Method, system and equipment for detecting xerophthalmia and medium

Abstract

The invention discloses a xerophthalmia detection method, system and device and a medium. The method comprises the steps that eyelid point position time sequence data, eyebrow point position time sequence data and eye redness probability in collected historical video information are recognized frame by frame; calculating an eyelid distance and an eyebrow distance to construct an eyelid distance time sequence data set and an eyebrow distance time sequence data set; calculating a blinking identification threshold value and an eyebrow identification threshold value, identifying eyebrow frames according to the eyelid distance time sequence data set, the blinking identification threshold value, the identified blinking frames, the eyebrow point position time sequence data set and the eyebrow identification threshold value, and calculating blinking frame data of each blinking frame and eyebrow frame data of each eyebrow frame; and training a xerophthalmia detection model by using the blink frame data, the rugosa frame data and the eye redness probability to obtain a target xerophthalmia detection model for real-time detection of xerophthalmia. According to the method provided by the invention, a convenient, practical and accurate xerophthalmia detection method capable of being popularized and applied on a large scale can be provided.

Description

Technical Field

[0001] The present invention relates to the field of medical technology, and in particular to a dry eye detection method, system, equipment and medium. Background Art

[0002] Dry eye is a common eye disease, often accompanied by symptoms such as dry eyes, burning sensation, stinging and foreign body sensation, and may even lead to blurred vision and decreased vision, seriously affecting the patient's quality of life. With the significant increase in the use of screen devices such as mobile phones and computers, the incidence of dry eye continues to remain high. Among the high-risk white-collar workers, the incidence rate exceeds 50%. However, due to the large number of dry eye patients and the difficulty in completely curing them, a heavy burden is placed on limited ophthalmic medical and health resources. A large number of dry eye patients find it difficult to obtain professional diagnosis and treatment from ophthalmologists in a timely manner, which leads to worsening of the disease, seriously affecting their quality of life, and even inducing blinding eye diseases such as corneal inflammation.

[0003] Existing dry eye detection methods have the following problems: 1. They require supporting hardware equipment and / or drug support, which is not convenient for daily use and large-scale promotion; 2. The data collection conditions are high and the data processing is insufficient, resulting in low accuracy of the test results.

[0004] It can be seen that how to improve the convenience, practicality and accuracy of dry eye detection has become a technical problem that technical personnel in this field need to solve urgently. Summary of the invention

[0005] The present invention provides a dry eye detection method and system to solve the technical problems of improving the convenience, practicality and accuracy of dry eye detection, and to achieve the effect of providing a dry eye detection method that is convenient, practical, accurate and can be widely promoted and applied.

[0006] In a first aspect, the present invention provides a method for detecting dry eye, the method comprising:

[0007] Collect historical video information containing human faces, and identify eyelid point time series data, eyebrow point time series data and eye redness probability in the historical video information frame by frame;

[0008] Calculating the eyelid distance frame by frame according to the eyelid point time series data to construct an eyelid distance time series data set, and calculating the eyebrow-eye distance frame by frame according to the eyebrow-eye point time series data to construct an eyebrow-eye distance time series data set;

[0009] Based on the fluctuation of the eyelid distance time series data set and the abnormal value of the blink amplitude, a blink recognition threshold is calculated; based on the fluctuation of the eyebrow-eye distance time series data set and the abnormal value of the eyebrow-eye movement amplitude, a frown recognition threshold is calculated;

[0010] According to the size relationship between the eyelid distance time series data set and the blink recognition threshold, the blink frame is identified; according to the size relationship between the eyebrow-eye distance time series data set and the frown recognition threshold, the frown frame is identified;

[0011] Based on the eyelid distance time series data set, the blink frame data of each blink frame is calculated; based on the eyebrow-eye distance time series data set, the frown frame data of each frown frame is calculated;

[0012] The constructed dry eye syndrome detection model is trained using the blink frame data, the frown frame data and the red eye probability to obtain a target dry eye syndrome detection model;

[0013] During the real-time detection process, the real-time blink frame data, real-time frown frame data and real-time red eye probability extracted from the collected real-time video information are input into the target dry eye detection model to obtain the dry eye detection result.

[0014] Preferably, the frame-by-frame identification of eyelid point time series data, eyebrow point time series data and eye redness probability in the historical video information includes:

[0015] The dlib model is used to identify 68 standard points of facial features in the video information frame by frame:

[0016] Selecting eyelid point data and eyebrow point data from the 68 points;

[0017] Combining the time series data of the video information of each frame with the eyelid point data to construct the eyelid point time series data;

[0018] Combining the time series data with the eyebrow and eye point data to construct the eyebrow and eye point time series data;

[0019] The video information of each frame is input into the trained red eye recognition model to obtain the red eye probability.

[0020] Preferably, the eyelid distance is defined as: the difference between the upper eyelid ordinate and the lower eyelid ordinate;

[0021] The eyebrow-eye distance is defined as the difference between the ordinate of the highest point of the eyebrow and the ordinate of the inner corner of the eye.

[0022] Preferably, the calculation of the blink recognition threshold based on the fluctuation of the eyelid distance time series data set and the blink amplitude abnormal value comprises:

[0023] Calculating the global mean of eyelid distance and the global standard deviation of eyelid distance of the eyelid distance time series data set, and calculating the global threshold of eyelid distance according to the global mean of eyelid distance and the global standard deviation of eyelid distance;

[0024] The calculation formula of the eyelid distance global threshold is:

[0025] L l =L mean -C l ×((L std )×(1-3L std ))

[0026] Among them, L l represents the global threshold of eyelid distance, L mean represents the global mean of eyelid distance, L std represents the global standard deviation of eyelid distance, C l Indicates the sensitive parameters of blink recognition;

[0027] Select video frames in a first preset time period before and after the historical video information to represent abnormal blink amplitude frames, calculate the local mean of eyelid distance and the local standard deviation of eyelid distance of the eyelid distance time series data set corresponding to the abnormal blink amplitude frames, and calculate the local threshold of eyelid distance according to the local mean of eyelid distance and the local standard deviation of eyelid distance;

[0028] The calculation formula of the eyelid distance local threshold is:

[0029] L l ′=L local-mean -C l ×((L local-std )×(1-3L local-std ))

[0030] Among them, L l ' represents the local threshold of eyelid distance, L local-mean represents the local mean of eyelid distance, L local-std represents the local standard deviation of eyelid distance;

[0031] The eyelid distance global threshold and the eyelid distance local threshold are fused according to a first preset weight to obtain the blink recognition threshold.

