visual function assessment result.

[0019] The multi-wavelength retinal image dataset includes wavelength distribution records, contrast adjustment results and noise index; the visual field defect area image includes the damage boundary, the size of the affected area and the abnormal light intensity area; the pupil detection report includes the maximum pupil diameter, the minimum pupil diameter and the rate of change; the comprehensive visual function assessment results include the visual function integrity analysis results, pupil reaction health indicators and retinal damage correlation.

[0020] See also Figure 2 , the steps to obtain the multi-wavelength retinal image dataset are: S111: Setting the wavelength of the light source to match the absorption characteristics of the retina, adjusting the light source to cover the entire spectrum from ultraviolet to infrared, and obtaining the adjusted wavelength parameters of the light source; When setting the wavelength of the light source to suit the absorption characteristics of the retina, a detailed analysis of the spectral range of the light source is performed, including measuring the output energy of different types of light sources such as LEDs and lasers at different wavelengths, and using a spectrum analyzer to record the intensity of each wavelength of light in order to identify which bands can be most effectively absorbed by the retinal tissue. Next, based on the obtained spectral data, the settings on the light source controller are manually adjusted to ensure that the wavelength output of the light source can be adjusted as needed. For example, if the infrared and near-infrared bands show higher absorption efficiency, the operator will adjust the ability of the light source to emit specific wavelengths. The specific operation includes adjusting the current input of the light source and adjusting the position of the wavelength filter to ensure a wide spectrum coverage from ultraviolet to infrared.

[0021] S112: activating the camera device and calibrating the photosensitivity settings using the adjusted light source wavelength parameters, initializing the capture efficiency and quality of the multi-wavelength image, and generating a multi-wavelength original image set; In the process of activating and calibrating the camera to capture multi-wavelength images using the adjusted light source wavelength parameters, the hardware settings of the camera are first checked and configured, including adjusting the lens focus and aperture of the camera to adapt to the light intensity of different wavelengths. The operator needs to manually set the exposure time and sensitivity according to the wavelength adjustment parameters, which is usually done through the control panel to ensure that the camera can receive uniform and clear images at each wavelength. Subsequently, a series of test shots are taken to observe the image quality obtained under different settings, for example, testing the clarity and color accuracy of the image under lower wavelength ultraviolet light, and adjusting the exposure and ISO settings to reduce image noise and improve resolution. The process not only relies on the operator's technical operation, but also requires continuous adjustment of settings by observing the images obtained to ensure that the final image can be effectively used for subsequent medical analysis and diagnosis, thereby generating a multi-wavelength raw image set.

[0022] S113: Signal processing is performed on the multi-wavelength original image set to perform denoising and contrast enhancement, using the formula: ; Calculate the pixel value of the enhanced image , we get a multi-wavelength retinal image dataset, where represents a single image pixel value in the multi-wavelength raw image set, represent The average pixel value, represent The pixel standard deviation of In a set of image data, the original image The pixel value of a certain image is 120, and the average pixel value of the entire image is is 100, standard deviation is 15. First, calculate and The absolute value of the difference: ; Put the difference into the formula for square root and standard deviation: ; Then use To normalize this result: ; Calculation results Represents the pixel value after enhancement, compared with the original It improves the contrast of the image while maintaining the original brightness, making the details in the image more obvious, which helps to identify the details during the diagnosis process, especially when identifying and analyzing fundus images. The subtle blood vessels and tissue structures are easier to observe and analyze, providing clearer and more accurate retinal images, thereby better supporting clinical decision-making.

[0023] See also Figure 3 , the steps to obtain light intensity gradient data are: S211: Based on the multi-wavelength retinal image dataset, data capture of the light intensity of each pixel in the image data is performed to obtain an initial light intensity dataset; First, images are selected from a multi-wavelength retinal image dataset. The selection process involves evaluating the clarity, contrast, and color depth of the images to ensure that the selected images are of sufficient quality for in-depth analysis. Specifically, the pixel resolution of each image is first checked to ensure that it meets or exceeds the specified resolution standard, because high resolution can provide more details and help with subsequent light intensity calculations. Next, the image is color corrected. During the correction process, the image color is adjusted according to the distribution of light colors of different wavelengths in the image to make it reflect a more realistic retinal state. In addition, noise is an important factor affecting image quality. By comparing the background noise levels of different images, the image with the lowest noise is selected for analysis.

