Image transmission method for remote detection of ophthalmology department

Abstract

The invention relates to the technical field of intelligent sensors, in particular to an image transmission method for remote detection of the ophthalmology department. According to the method, multiple frames of fundus grayscale images are continuously collected in preset exposure time, the detail definition degree in the images is evaluated according to the grayscale distribution condition, and the ambient light suitability degree of the current collection environment is obtained; and then the micro-motion degree of the eyeballs is evaluated according to the difference between the fundus grayscale images between the adjacent frames, so that the preset exposure time is adjusted in combination with the ambient light suitability, the high-quality fundus grayscale images are obtained for remote transmission, and the remote transmission effect of the fundus grayscale images is improved.

Description

Technical Field

[0001] The present invention relates to the technical field of intelligent sensors, and in particular to an image transmission method for remote ophthalmic detection. Background Art

[0002] With the development and advancement of Internet communications, the remote transmission technology of ophthalmic imaging data has become more stable and efficient, and a large number of ophthalmic images can be transmitted in a relatively short period of time.

[0003] Since remote transmission will cause a certain degree of loss in ophthalmic images, which may affect the relevant medical staff's analysis and evaluation of the detailed features of the received images, the quality of ophthalmic images is crucial during the acquisition stage; currently, fundus images are mainly collected through equipment and technology such as fundus cameras, but for some remote areas, their medical equipment technology is relatively backward. During the acquisition of fundus images, the exposure time of the fundus camera needs to be manually adjusted; inappropriate exposure time will affect the length of time the smart sensor in the fundus camera receives light, resulting in overexposure or too dark images, blurring the detailed features in the ophthalmic images, and then resulting in poor acquisition quality of the fundus images, which in turn affects the subsequent remote transmission effect. Summary of the invention

[0004] In order to solve the technical problem that inappropriate exposure time leads to poor fundus image acquisition quality and thus poor image remote transmission effect, the purpose of the present invention is to provide an image transmission method for ophthalmic remote detection, and the technical solution adopted is as follows:

[0005] Acquire at least two consecutive frames of fundus grayscale images with a preset exposure time;

[0006] Acquire a target area and a background area in each frame of the fundus grayscale image, wherein the target area is a fundus structure area; and evaluate the suitability of ambient light when acquiring all the fundus grayscale images according to the grayscale distribution difference between the target area and the background area in each frame of the fundus grayscale image;

[0007] Between the fundus grayscale images of adjacent frames, the degree of micro-movement of the eyeball when all the fundus grayscale images are acquired is evaluated according to the similarity of the grayscale differences between all pixels at the same position;

[0008] The preset exposure time is adjusted according to the suitability of the ambient light and the degree of micro-movement to obtain an optimal exposure time; the fundus grayscale image to be transmitted is collected using the optimal exposure time and transmitted remotely.

[0009] Furthermore, the method for obtaining the suitability of ambient light includes:

[0010] In each frame of the fundus grayscale image, obtaining a first parameter according to a difference between corresponding grayscale distributions of the target area and the background area;

[0011] Acquire a second parameter according to the difference between the number of corresponding pixels at each same gray level between the target area and the background area;

[0012] According to the first parameter and the second parameter, obtaining a distinction difficulty coefficient between the target area and the background area in each frame of the fundus grayscale image, wherein the first parameter and the second parameter are both positively correlated with the distinction difficulty coefficient;

[0013] All the fundus grayscale images are negatively normalized with respect to the minimum value of the differentiation difficulty coefficient, so as to obtain the suitability of the ambient light when all the fundus grayscale images are obtained.

[0014] Furthermore, the method for obtaining the first parameter includes:

[0015] In each frame of the fundus grayscale image, a grayscale set of the target area and a grayscale set of the background area are respectively obtained, and the Jaccard similarity coefficient between the two grayscale sets is used as the first parameter.

[0016] Furthermore, the method for obtaining the micro-motion degree includes:

[0017] Acquire a difference image between fundus grayscale images of adjacent frames;

[0018] Obtaining the overall micro-motion coefficient according to the discrete degree of the grayscale values ​​corresponding to all pixels in the difference image;

[0019] In the difference image, all target points are screened out from all pixels according to the grayscale distribution of each pixel within a preset range; and the local micro-motion coefficient is obtained according to the grayscale difference between each target point and other pixels within the corresponding preset range;

[0020] The micromotion degree is obtained according to the overall micromotion coefficient and the local micromotion coefficient, and both the overall micromotion coefficient and the local micromotion coefficient are positively correlated with the micromotion degree.

[0021] Furthermore, the method for acquiring the target point includes:

[0022] Sort all the pixels in the difference image in descending order according to the grayscale value, and select the first preset number of pixels in the sequence as preliminary screening points;

[0023] At least two target points are obtained from the preliminary screening points, wherein the distance between any two target points is greater than a preset length value.