[0032] Preferably, the frown recognition threshold is calculated based on the fluctuation of the eyebrow-eye distance time series data set and the eyebrow-eye movement amplitude abnormal value, including:

[0033] Calculate the global mean of the eyebrow-eye distance and the global standard deviation of the eyebrow-eye distance time series data set, and calculate the global threshold of the eyebrow-eye distance according to the global mean of the eyebrow-eye distance and the global standard deviation of the eyebrow-eye distance;

[0034] The calculation formula of the eyebrow-eye distance global threshold is:

[0035] Lm =M mean -C m ×((M std )×(1-3M std ))

[0036] Among them, L m represents the global threshold of eyebrow-eye distance, M mean represents the global mean of eyebrow-eye distance, M std represents the global standard deviation of the eyebrow-eye distance, C m Indicates frown recognition sensitive parameters;

[0037] Select the video frames of the first preset time period to represent the frames with abnormal eyebrow and eye movement amplitude, calculate the local mean of the eyebrow and eye distance and the local standard deviation of the eyebrow and eye distance time series data set corresponding to the frames with abnormal eyebrow and eye movement amplitude, and calculate the local threshold of the eyebrow and eye distance according to the local mean of the eyebrow and eye distance and the local standard deviation of the eyebrow and eye distance;

[0038] The calculation formula of the local threshold of the eyebrow-eye distance is:

[0039] L m ′=M local-mean -C m ×((M local-std )×(1-3M local-std ))

[0040] Among them, L m ' represents the local threshold of eyebrow distance, M local-mean represents the local mean of the eyebrow distance, M local-std represents the local standard deviation of the eyebrow-eye distance;

[0041] The eyebrow-eye distance global threshold and the eyebrow-eye distance local threshold are fused according to a second preset weight to obtain the frown recognition threshold.

[0042] Preferably, the step of calculating the blink frame data of each blink frame based on the eyelid distance time series data set includes:

[0043] Selecting video frames within a second preset time period before and after each of the blink frames to represent blink-related frames, and selecting the maximum eyelid distance and the minimum eyelid distance corresponding to the blink-related frames from the eyelid distance time series data set;

[0044] Calculate the first time series interval in which the eyelid distance of each blink-related frame is less than the blink recognition threshold, and use the first time series interval as the blink time;

[0045] Calculating the mean of the blink time to obtain an average blink time;

[0046] According to the size of the blink time and a preset blink threshold, determining whether the blink frame is a complete blink or an incomplete blink, and counting the number of complete blinks and the number of incomplete blinks in the blink frame;

[0047] Calculating a complete blink time standard deviation and a complete blink time average value of the blink time corresponding to the complete blink, and taking the ratio of the complete blink time standard deviation to the complete blink time average value as the blink interval time variability;

[0048] Determine a minimum blink time value according to each of the blink times;

[0049] Calculate the blink interval according to the second time series interval between the minimum values ​​of the eyelid distances of each blink frame, and select the maximum blink interval from the blink intervals;

[0050] Calculate a third time series interval of the eyelid distance of each blink frame from being less than the blink recognition threshold to the minimum value of the eyelid distance, and use the third time series interval as the eye closing time;

[0051] Calculate the eyelid distance of each of the blink frames from the minimum eyelid distance to a fourth time series interval greater than the blink recognition threshold, and use the fourth time series interval as the eye-opening time;

[0052] Selecting a minimum eye closing time from the eye closing times, and calculating an average eye closing time based on each of the eye closing times;

[0053] Selecting a minimum eye-opening time from the eye-opening times, and calculating an average eye-opening time based on each of the eye-opening times;

[0054] The maximum eyelid distance, the minimum eyelid distance, the number of complete blinks, the number of incomplete blinks, the minimum blink time, the average blink time, the maximum blink interval, the blink interval time variability, the minimum eye closing time, the average eye closing time, the minimum eye opening time and the average eye opening time are taken as blink frame data.

[0055] Preferably, the step of calculating the frown frame data of each frown frame based on the eyebrow distance time series data set includes:

[0056] Select video frames within a third preset time period before and after each frown frame to represent frown-related frames, and select the maximum eyebrow-eye distance and the minimum eyebrow-eye distance corresponding to the frown-related frames from the eyebrow-eye distance time series data set;

[0057] Calculate the fifth time series interval at which the eyebrow-eye distance of each frown-associated frame is less than the frown recognition threshold, and determine the frown time and frown frequency according to the fifth time series interval;

[0058] Calculating the frown time standard deviation and the frown time average value of the frown time, and taking the ratio of the frown time standard deviation to the frown time average value as the frown interval time variability;

[0059] Determine a maximum frown time according to each of the frown times;

[0060] Calculate the frown interval according to the sixth time series interval between the minimum values ​​of the eyebrow-eye distance of each frown frame, and select the minimum frown interval from the frown intervals;

[0061] The maximum eyebrow-eye distance, the minimum eyebrow-eye distance, the frowning time, the number of frowning times, the frowning interval time variability, the maximum frowning time and the minimum frowning interval are used as frowning frame data.

[0062] In a second aspect, the present invention further provides a dry eye detection system to implement the above-mentioned dry eye detection method, the system comprising: a data acquisition processing unit, a first calculation unit, a recognition threshold calculation unit, a recognition unit, a second calculation unit, a model training unit and a real-time detection unit;

[0063] The data acquisition processing unit is used to acquire historical video information containing human faces, and identify eyelid point time series data, eyebrow point time series data and eye redness probability in the historical video information frame by frame;

[0064] The first calculation unit is used to calculate the eyelid distance frame by frame according to the eyelid point time series data to construct an eyelid distance time series data set, and calculate the eyebrow-eye distance frame by frame according to the eyebrow-eye point time series data to construct an eyebrow-eye distance time series data set;

[0065] The recognition threshold calculation unit is used to calculate the blink recognition threshold based on the fluctuation of the eyelid distance time series data set and the abnormal value of the blink amplitude, and calculate the frown recognition threshold based on the fluctuation of the eyebrow-eye distance time series data set and the abnormal value of the eyebrow-eye movement amplitude;

[0066] The recognition unit is used to recognize the blink frame according to the size relationship between the eyelid distance time series data set and the blink recognition threshold, and recognize the frown frame according to the size relationship between the eyebrow distance time series data set and the frown recognition threshold;

[0067] The second calculation unit is used to calculate the blink frame data of each blink frame based on the eyelid distance time series data set, and calculate the frown frame data of each frown frame based on the eyebrow-eye distance time series data set;

[0068] The model training unit is used to train the constructed dry eye detection model using the blink frame data, the frown frame data and the red eye probability to obtain a target dry eye detection model;

[0069] The real-time detection unit is used to input the real-time blink frame data, real-time frown frame data and real-time red eye probability extracted from the collected real-time video information into the target dry eye detection model during the real-time detection process to obtain the dry eye detection result.

[0070] In a third aspect, the present invention also provides a computer device, comprising a memory, a processor and a transceiver, which are connected via a bus; the memory is used to store a set of computer program instructions and data, and transmit the stored data to the processor, and the processor executes the program instructions stored in the memory to execute the dry eye detection method described above.

[0071] In a fourth aspect, the present invention further provides a computer-readable storage medium, wherein a computer program is stored in the computer-readable storage medium. When the computer program is executed, the above-mentioned L dry eye detection method is implemented.