[0024] S212: Based on the initial light intensity data set, using the formula: ; Calculate the light intensity difference between each pixel and its neighboring pixels , and obtain the light intensity difference data set, where Represents pixel The light intensity, Representative Points The surrounding pixels are is the number of pixels in the set; Set the center pixel Light intensity is 150. Assume that there are 8 pixels around it, and their light intensity values ​​are 152, 148, 147, 149, 153, 151, 150, and 148 respectively. First, calculate the average light intensity of these 8 neighboring pixels: ; Subtract the light intensity of the center pixel from the average value to get the light intensity difference: ; Calculation results It shows that the central pixel is smaller than its surrounding pixels. The light intensity difference is very small. This means that in this target area, the light intensity change of the retinal image is not enough to identify it as a damaged or diseased area, which provides a basis for identifying the retinal area that needs further diagnosis and helps to detect and treat retinal diseases at an early stage.

[0025] S213: Analyze the light intensity difference data set and set the threshold , identifying light intensity changes exceeding a threshold The corresponding area is marked as an area with significant light intensity gradient change, and light intensity gradient data is generated; In the process of analyzing the light intensity difference data set, it is necessary to determine which areas have light intensity changes that exceed the normal range. First, obtain data from the light intensity difference data set, which represents the relative difference in light intensity between each pixel and its surrounding pixels. Next, based on the light intensity distribution of the entire data set, calculate the mean and standard deviation of the light intensity change, and set a threshold The threshold is usually set as the mean light intensity change plus two standard deviations to ensure that only significant changes are marked. These pixels are then marked as areas where the light intensity gradient changes significantly. The marking of these areas not only helps to identify potential retinal damage, but also provides a basis for further medical diagnosis.

[0026] See also Figure 4 , the steps for obtaining the image of the visual field defect area are: S221: Based on the light intensity gradient data, determine the area where the light intensity changes significantly, compare it with the critical threshold, and mark the part exceeding the critical threshold as a potential retinal damage area to generate damage identification data; Based on the foregoing, the area of ​​significant light intensity change on the retina is determined, and the light intensity difference of the area of ​​significant light intensity change is compared with a critical threshold. This critical threshold is used to distinguish between normal light intensity changes and changes that indicate potential damage. The light intensity difference of each pixel in the area of ​​significant light intensity change is compared with the critical threshold. If it exceeds this threshold, the pixel is marked as a potential damage area.

[0027] S222: Based on the damage identification data, spatially map the potential retinal damage area, depict the boundary of the damage area and superimpose it on the standard model of the retina to generate a retinal damage map; The marked potential lesion areas are mapped in detail, first using basic image processing operations such as dilation and erosion to enhance the visual recognition of the marked areas. Then, contour tracing is applied to each marked area to determine the precise boundaries. The contours are then superimposed on a standard retinal model, and the exact position and shape of each contour is adjusted according to the retinal anatomy to ensure a match with the actual retina.

[0028] S223: using the retinal damage map, converting the mapping of the damaged area into a visual field defect map, emphasizing the damaged area through differential color marking, visually displaying the location and range of the retinal damage, and obtaining an image of the visual field defect area; The aforementioned retinal damage map is converted into a visual field loss map that can be displayed intuitively. First, a series of color codes are defined to represent different degrees of damage: mild, moderate, and severe damage are represented by green, yellow, and red, respectively. Next, the area of ​​each damage level is colored accordingly according to the degree of damage to ensure immediate visual identification. Finally, these color-marked areas are synthesized into the final visual field loss image, which is completed through simple image overlay technology and color processing without relying on any external software, and can intuitively evaluate and analyze the damage status of the retina.

[0029] See also Figure 5 , the measurement steps of pupil diameter change are: S311: Record the dynamic process of the pupil in the continuous light change by a high-speed camera, adjust the shooting frequency of the camera, match the contraction and expansion of the pupil each time, and generate a pupil dynamic video; The dynamic process of the pupil in continuous light changes is recorded by a high-speed camera. In the specific process, the high-speed camera is first set to aim at the subject's eyes, and the focal length and exposure parameters of the camera are adjusted to adapt to the indoor light conditions to ensure image clarity. Subsequently, by comparing the pupil images before and after the light changes, each change of the pupil is captured frame by frame. At this time, each frame of the image needs to be sent to the computer through real-time image transmission. On the computer side, these image data will be received and stored, and each frame of the image will be marked with a timestamp to facilitate subsequent analysis of the relationship between pupil changes and time. Next, the pupil area is located from each frame of the image, the diameter of the pupil is measured, the pupil boundary in the image is identified, and its size is calculated.

[0030] S312: extracting a pupil image of each frame from the pupil dynamic video, analyzing the geometric features of the pupil, and accurately measuring the diameter of the pupil of each frame to obtain pupil diameter sequence data; The pupil image of each frame is extracted from the generated pupil dynamic video. During the process, each frame of the image first undergoes a preprocessing stage to adjust the contrast and brightness of the image to highlight the pupil area. Then edge detection is used to identify the pupil boundary, which mainly relies on detecting color and brightness changes in the image. When a significant color jump is detected, it is marked as the edge of the pupil. Each measurement of the pupil diameter is achieved by calculating the maximum and minimum boundary distances of the pupil in the edge detection results, which can effectively reduce the influence of ambient light and the errors caused by the subject's blinking. After the measurement is completed, all pupil diameter data are collected and sorted to form a continuous data sequence for analyzing the trend of pupil diameter changes over time, thereby providing detailed insights into the dynamic changes of the pupil.

[0031] S313: Perform frame-by-frame differential processing on the pupil diameter sequence data, using the formula: ; Calculate the change in pupil diameter between the i-th frame and the previous frame , generate pupil diameter change data, where, and Represent the pupil diameter in two consecutive frames respectively; Assuming that the pupil diameters of two consecutive frames are 2.4mm and 2.5mm respectively, the calculation process is: , indicating that the pupil diameter has increased by 0.1mm from the previous frame to the current frame. This result shows the pupil's reaction speed and adjustment ability to light. Further research on the changes can help understand the health of the visual system; and They represent the pupil diameters in two consecutive frames respectively. The formula calculates the change by directly comparing the difference in pupil diameters between two adjacent frames, which concisely and effectively provides data on the pupil's immediate response to light changes.

[0032] See also Figure 6 , the steps to obtain the pupil detection report are: S321: Based on pupil diameter change data, the formula is used: ; Calculate the pupil's reaction speed to each change in light intensity , generate pupil response speed data, where represents the change in pupil diameter, Indicates a time interval; Consider a typical case where the pupil diameter changes from 4.0 mm to 4.5 mm in 0.1 seconds. The response speed is calculated according to the formula: This result shows that the pupil's reaction speed is 5.0 mm per second. This faster change speed indicates that the pupil is highly sensitive to changes in light intensity. This result shows that the pupil has good accommodation function and can quickly respond to changes in external light. The speed index helps to diagnose whether there is any abnormality in the pupil's reaction function.

[0033] S322: Analyze pupil response speed data, perform statistical processing on the data, evaluate the sensitivity of pupil adjustment, determine the pupil's adaptability to light intensity changes and the adjustment speed, and obtain a pupil adjustment sensitivity evaluation result; Based on the pupil response speed data, a detailed statistical analysis was performed to evaluate the sensitivity of pupil accommodation. First, the collected response speed data were sorted and classified. To ensure the accuracy of data processing, each data point was manually checked to exclude any outliers or erroneous inputs. After that, the mean and standard deviation of the data were calculated. The calculated mean provides the central trend of the pupil response speed, while the standard deviation reflects the variability of the pupil response speed. The two statistical indicators are obtained by calculating the distance from the mean of each data point one by one, which can quantitatively describe the sensitivity of pupil accommodation and provide a scientific basis for further evaluation of pupil health status.