[0072] The present invention provides a dry eye detection method, system, device and medium. Compared with the prior art, the beneficial effects of the embodiments of the present invention are at least one of the following:

[0073] (1) To provide a convenient, practical, accurate and large-scale dry eye detection method, so that patients can complete dry eye screening independently and obtain corresponding treatment or relief suggestions, and provide patients with timely and effective self-management plans to improve their quality of life.

[0074] (2) Screen patients with dry eye disease in a timely manner, provide self-help diagnosis and treatment advice for patients with mild symptoms, help refer patients with moderate to severe symptoms for treatment, and promote the efficient allocation of medical resources. BRIEF DESCRIPTION OF THE DRAWINGS

[0075] Figure 1 This is a schematic diagram of the steps of a dry eye detection method provided by a preferred embodiment of the present invention;

[0076] Figure 2 68 key points of a face provided by a preferred embodiment of the present invention;

[0077] Figure 3A blink determination line graph provided by a preferred embodiment of the present invention;

[0078] Figure 4 It is a structural schematic diagram of a dry eye detection system provided by a preferred embodiment of the present invention;

[0079] Figure 5 It is a schematic diagram of the structure of a computer device provided in one embodiment of the present invention. DETAILED DESCRIPTION

[0080] The following is a detailed explanation of the implementation mode of the present invention in conjunction with the accompanying drawings. The embodiments are provided only for illustrative purposes and cannot be understood as limiting the present invention. The accompanying drawings are only for reference and illustration purposes and do not constitute a limitation on the scope of protection of the present invention. Based on the embodiments in the present invention, all other embodiments obtained by ordinary technicians in this field without making creative work are within the scope of protection of the present invention. In the description of the present invention, the terms "first", "second", "third", etc. are only used for descriptive purposes and cannot be understood as indicating or implying relative importance or implicitly indicating the number of indicated technical features. Therefore, the features defined as "first", "second", "third", etc. may explicitly or implicitly include one or more of the features. In the description of the present invention, unless otherwise specified, the meaning of "multiple" is two or more.

[0081] In the description of the present invention, it should be noted that, unless otherwise clearly specified and limited, the terms "installed", "connected" and "connected" should be understood in a broad sense. For example, it can be a fixed connection, a detachable connection, or an integral connection; it can be a mechanical connection or an electrical connection; it can be a direct connection, or it can be indirectly connected through an intermediate medium, or it can be a connection between the two elements. The terms "vertical", "horizontal", "left", "right", "upper", "lower" and similar expressions used herein are only for illustrative purposes, and do not indicate or imply 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 a limitation on the present invention. The term "and / or" used herein includes any and all combinations of one or more related listed items. For those of ordinary skill in the art, the specific meanings of the above terms in the present invention can be understood according to specific circumstances.

[0082] In the description of the present invention, it should be noted that, unless otherwise defined, all technical and scientific terms used in the present invention have the same meanings as those commonly understood by those skilled in the art. The terms used in the specification of the present invention are only for the purpose of describing specific embodiments and are not intended to limit the present invention. For those of ordinary skill in the art, the specific meanings of the above terms in the present invention can be understood by specific circumstances.

[0083] See also Figure 1 In an embodiment of the present invention, a method for detecting dry eye syndrome is provided, the method comprising:

[0084] S1. Collect historical video information containing human faces, and identify eyelid point time series data, eyebrow point time series data and red eye probability in the historical video information frame by frame.

[0085] S2. Calculate the eyelid distance frame by frame according to the eyelid point time series data to construct an eyelid distance time series data set, and calculate the eyebrow-eye distance frame by frame according to the eyebrow-eye point time series data to construct an eyebrow-eye distance time series data set.

[0086] S3. Calculate the blink recognition threshold based on the fluctuation of the eyelid distance time series data set and the abnormal value of the blink amplitude, and calculate the frown recognition threshold based on the fluctuation of the eyebrow distance time series data set and the abnormal value of the eyebrow movement amplitude.

[0087] S4. According to the size relationship between the eyelid distance time series data set and the blink recognition threshold, identify the blink frame; according to the size relationship between the eyebrow distance time series data set and the frown recognition threshold, identify the frown frame.

[0088] S5. Calculate the blink frame data of each blink frame based on the eyelid distance time series data set, and calculate the frown frame data of each frown frame based on the eyebrow distance time series data set.

[0089] S6. Using the blink frame data, the frown frame data and the red eye probability, the constructed dry eye detection model is trained to obtain a target dry eye detection model.

[0090] S7. During the real-time detection process, the real-time blink frame data, real-time frown frame data and real-time red eye probability extracted from the collected real-time video information are input into the target dry eye detection model to obtain the dry eye detection result.

[0091] In a preferred embodiment of the present invention, a dedicated video acquisition device or a smart phone can be used to collect video information, so that users can perform daily detection of dry eye syndrome. In the present invention, the use of a smart phone to collect video information is used as an example for explanation. A mobile phone is used to collect historical video information containing faces. The collection of historical video information can be in the form of live recording, or the recorded video can be uploaded. Various video information collection methods improve the convenience of information collection. The video collection time can be set according to demand, generally 1 minute.

[0092] After the historical video information is collected, the information related to the face in the historical video information is identified frame by frame. In the embodiment of the present application, the dlib model is used to identify the 68 standard points of the facial features in the video information frame by frame. The dlib model is a C++ library containing machine learning algorithms, which is particularly good at image processing and deep learning and is widely used in computer vision, machine learning and data analysis. In the field of face recognition, dlib can detect 68 key points of the face (such as Figure 2 As shown in the figure, these key points are crucial for the subsequent eyelid feature extraction and eyebrow feature extraction. The dlib facial key point detector has the characteristics of fast speed and high accuracy. It selects eyelid point data and eyebrow point data from 68 points, where the eyelid point data are points numbered 37-48, and the eyebrow point data are points numbered 18-27 and 37-48 of the eyelid point data. The time series data of each frame of video information is combined with the eyelid point data to construct the eyelid point time series data, and the time series data is combined with the eyebrow point data to construct the eyebrow point time series data.

[0093] To obtain the probability of red eyes, each frame of video information is input into the trained red eyes recognition model to obtain the probability of red eyes. The red eyes recognition model is generally constructed by combining image processing and computer vision recognition to detect and identify the probability of the eyes being red. Common model construction algorithms include support vector machines, random forests, and deep learning classifiers.

[0094] Furthermore, after the historical video information is identified and processed, the eyelid distance is calculated frame by frame for the eyelid point time series data. The eyelid distance is defined as the difference between the upper eyelid ordinate and the lower eyelid ordinate, such as Figure 2 As shown, for the right eye, it is the difference between the ordinates of the two groups of upper and lower eyelid points 38 and 42, and 39 and 41, and for the left eye, it is the difference between the ordinates of the two groups of upper and lower eyelid points 44 and 46, and 45 and 47. The differences in the ordinates of the four groups of upper and lower eyelid points of the right eye and the left eye in each frame are averaged to obtain the eyelid distance of the video frame, and the eyelid distances of all frames are stored in a list data to construct an eyelid distance time series dataset.

[0095] For the time series data of eyebrow and eye position, the eyebrow and eye distance is calculated frame by frame. The eyebrow and eye distance is defined as the difference between the ordinate of the highest point of the eyebrow and the ordinate of the inner corner of the eye. Figure 2As shown, for the right eye, it is the difference between the ordinates of the 20 and 40 points, and for the left eye, it is the difference between the ordinates of the 25 and 43 points. When a person frowns, the distance between the highest point of the eyebrow and the middle line of the eye changes. Therefore, in the present invention, the highest point of the eyebrow and the inner corner of the eye are selected to characterize the change of the eyebrow-eye distance affected by frowning. The difference between the ordinates of the highest point of the eyebrow and the inner corner of the eye of the right eye and the left eye of each frame is averaged to obtain the eyebrow-eye distance of the video frame, and the eyebrow-eye distance of all frames is stored in a list data to construct an eyebrow-eye distance time series data set.

[0096] Furthermore, based on the fluctuation of the eyelid distance time series dataset and the blink amplitude anomalies, the blink recognition threshold is calculated, as Figure 3 As shown, it is a blink judgment line graph. In the line graph, the blue discount is the blink recognition threshold, and the blink recognition threshold is also a dynamically changing line. In the preferred embodiment of the present invention, the calculation of the blink recognition threshold needs to take into account special circumstances, such as when some users open their eyes, the distance between their eyelids will become larger or smaller as a whole; some users will not blink throughout the whole process; some users cannot open their eyes throughout the whole process, and the distance between the upper and lower eyelids changes little; some users have eye size problems, and the distance between their eyelids changes little throughout the whole process. In order to eliminate the noise generated by the above situations, this application uses the local threshold of eyelid distance for correction. Specifically, the global mean of eyelid distance and the global standard deviation of eyelid distance of the eyelid distance time series data set are calculated, and the global threshold of eyelid distance is calculated based on the global mean of eyelid distance and the global standard deviation of eyelid distance. The eyelid distance time series dataset is the distance between the upper and lower eyelid points pixels identified in each frame of the video. The specific format is: [62.0, 62.0, 63.5, 62.0, 62.5, 59.5, 56.0, 59.0, 58.5, 60.5, 61.5, 61.5, 63.0, 63.5, 65.0, 65.0, 66.0, 66.0, 67.5, 65.0, 66.5, 66.0, 65.5, 65.5, 66.0, 65.0, 65.0, 65.0, 64.5, 65.5, ...]. The video is generally 16 frames per second. The pixel distance is normalized to 0 according to the global maximum value. The global mean and global standard deviation of the eyelid distance of the normalized eyelid distance time series data set are calculated. The global mean of the eyelid distance is used as the threshold baseline, and the global standard deviation of the eyelid distance is used as the judgment of the overall fluctuation degree of the eyelid distance. The global threshold of the eyelid distance is calculated. The calculation formula of the global threshold of the eyelid distance is:

[0097] L l =L mean -C l ×((L std )×(1-3L std ))

[0098] Among them, L l represents the global threshold of eyelid distance, L mean represents the global mean of eyelid distance, L std represents the global standard deviation of eyelid distance, C l Indicates the sensitive parameters of blink recognition.

[0099] Formula C l ×((L std )×(1-3L std ))Use the global standard deviation of eyelid distance to control the accuracy of blink recognition threshold. The larger the global standard deviation of eyelid distance, the greater the noise of the patient's blink data. The smaller the distance between the eyelid distance and the global mean of eyelid distance each time the patient blinks, the smaller the threshold closer to the mean is needed for detection. Formula C l ×((L std )×(1-3L std ))Adjust the situation where the global standard deviation of eyelid distance is large to improve the accuracy of the global threshold of eyelid distance.

[0100] For the calculation of the local threshold of eyelid distance, the video frames of the first preset time period before and after the historical video information are selected to represent the abnormal frames of blink amplitude, and the local mean and local standard deviation of the eyelid distance of the eyelid distance time series data set corresponding to the abnormal frames of blink amplitude are calculated. The local threshold of eyelid distance is calculated according to the local mean and local standard deviation of eyelid distance. The calculation formula of the local threshold of eyelid distance is:

[0101] L l ′=L local-mean -C l ×((L local-std )×(1-3L local-std ))

[0102] Among them, L l ' represents the local threshold of eyelid distance, L local-mean represents the local mean of eyelid distance, L local-std Represents the local standard deviation of eyelid distance.

[0103] After obtaining the global eyelid distance threshold and the local eyelid distance threshold, the weights of the global eyelid distance threshold and the local eyelid distance threshold are set as the first preset weight, and the global eyelid distance threshold and the local eyelid distance threshold are fused according to the first preset weight to obtain the blink recognition threshold. In the preferred embodiment of the present application, the weight of the global eyelid distance threshold is 2 / 3, and the weight of the local eyelid distance threshold is 1 / 3. Finally, the blink recognition threshold is obtained as follows: Figure 3 Shown by the blue line.

[0104] After obtaining the blink recognition threshold, the eyelid distance in the eyelid distance time series dataset is compared with the blink recognition threshold. The video frame corresponding to the eyelid distance less than the blink recognition threshold is the blink frame, such as Figure 3 The orange line shown is the eyelid distance time series dataset, and the green line is the identified blink frame.

[0105] Similarly, based on the fluctuation of the eyebrow-eye distance time series data set and the abnormal value of the eyebrow-eye movement amplitude, the blink recognition threshold is calculated. Specifically, the global mean and global standard deviation of the eyebrow-eye distance time series data set are calculated, and the global threshold is calculated based on the global mean and global standard deviation of the eyebrow-eye distance. The calculation formula of the global threshold of the eyebrow-eye distance is:

[0106] L m =M mean -C m ×((M std )×(1-3M std ))

[0107] Among them, L m represents the global threshold of eyebrow-eye distance, M mean represents the global mean of eyebrow-eye distance, M std represents the global standard deviation of the eyebrow-eye distance, C m Indicates frown recognition sensitive parameters.

[0108] Select the video frames of the first preset time period to represent the abnormal eyebrow and eye movement amplitude frames, calculate the local mean of the eyebrow and eye distance and the local standard deviation of the eyebrow and eye distance time series data set corresponding to the abnormal eyebrow and eye movement amplitude frames, and calculate the local threshold of the eyebrow and eye distance according to the local mean of the eyebrow and eye distance and the local standard deviation of the eyebrow and eye distance. The calculation formula of the local threshold of the eyebrow and eye distance is:

[0109] L m ′=M local-mean -C m ×((M local-std )×(1-3M local-std ))

[0110] Among them, L m ' represents the local threshold of eyebrow distance, M local-mean represents the local mean of the eyebrow distance, M local-std Represents the local standard deviation of the eyebrow-eye distance.