[0034] S323: Integrate pupil response speed data and pupil accommodation sensitivity assessment results, compare the data with health standards, evaluate pupil function, and generate a pupil detection report; After completing the assessment of pupil response speed and accommodation sensitivity, these data are used in combination to conduct a comprehensive assessment of pupil health. First, the statistical data obtained from the above calculations are compared with health standards, which are established based on a large amount of research and clinical data to ensure the reliability of the assessment results. In the comparative analysis, special attention is paid to data points that deviate from the normal range. Each such data point will be recorded and analyzed in detail, including its possible health risks and reasons for further diagnosis. In addition, considering the stability and adaptability of pupil accommodation, these characteristics are combined with the data of pupil response speed and sensitivity, and by analyzing how each parameter affects the overall performance of the pupil one by one, data support is provided for the final health assessment report.

[0035] See also Figure 7 , the steps to obtain the comprehensive visual function evaluation results are: S411: Integrate the visual field defect area image and pupil detection report, count the coordinates of the visual field defect and the time series of pupil dynamic changes, and obtain the retina and pupil data set; When integrating the images of the visual field defect area and the pupil detection report, the retinal image is first obtained through high-resolution scanning, and the edges and centers of the retinal damage areas are manually marked. Image processing techniques, such as edge detection and region growing algorithms, are used for each marked area to determine the exact coordinates and area of ​​the damaged area. Subsequently, the dynamic response of the pupil to light stimulation is recorded through a non-invasive eye tracking device. This device captures the time series data of pupil size changes and position movements, and the pupil reaction speed and amplitude are evaluated by quantitatively analyzing the changes. The captured images and time series data are synchronized, and the retinal damage data are associated with the pupil response data through data matching technology to form a comprehensive data set that includes both the spatial information of the damage and the temporal response characteristics of the pupil.

[0036] S412: Based on the retina and pupil data set, perform data cross-validation to confirm the possibility and scope of visual function abnormality and generate visual function abnormality correlation analysis results; When conducting the correlation analysis of visual function abnormalities, we first extracted retinal damage data and pupil response data from the comprehensive data set, and performed preliminary statistical analysis on the data, including the calculation of mean, variance and distribution characteristics. The statistical values ​​help to understand the basic behavior of the data and possible abnormal patterns. Next, the data cross-validation method is used to determine the correlation between retinal damage and pupil response, including calculating the correlation coefficient between the two sets of data to evaluate the direct connection between them. After that, a logistic regression model is established, using retinal damage data as the independent variable and pupil response characteristics as the dependent variable to minimize the prediction error, and finally determining the probability of visual function abnormalities and the range of damage. The analysis process not only considers single data points, but also analyzes the overall trend in the data set to ensure the accuracy and reliability of the results.

[0037] S413: Based on the results of the correlation analysis of visual function abnormalities, the formula is used: ; Calculation of comprehensive visual function assessment score , generate comprehensive visual function assessment results, where represents the retinal damage score obtained from the correlation analysis of visual function abnormalities, is the mean of all reaction times in the pupil reaction dataset, is the total number of data items; S = 85 is set to represent the retinal damage score obtained from the correlation analysis of visual function abnormalities, M = 75 represents the mean of all reaction times in the pupil reaction data set, H = 10 is the total number of data items, and the formula becomes: ; The calculation results are: ; The results showed that the comprehensive visual function assessment score was 0.316, indicating that in the current dataset, the correlation between retinal damage and pupil dynamic response was low, thereby assessing a smaller possibility of visual function abnormality, which helps clinicians to more accurately judge the functional status of the retina and pupil during diagnosis and treatment.

[0038] Visual function detection system based on image analysis, including: The light source setting module sets the wavelength of the light source to match the absorption characteristics of the retina, adjusts the light source to cover all spectra, activates the camera device and calibrates the photosensitivity settings, performs denoising and contrast enhancement, calculates the pixel values ​​of the enhanced image, and obtains a multi-wavelength retinal image dataset; The retinal analysis module is based on a multi-wavelength retinal image data set. It captures the light intensity of each pixel in the image data, calculates the light intensity difference between each pixel and its neighboring pixels, identifies the area where the light intensity changes exceed the threshold, and compares it with the critical threshold. The part exceeding the critical threshold is marked as a potential retinal damage area. The potential retinal damage area is spatially mapped, and the damaged area is emphasized by differential color marking to obtain an image of the visual field defect area. The pupil dynamic monitoring module records the dynamic process of the pupil in continuous light changes through a high-speed camera, extracts the pupil image of each frame, analyzes the geometric characteristics of the pupil, and accurately measures the pupil diameter of each frame, calculates the change in pupil diameter between the previous frame and the pupil's reaction speed to each light intensity change, determines the pupil's adaptability and adjustment speed to light intensity changes, compares it with the health standard, evaluates pupil function, and generates a pupil detection report; The comprehensive visual function assessment module integrates the visual field defect area image and pupil detection report, counts the coordinates of the visual field defect and the time series of pupil dynamic changes, performs data cross-validation, calculates the comprehensive visual function assessment score, and generates a comprehensive visual function assessment result.