[0111] After obtaining the global threshold of eyebrow-eye distance and the local threshold of eyebrow-eye distance, the weights of the global threshold of eyebrow-eye distance and the local threshold of eyebrow-eye distance are set as the second preset weight, and the global threshold of eyebrow-eye distance and the local threshold of eyebrow-eye distance are fused according to the second preset weight to obtain the frown recognition threshold. In the preferred embodiment of the present application, the weight of the global threshold of eyebrow-eye distance is 2 / 3, and the weight of the local threshold of eyebrow-eye distance is 1 / 3, and finally the frown recognition threshold is obtained.

[0112] After obtaining the frown recognition threshold, the eyebrow-eye distance in the eyebrow-eye distance time series data set is compared with the frown recognition threshold, and the video frame corresponding to the eyebrow-eye distance smaller than the frown recognition threshold is the frown frame.

[0113] Further, in a preferred embodiment of the present invention, based on the eyelid distance time series data set, the blink frame data of each blink frame is calculated, and the blink frame data includes the maximum eyelid distance, the minimum eyelid distance, the number of complete blinks, the number of incomplete blinks, the minimum blink time, the average blink time, the maximum blink interval, the blink interval time variability, the minimum eye closure time, the average eye closure time, the minimum eye opening time and the average eye opening time. Among them, the maximum eyelid distance and the minimum eyelid distance are as follows Figure 3 The maximum and minimum values ​​of the eyelid distance for each blink in the eyelid distance polyline shown. The video frames in the second preset time period before and after each blink frame are selected to represent the blink-related frames, and the maximum and minimum eyelid distance values ​​corresponding to the blink-related frames are selected from the eyelid distance time series data set.

[0114] The blink time is the duration of each blink. Specifically, the duration of the eyelid distance of each blink-related frame being less than the blink recognition threshold is calculated as the first time series interval, and the first time series interval is used as the blink time of each blink. Figure 3 As shown, the blink time is the duration when the orange line is less than the blue line. The average of each blink time is calculated to obtain the average blink time, and the minimum blink time is determined based on each blink time.

[0115] Furthermore, according to the size of the blink time and the preset blink threshold, the blink frame is judged as a complete blink or an incomplete blink, and the number of complete blinks and incomplete blinks in the blink frame is counted. In the preferred embodiment of the present application, the preset blink threshold is 0.2s. When the blink time is greater than or equal to 0.2s, it is a complete blink, and when the blink time is less than 0.2s, it is an incomplete blink.

[0116] Furthermore, the complete blink time standard deviation and the complete blink time average of the blink time corresponding to the complete blink are calculated, and the ratio of the complete blink time standard deviation to the complete blink time average is used as the blink interval time variability.

[0117] Furthermore, the duration between the minimum values ​​of the eyelid distances in each blink frame is calculated as the second time series interval, the second time series interval is used as the blink interval, and the maximum blink interval is selected from the blink intervals.

[0118] Furthermore, the duration of the eyelid distance from being less than the blink recognition threshold to the minimum eyelid distance in each blink frame is calculated as the third time series interval, that is, Figure 3 The duration of the orange line is less than that of the blue line, and the third time series interval is taken as the eye closing time. The minimum eye closing time is selected from the eye closing time, and the average eye closing time is calculated based on each eye closing time.

[0119] Further, the duration of the eyelid distance from the minimum eyelid distance to the time when it is greater than the blink recognition threshold in each blink frame is calculated as the fourth time series interval, and the fourth time series interval is used as the eye opening time. The minimum eye opening time is selected from the eye opening time, and the average eye opening time is calculated based on each eye opening time.

[0120] Similarly, for frown frames, based on the eye-brow distance time series data set, the frown frame data of each frown frame is calculated, and the frown frame data includes the maximum eye-brow distance, the minimum eye-brow distance, the frown time, the number of frowns, the variability of the frown interval time, the maximum frown time and the minimum frown interval.

[0121] Among them, the maximum eyebrow distance and the minimum eyebrow distance are the maximum and minimum eyebrow distance of each frown. The frown-related frames are represented by selecting video frames in the third preset time period before and after each frown frame, and the maximum eyebrow distance and the minimum eyebrow distance corresponding to the frown-related frames are selected from the eyebrow distance time series data set.

[0122] The frowning time is the duration of each frown. Specifically, the duration of the eyebrow-eye distance of each frown-associated frame that is less than the frown recognition threshold is calculated as the fifth time series interval. The fifth time series interval is used as the frowning time of each frown, and the total number of frowns in the entire historical video information is determined. The maximum frowning time is determined based on each frowning time.

[0123] Furthermore, the frown time standard deviation and the frown time average value corresponding to the frown time are calculated, and the ratio of the frown time standard deviation to the frown time average value is taken as the frown interval time variability.

[0124] Furthermore, the duration between the minimum values ​​of the eyebrow-eye distance in each frowning frame is calculated as the sixth time series interval, the sixth time series interval is used as the frowning interval, and the minimum frowning interval is selected from the frowning intervals.

[0125] Further, the constructed dry eye detection model is trained using blink frame data, frown frame data and red eye probability to obtain the target dry eye detection model. In the preferred embodiment of the present application, the dry eye detection model is constructed using the Catboost framework. Catboost has made a lot of improvements to the processing of category features, and an innovative algorithm that automatically processes category features into numerical features is embedded. Specifically, some statistics are made on the category features, the frequency of occurrence of a certain category feature is calculated, and then hyperparameters are added to generate new numerical features. The use of combined category features can utilize the connection between features, greatly enrich the feature dimension, and enable more efficient training. CatBoost uses the concept of sorting promotion, which is a permutation-driven method that trains the model on a subset of data and calculates the residual on another subset, thereby preventing target leakage and overfitting. CatBoost can be easily integrated with a deep learning framework, supports various features of numbers, classifications or texts, saves time and effort in preprocessing, and can greatly improve the convenience and efficiency of dry eye detection on smartphones in this application.

[0126] The blink frame data, frown frame data and red eye probability are input into the constructed dry eye detection model for training to obtain a target dry eye detection model. During the real-time detection process, the user to be identified uploads real-time video information via a smartphone or uses the front camera of a smartphone to record real-time video information. After information extraction, the blink frame data, real-time frown frame data and real-time red eye probability are obtained. The blink frame data, real-time frown frame data and real-time red eye probability are input into the target dry eye detection model to obtain the dry eye detection result.

[0127] In a preferred embodiment of the present invention, historical video information containing human faces is collected, and the eyelid point time series data, eyebrow and eye point time series data and eye redness probability in the historical video information are identified frame by frame; the eyelid distance is calculated frame by frame according to the eyelid point time series data to construct an eyelid distance time series data set, and the eyebrow and eye distance is calculated frame by frame according to the eyebrow and eye point time series data to construct an eyebrow and eye distance time series data set; based on the fluctuation of the eyelid distance time series data set and the abnormal value of the blink amplitude, the blink recognition threshold is calculated, and based on the fluctuation of the eyebrow and eye distance time series data set and the abnormal value of the eyebrow and eye movement amplitude, the frown recognition threshold is calculated; based on the fluctuation of the eyelid distance time series data set and the abnormal value of the eyebrow and eye movement amplitude, the frown recognition threshold is calculated According to the size relationship of the eye recognition threshold, the blink frame is identified, and the frown frame is identified according to the size relationship between the eyebrow-eye distance time series data set and the frown recognition threshold; based on the eyelid distance time series data set, the blink frame data of each blink frame is calculated, and based on the eyebrow-eye distance time series data set, the frown frame data of each frown frame is calculated; the constructed dry eye detection model is trained using the blink frame data, frown frame data and eye redness probability to obtain the target dry eye detection model; in the real-time detection process, the real-time blink frame data, real-time frown frame data and real-time eye redness probability extracted from the collected real-time video information are input into the target dry eye detection model to obtain the dry eye detection result. The dry eye detection method provided in the present application can achieve the effect of providing a convenient, practical, accurate and large-scale popularization and application of dry eye detection method, which helps to improve the quality of life of patients and promote the efficient allocation of limited medical resources.

[0128] Accordingly, if Figure 4 As shown, based on a dry eye detection method, an embodiment of the present invention further provides a dry eye detection system to implement the dry eye detection method disclosed in the embodiment of the present invention, including: a data acquisition processing unit 1, a first calculation unit 2, a recognition threshold calculation unit 3, a recognition unit 4, a second calculation unit 5, a model training unit 6 and a real-time detection unit 7;

[0129] The data acquisition processing unit 1 is used to acquire historical video information containing human faces, and identify eyelid point time series data, eyebrow point time series data and eye redness probability in the historical video information frame by frame;

[0130] The first calculation unit 2 is used to calculate the eyelid distance frame by frame according to the eyelid point time series data to construct an eyelid distance time series data set, and calculate the eyebrow-eye distance frame by frame according to the eyebrow-eye point time series data to construct an eyebrow-eye distance time series data set;

[0131] The recognition threshold calculation unit 3 is used to calculate the blink recognition threshold based on the fluctuation of the eyelid distance time series data set and the abnormal value of the blink amplitude, and calculate the frown recognition threshold based on the fluctuation of the eyebrow-eye distance time series data set and the abnormal value of the eyebrow-eye movement amplitude;

[0132] The recognition unit 4 is used to recognize the blink frame according to the size relationship between the eyelid distance time series data set and the blink recognition threshold, and recognize the frown frame according to the size relationship between the eyebrow distance time series data set and the frown recognition threshold;

[0133] The second calculation unit 5 is used to calculate the blink frame data of each blink frame based on the eyelid distance time series data set, and calculate the frown frame data of each frown frame based on the eyebrow-eye distance time series data set;

[0134] The model training unit 6 is used to train the constructed dry eye detection model using the blink frame data, the frown frame data and the red eye probability to obtain a target dry eye detection model;

[0135] The real-time detection unit 7 is used to input the real-time blink frame data, real-time frown frame data and real-time red eye probability extracted from the collected real-time video information into the target dry eye detection model during the real-time detection process to obtain the dry eye detection result.

[0136] For the specific definition of a dry eye detection system, please refer to the above-mentioned definition of a dry eye detection method, which will not be repeated here. A person of ordinary skill in the art will appreciate that the modules and steps described in conjunction with the embodiments disclosed in the present invention can be implemented in hardware, software, or a combination of both. Whether these functions are performed in hardware or software depends on the specific application and design constraints of the technical solution. Professional and technical personnel can use different methods to implement the described functions for each specific application, but such implementation should not be considered to exceed the scope of the present invention.

[0137] like Figure 5 As shown, an embodiment of the present invention provides a computer device, including a processor, a memory, and a computer program stored in the memory and configured to be executed by the processor. When the processor executes the computer program, the steps in the above-mentioned dry eye detection method embodiment are implemented, for example Figure 1 Steps S1 to S7 described in .

[0138] Those skilled in the art will understand that the Figure 5It is only an example of a computer device and does not constitute a limitation of the computer device. It may include more or fewer components than shown in the figure, or a combination of certain components, or different components. For example, the computer device may also include input and output devices, network access devices, buses, etc.

[0139] The processor may be a central processing unit (CPU), or other general-purpose processors, digital signal processors (DSP), application-specific integrated circuits (ASIC), field-programmable gate arrays (FPGA) or other programmable logic devices, discrete gate or transistor logic devices, discrete hardware components, etc. A general-purpose processor may be a microprocessor or any conventional processor, etc. The processor is the control center of the computer device, and uses various interfaces and lines to connect various parts of the entire computer device.

[0140] The memory can be used to store the computer program and / or module, and the processor realizes various functions of the computer device by running or executing the computer program and / or module stored in the memory, and calling the data stored in the memory. The memory can mainly include a program storage area and a data storage area, wherein the program storage area can store an operating system, an application required for at least one function (such as a sound playback function, an image playback function, etc.), etc.; the data storage area can store data created according to the use of the mobile phone (such as audio data, a phone book, etc.), etc. In addition, the memory can include a high-speed random access memory, and can also include a non-volatile memory, such as a hard disk, a memory, a plug-in hard disk, a smart memory card (Smart Media Card, SMC), a secure digital (Secure Digital, SD) card, a flash card (Flash Card), at least one disk storage device, a flash memory device, or other volatile solid-state storage devices.

[0141] Wherein, if the module integrated in the computer device is implemented in the form of a software functional unit and sold or used as an independent product, it can be stored in a computer-readable storage medium. Based on such an understanding, the present invention implements all or part of the processes in the above-mentioned embodiment method, and can also be completed by instructing the relevant hardware through a computer program. The computer program can be stored in a computer-readable storage medium, and the computer program can implement the steps of the above-mentioned various method embodiments when executed by the processor. Wherein, the computer program includes computer program code, and the computer program code can be in source code form, object code form, executable file or some intermediate form, etc. The computer-readable medium may include: any entity or device capable of carrying the computer program code, recording medium, U disk, mobile hard disk, disk, optical disk, computer memory, read-only memory (ROM, Read-Only Memory), random access memory (RAM, Random Access Memory), electrical carrier signal, telecommunication signal and software distribution medium, etc.

[0142] A person skilled in the art can understand that all or part of the processes in the above-mentioned embodiments can be implemented by instructing the relevant hardware through a computer program, and the program can be stored in a computer-readable storage medium, and when the program is executed, it can include the processes of the embodiments of the above-mentioned methods. The storage medium can be a disk, an optical disk, a read-only memory (ROM) or a random access memory (RAM), etc.