[0039] The above are only preferred embodiments of the present invention and are not intended to limit the present invention in other forms. Any technician familiar with the profession may use the technical contents disclosed above to change or modify them into equivalent embodiments with equivalent changes and apply them to other fields. However, any simple modification, equivalent change and modification made to the above embodiments based on the technical essence of the present invention without departing from the technical solution of the present invention still falls within the protection scope of the technical solution of the present invention.

Claims

1. A visual function detection method based on image analysis, characterized in that: The following steps are involved: S1: Collect multi-wavelength reflectance images of the fundus, adjust the wavelength parameters of the light source, activate the camera to receive images, denoise and adjust the contrast of the collected images, and generate a multi-wavelength retinal image dataset; S2: Based on the multi-wavelength retinal image data set, the difference between the light intensity of each pixel and the light intensity of surrounding pixels is calculated, the light intensity gradient is collected, and the retinal damage and functional defect areas are revealed according to the difference calculation results, the retinal damage range is mapped, and the visual field defect area image is obtained; S3: Capture the dynamic process of pupil changes in response to light through high-speed video, adjust the shooting frequency to match the speed of pupil contraction and expansion, measure the change of pupil diameter for each frame of image, analyze the reaction speed of pupil to instantaneous light intensity change based on the measurement results, determine the sensitivity of pupil adjustment, and generate a pupil detection report; S4: Combining the visual field defect area image with the pupil detection report, cross-validating the two data, analyzing the state of visual function, and generating a comprehensive visual function evaluation result.

2. The method for detecting visual function based on image analysis according to claim 1, characterized in that: The steps of acquiring the multi-wavelength retinal image dataset are as follows: S111: Setting the wavelength of the light source to match the absorption characteristics of the retina, adjusting the light source to cover the entire spectrum from ultraviolet to infrared, and obtaining the adjusted wavelength parameters of the light source; S112: activating the camera device and calibrating the photosensitivity settings using the adjusted light source wavelength parameters, initializing the capture efficiency and quality of the multi-wavelength image, and generating a multi-wavelength original image set; S113: Perform signal processing on the multi-wavelength original image set, perform denoising and contrast enhancement, using the formula: ； Calculate the pixel value of the enhanced image , we get a multi-wavelength retinal image dataset, where represents a single image pixel value in the multi-wavelength raw image set, represent The average pixel value, represent The pixel standard deviation.

3. The visual function detection method based on image analysis according to claim 2, characterized in that: The steps for obtaining the light intensity gradient data are: S211: Based on the multi-wavelength retinal image dataset, data capture of the light intensity of each pixel in the image data is performed to obtain an initial light intensity dataset; S212: Based on the initial light intensity data set, using the formula: ； Calculate the light intensity difference between each pixel and its neighboring pixels , and obtain the light intensity difference data set, where Represents pixel The light intensity, Representative Points The surrounding pixels are is the number of pixels in the set; S213: Analyze the light intensity difference data set and set a threshold , identifying light intensity changes exceeding a threshold The corresponding area is marked as the area with significant changes in light intensity gradient, and light intensity gradient data is generated.

4. The method for detecting visual function based on image analysis according to claim 3, characterized in that: The steps of acquiring the image of the visual field defect area are: S221: Based on the light intensity gradient data, determine the area where the light intensity changes significantly, compare it with the critical threshold, and mark the part exceeding the critical threshold as a potential retinal damage area to generate damage identification data; S222: Based on the damage identification data, spatially map the potential retinal damage area, depict the boundary of the damage area and superimpose it on the standard model of the retina to generate a retinal damage map; S223: Using the retinal damage map, converting the mapping of the damaged area into a visual field defect map, emphasizing the damaged area through differential color marking, intuitively displaying the location and range of the retinal damage, and obtaining a visual field defect area image.