[0143] Accordingly, an embodiment of the present invention provides a computer-readable storage medium, wherein the computer-readable storage medium includes a stored computer program, wherein when the computer program is executed, the device where the computer-readable storage medium is located is controlled to perform the steps in the dry eye detection method of the above embodiment, for example Figure 1 Steps S1 to S7 described in .

[0144] A dry eye detection method, system, device and medium provided in this embodiment are used to solve the technical problem of improving the convenience, practicality and accuracy of dry eye detection. Collect historical video information containing human faces, and identify the eyelid point time series data, eyebrow and eye point time series data and red eye probability in the historical video information frame by frame; calculate the eyelid distance frame by frame according to the eyelid point time series data to construct an eyelid distance time series data set, and calculate the eyebrow and eye distance frame by frame according to the eyebrow and eye point time series data to construct an eyebrow and eye distance time series data set; calculate the blink recognition threshold based on the fluctuation of the eyelid distance time series data set and the abnormal value of the blink amplitude, and calculate the frown recognition threshold based on the fluctuation of the eyebrow and eye distance time series data set and the abnormal value of the eyebrow and eye movement amplitude; calculate the eyelid distance time series data set according to the fluctuation of the eyelid distance time series data set and the abnormal value of the eyebrow and eye movement amplitude According to the size relationship, the blink frame is identified, and the frown frame is identified according to the size relationship between the eyebrow-eye distance time series data set and the frown recognition threshold; based on the eyelid distance time series data set, the blink frame data of each blink frame is calculated, and based on the eyebrow-eye distance time series data set, the frown frame data of each frown frame is calculated; the constructed dry eye detection model is trained using the blink frame data, frown frame data and eye redness probability to obtain the target dry eye detection model; in the real-time detection process, the real-time blink frame data, real-time frown frame data and real-time eye redness probability extracted from the collected real-time video information are input into the target dry eye detection model to obtain the dry eye detection result. The dry eye detection method provided in the present application can achieve the effect of providing a convenient, practical, accurate and large-scale popularization and application of dry eye detection method, which helps to improve the quality of life of patients and promote the efficient allocation of limited medical resources.

[0145] Each embodiment in this specification is described in a progressive manner, and the same or similar parts of each embodiment can be directly referred to each other, and each embodiment focuses on the differences from other embodiments. In particular, for the system embodiment, since it is basically similar to the method embodiment, the description is relatively simple, and the relevant parts can be referred to the partial description of the method embodiment. It should be noted that the technical features of the above-mentioned embodiments can be combined arbitrarily. In order to make the description concise, not all possible combinations of the technical features in the above-mentioned embodiments are described. However, as long as there is no contradiction in the combination of these technical features, they should be considered to be within the scope of this specification.

[0146] The above-described embodiments only express several preferred implementations of the present invention, and the descriptions thereof are relatively specific and detailed, but they cannot be construed as limiting the scope of the invention. It should be pointed out that, for those skilled in the art, several improvements and substitutions can be made without departing from the technical principles of the present invention, and these improvements and substitutions should also be regarded as the protection scope of the present invention. Therefore, the protection scope of the present invention shall be subject to the protection scope of the claims.

Claims

1. A method for detecting dry eye, characterized in that: The method comprises: Collect historical video information containing human faces, and identify eyelid point time series data, eyebrow point time series data and eye redness probability in the historical video information frame by frame; Calculating the eyelid distance frame by frame according to the eyelid point time series data to construct an eyelid distance time series data set, and calculating the eyebrow-eye distance frame by frame according to the eyebrow-eye point time series data to construct an eyebrow-eye distance time series data set; Based on the fluctuation of the eyelid distance time series data set and the abnormal value of the blink amplitude, a blink recognition threshold is calculated; based on the fluctuation of the eyebrow-eye distance time series data set and the abnormal value of the eyebrow-eye movement amplitude, a frown recognition threshold is calculated; According to the size relationship between the eyelid distance time series data set and the blink recognition threshold, the blink frame is identified; according to the size relationship between the eyebrow-eye distance time series data set and the frown recognition threshold, the frown frame is identified; Based on the eyelid distance time series data set, the blink frame data of each blink frame is calculated; based on the eyebrow-eye distance time series data set, the frown frame data of each frown frame is calculated; Using the blink frame data, the frown frame data and the red eye probability, the constructed dry eye detection model is trained to obtain a target dry eye detection model; During the real-time detection process, the real-time blink frame data, real-time frown frame data and real-time red eye probability extracted from the collected real-time video information are input into the target dry eye detection model to obtain the dry eye detection result.

2. The dry eye detection method according to claim 1, characterized in that: The step of identifying the eyelid point time series data, eyebrow and eye point time series data and eye redness probability in the historical video information frame by frame includes: The dlib model is used to identify 68 standard points of facial features in the video information frame by frame: Selecting eyelid point data and eyebrow point data from the 68 points; Combining the time series data of the video information of each frame with the eyelid point data to construct the eyelid point time series data; Combining the time series data with the eyebrow and eye point data to construct the eyebrow and eye point time series data; The video information of each frame is input into the trained red eye recognition model to obtain the red eye probability.

3. The dry eye detection method according to claim 1, characterized in that: The eyelid distance is defined as: the difference between the upper eyelid ordinate and the lower eyelid ordinate; The eyebrow-eye distance is defined as the difference between the ordinate of the highest point of the eyebrow and the ordinate of the inner corner of the eye.

4. The dry eye detection method according to claim 1, characterized in that: The step of calculating the blink recognition threshold based on the fluctuation of the eyelid distance time series data set and the blink amplitude abnormal value comprises: Calculating the global mean of eyelid distance and the global standard deviation of eyelid distance of the eyelid distance time series data set, and calculating the global threshold of eyelid distance according to the global mean of eyelid distance and the global standard deviation of eyelid distance; The calculation formula of the eyelid distance global threshold is: L l =L mean -C l ×((L std )×(1-3L std )) Among them, L l represents the global threshold of eyelid distance, L mean represents the global mean of eyelid distance, L std represents the global standard deviation of eyelid distance, C l Indicates the sensitive parameters of blink recognition; Select video frames in a first preset time period before and after the historical video information to represent abnormal blink amplitude frames, calculate the local mean of eyelid distance and the local standard deviation of eyelid distance of the eyelid distance time series data set corresponding to the abnormal blink amplitude frames, and calculate the local threshold of eyelid distance according to the local mean of eyelid distance and the local standard deviation of eyelid distance; The calculation formula of the eyelid distance local threshold is: L l ′=L local-mean -C l ×((L local-std )×(1-3L local-std )) Among them, L l ' represents the local threshold of eyelid distance, L local-mean represents the local mean of eyelid distance, L local-std represents the local standard deviation of eyelid distance; The eyelid distance global threshold and the eyelid distance local threshold are fused according to a first preset weight to obtain the blink recognition threshold.