5. The method for detecting visual function based on image analysis according to claim 4, characterized in that: The measuring steps of the pupil diameter change are: S311: Record the dynamic process of the pupil in the continuous light change by a high-speed camera, adjust the shooting frequency of the camera, match the contraction and expansion of the pupil each time, and generate a pupil dynamic video; S312: extracting a pupil image of each frame from the pupil dynamic video, analyzing the geometric features of the pupil, and accurately measuring the diameter of the pupil of each frame to obtain pupil diameter sequence data; S313: Perform frame-by-frame differential processing on the pupil diameter sequence data, using the formula: ； Calculate the change in pupil diameter between the i-th frame and the previous frame , generate pupil diameter change data, where, and Represent the pupil diameters in two consecutive frames respectively.

6. The method for detecting visual function based on image analysis according to claim 5, characterized in that: The steps for obtaining the pupil detection report are: S321: Based on the pupil diameter change data, the formula is used: ； Calculate the pupil's reaction speed to each change in light intensity , generate pupil response speed data, where represents the change in pupil diameter, Indicates a time interval; S322: Analyze the pupil response speed data, perform statistical processing on the data, evaluate the sensitivity of pupil adjustment, determine the pupil's adaptability to light intensity changes and the adjustment speed, and obtain a pupil adjustment sensitivity evaluation result; S323: Integrate the pupil response speed data and the pupil accommodation sensitivity evaluation result, compare the data with the health standard, evaluate the pupil function, and generate a pupil detection report.

7. The method for detecting visual function based on image analysis according to claim 6, characterized in that: The steps for obtaining the comprehensive visual function evaluation result are as follows: S411: Integrate the visual field defect area image and the pupil detection report, count the coordinates of the visual field defect and the time series of pupil dynamic changes, and obtain a retinal and pupil data set; S412: performing data cross-validation based on the retina and pupil data set to confirm the possibility and scope of visual function abnormality and generate a visual function abnormality correlation analysis result; S413: Based on the visual function abnormality correlation analysis results, the formula is used: ； Calculation of comprehensive visual function assessment score , generate comprehensive visual function assessment results, where represents the retinal damage score obtained from the correlation analysis of visual function abnormalities, is the mean of all reaction times in the pupil reaction dataset, is the total number of data items.

8. A visual function detection system based on image analysis, characterized in that: The visual function detection system based on image analysis is used to execute the visual function detection method based on image analysis according to any one of claims 1 to 7, comprising: The light source setting module sets the wavelength of the light source to match the absorption characteristics of the retina, adjusts the light source to cover all spectra, activates the camera device and calibrates the photosensitivity settings, performs denoising and contrast enhancement, calculates the pixel values ​​of the enhanced image, and obtains a multi-wavelength retinal image dataset; The retinal analysis module captures the light intensity of each pixel in the image data based on the multi-wavelength retinal image data set, calculates the light intensity difference between each pixel and the adjacent pixel, identifies the area where the light intensity change exceeds the threshold, compares it with the critical threshold, and marks the part exceeding the critical threshold as a potential retinal damage area. The potential retinal damage area is spatially mapped, and the damaged area is emphasized by differential color marking to obtain an image of the visual field defect area; The pupil dynamic monitoring module records the dynamic process of the pupil in continuous light changes through a high-speed camera, extracts the pupil image of each frame, analyzes the geometric characteristics of the pupil, and accurately measures the pupil diameter of each frame, calculates the change in pupil diameter between the previous frame and the pupil's reaction speed to each light intensity change, determines the pupil's adaptability and adjustment speed to light intensity changes, compares it with the health standard, evaluates pupil function, and generates a pupil detection report; The comprehensive visual function evaluation module integrates the visual field defect area image and pupil detection report, counts the coordinates of the visual field defect and the time series of pupil dynamic changes, performs data cross-validation, calculates the comprehensive visual function evaluation score, and generates a comprehensive visual function evaluation result.

Images