5. The dry eye detection method according to claim 1, characterized in that: The step of calculating the frown recognition threshold based on the fluctuation of the eyebrow-eye distance time series data set and the eyebrow-eye movement amplitude abnormal value comprises: Calculate the global mean of the eyebrow-eye distance and the global standard deviation of the eyebrow-eye distance time series data set, and calculate the global threshold of the eyebrow-eye distance according to the global mean of the eyebrow-eye distance and the global standard deviation of the eyebrow-eye distance; The calculation formula of the eyebrow-eye distance global threshold is: L m =M mean -C m ×((M std )×(1-3M std )) Among them, L m represents the global threshold of eyebrow-eye distance, M mean represents the global mean of eyebrow-eye distance, M std represents the global standard deviation of the eyebrow-eye distance, C m Indicates frown recognition sensitive parameters; Select video frames in a first preset time period before and after the historical video information to represent frames with abnormal eyebrow and eye movement amplitude, calculate the local mean of the eyebrow and eye distance and the local standard deviation of the eyebrow and eye distance time series data set corresponding to the frames with abnormal eyebrow and eye movement amplitude, and calculate the local threshold of the eyebrow and eye distance according to the local mean of the eyebrow and eye distance and the local standard deviation of the eyebrow and eye distance; The calculation formula of the local threshold of the eyebrow-eye distance is: L m ′=M local-mean -C m ×((M local-std )×(1-3M local-std )) Among them, L m ' represents the local threshold of eyebrow distance, M local-mean represents the local mean of the eyebrow distance, M local-std represents the local standard deviation of the eyebrow-eye distance; The eyebrow-eye distance global threshold and the eyebrow-eye distance local threshold are fused according to a second preset weight to obtain the frown recognition threshold.

6. The dry eye detection method according to claim 1, characterized in that: The step of calculating the blink frame data of each blink frame based on the eyelid distance time series data set comprises: Selecting video frames within a second preset time period before and after each of the blink frames to represent blink-related frames, and selecting the maximum eyelid distance and the minimum eyelid distance corresponding to the blink-related frames from the eyelid distance time series data set; Calculate the first time series interval in which the eyelid distance of each blink-related frame is less than the blink recognition threshold, and use the first time series interval as the blink time; Calculating the mean of the blink time to obtain an average blink time; According to the size of the blink time and a preset blink threshold, determining whether the blink frame is a complete blink or an incomplete blink, and counting the number of complete blinks and the number of incomplete blinks in the blink frame; Calculating a complete blink time standard deviation and a complete blink time average value of the blink time corresponding to the complete blink, and taking the ratio of the complete blink time standard deviation to the complete blink time average value as the blink interval time variability; Determine a minimum blink time value according to each of the blink times; Calculate the blink interval according to the second time series interval between the minimum values ​​of the eyelid distances of each blink frame, and select the maximum blink interval from the blink intervals; Calculate a third time series interval of the eyelid distance of each blink frame from being less than the blink recognition threshold to the minimum value of the eyelid distance, and use the third time series interval as the eye closing time; Calculate the eyelid distance of each of the blink frames from the minimum eyelid distance to a fourth time series interval greater than the blink recognition threshold, and use the fourth time series interval as the eye-opening time; Selecting a minimum eye closing time from the eye closing times, and calculating an average eye closing time based on each of the eye closing times; Selecting a minimum eye-opening time from the eye-opening times, and calculating an average eye-opening time based on each of the eye-opening times; The maximum eyelid distance, the minimum eyelid distance, the number of complete blinks, the number of incomplete blinks, the minimum blink time, the average blink time, the maximum blink interval, the blink interval time variability, the minimum eye closing time, the average eye closing time, the minimum eye opening time and the average eye opening time are taken as blink frame data.

7. The dry eye detection method according to claim 1, characterized in that: The step of calculating the frown frame data of each frown frame based on the eyebrow distance time series data set comprises: Select video frames within a third preset time period before and after each frown frame to represent frown-related frames, and select the maximum eyebrow-eye distance and the minimum eyebrow-eye distance corresponding to the frown-related frames from the eyebrow-eye distance time series data set; Calculate the fifth time series interval at which the eyebrow-eye distance of each frown-associated frame is less than the frown recognition threshold, and determine the frown time and frown frequency according to the fifth time series interval; Calculating the frown time standard deviation and the frown time average value of the frown time, and taking the ratio of the frown time standard deviation to the frown time average value as the frown interval time variability; Determine a maximum frown time according to each of the frown times; Calculate the frown interval according to the sixth time series interval between the minimum values ​​of the eyebrow-eye distance of each frown frame, and select the minimum frown interval from the frown intervals; The maximum eyebrow-eye distance, the minimum eyebrow-eye distance, the frowning time, the number of frowning times, the frowning interval time variability, the maximum frowning time and the minimum frowning interval are used as frowning frame data.

8. A dry eye detection system, implementing the dry eye detection method according to any one of claims 1 to 7, characterized in that: The system comprises: a data acquisition processing unit, a first calculation unit, a recognition threshold calculation unit, a recognition unit, a second calculation unit, a model training unit and a real-time detection unit; The data acquisition processing unit is used to acquire historical video information containing human faces, and identify eyelid point time series data, eyebrow point time series data and eye redness probability in the historical video information frame by frame; The first calculation unit is used to calculate the eyelid distance frame by frame according to the eyelid point time series data to construct an eyelid distance time series data set, and calculate the eyebrow-eye distance frame by frame according to the eyebrow-eye point time series data to construct an eyebrow-eye distance time series data set; The recognition threshold calculation unit is used to calculate the blink recognition threshold based on the fluctuation of the eyelid distance time series data set and the abnormal value of the blink amplitude, and calculate the frown recognition threshold based on the fluctuation of the eyebrow-eye distance time series data set and the abnormal value of the eyebrow-eye movement amplitude; The recognition unit is used to recognize the blink frame according to the size relationship between the eyelid distance time series data set and the blink recognition threshold, and recognize the frown frame according to the size relationship between the eyebrow distance time series data set and the frown recognition threshold; The second calculation unit is used to calculate the blink frame data of each blink frame based on the eyelid distance time series data set, and calculate the frown frame data of each frown frame based on the eyebrow-eye distance time series data set; The model training unit is used to train the constructed dry eye detection model using the blink frame data, the frown frame data and the red eye probability to obtain a target dry eye detection model; The real-time detection unit is used to input the real-time blink frame data, real-time frown frame data and real-time red eye probability extracted from the collected real-time video information into the target dry eye detection model during the real-time detection process to obtain the dry eye detection result.

9. A computer device, characterized in that: The computer device includes a memory, a processor and a transceiver, which are connected via a bus; the memory is used to store a set of computer program instructions and data, and transmit the stored data to the processor, and the processor executes the program instructions stored in the memory to perform the dry eye detection method as described in any one of claims 1 to 7.

10. A computer-readable storage medium, characterized in that: The computer-readable storage medium stores a computer program, and when the computer program is executed, the dry eye detection method according to any one of claims 1 to 7 is implemented.

Images