[0024] Furthermore, the method for obtaining the preset exposure time includes:

[0025] The median value of the preset standard exposure time range is used as the preset exposure time.

[0026] Furthermore, the method for obtaining the optimal exposure time includes:

[0027] Obtaining a total length of an interval within a preset standard exposure time range, and using a preset proportion of the total length of the interval as an adjustment reference value;

[0028] The negative correlation of the ambient light suitability is normalized as the weight of the adjustment reference value to obtain an upward adjustment range of the preset exposure time;

[0029] Taking the micro-motion degree as the weight of the adjustment reference value, obtaining a downward adjustment amplitude of the preset exposure time;

[0030] The optimal exposure time is obtained according to the upward adjustment range and the downward adjustment range of the preset exposure time, wherein the upward adjustment range is positively correlated with the optimal exposure time, and the downward adjustment range is negatively correlated with the optimal exposure time.

[0031] Furthermore, the method for obtaining the target area and the background area in each frame of the fundus grayscale image includes:

[0032] Each frame of the fundus grayscale image is segmented based on a convolutional neural network to obtain the target area and the background area.

[0033] Furthermore, the remote transmission method of the fundus grayscale image to be transmitted includes:

[0034] A secure transmission channel is established through VPN technology, and the SSL encryption protocol is used to encrypt the fundus grayscale image to be transmitted and transmit it remotely.

[0035] Furthermore, the preset ratio is one-half.

[0036] The present invention has the following beneficial effects:

[0037] The present invention first obtains continuous frame fundus grayscale images with a preset exposure time; further obtains the target area and background area in each frame of the fundus grayscale image, the more obvious the grayscale difference between the target area and the background area is, the higher the clarity or recognizability of the detail texture or structure corresponding to the target area is, and the higher the quality of the acquired fundus grayscale image is, which indirectly reflects that the more suitable the ambient light is when acquiring the fundus grayscale image, the lower the need to adjust the exposure time is; between adjacent frames of fundus grayscale images, the degree of micro-movement of the eyeball when acquiring all fundus grayscale images is evaluated according to the similarity of the grayscale differences between all pixels at the same position, the greater the degree of micro-movement of the eyeball, the need to appropriately reduce the preset exposure time to reduce motion blur; and then adjust the preset exposure time according to the suitability of the ambient light and the degree of micro-movement to obtain the optimal exposure time; use the optimal exposure time to acquire the fundus grayscale image to be transmitted, and transmit it remotely. The present invention continuously collects multiple frames of fundus grayscale images with a preset exposure time, compares and analyzes the detail clarity and grayscale distribution differences of adjacent frames of fundus grayscale images, evaluates the influence of ambient light and eye movement on the fundus grayscale image collection during collection, and thus adjusts the preset exposure time to obtain high-quality fundus grayscale images for remote transmission. By improving the collection quality of ophthalmic images, even if there is a certain degree of loss in the remote transmission process, it can ensure that relevant ophthalmic technicians can analyze and evaluate fundus details to improve the transmission effect. BRIEF DESCRIPTION OF THE DRAWINGS

[0038] In order to more clearly illustrate the technical solutions and advantages in the embodiments of the present invention or the prior art, the drawings required for use in the embodiments or the prior art descriptions are briefly introduced below. Obviously, the drawings described below are only some embodiments of the present invention. For ordinary technicians in this field, other drawings can be obtained based on these drawings without paying creative work.

[0039] Figure 1 A flowchart of an image transmission method for ophthalmic remote detection provided by one embodiment of the present invention;

[0040] Figure 2 Two consecutive frames of fundus grayscale images provided by an embodiment of the present invention;

[0041] Figure 3 A flow chart of a method for obtaining the suitability of ambient light provided by an embodiment of the present invention;

[0042] Figure 4 A flowchart of a method for acquiring the degree of micro-motion provided by an embodiment of the present invention. DETAILED DESCRIPTION

[0043] In order to further explain the technical means and effects adopted by the present invention to achieve the predetermined invention purpose, the following is a detailed description of the image transmission method for remote ophthalmic detection proposed by the present invention, its specific implementation, structure, features and effects, in conjunction with the accompanying drawings and preferred embodiments. In the following description, different "one embodiment" or "another embodiment" does not necessarily refer to the same embodiment. In addition, specific features, structures or characteristics in one or more embodiments may be combined in any suitable form.

[0044] Unless defined otherwise, all technical and scientific terms used herein have the same meaning as commonly understood by one of ordinary skill in the art to which this invention belongs.

[0045] The specific scheme of the image transmission method for ophthalmic remote detection provided by the present invention is described in detail below with reference to the accompanying drawings.

[0046] The present invention continuously collects multiple frames of fundus grayscale images with a preset exposure time, compares and analyzes the detail clarity and grayscale distribution differences of adjacent frames of fundus grayscale images, evaluates the influence of ambient light and eye movement on fundus grayscale image collection during collection, and thus adjusts the preset exposure time to obtain high-quality fundus grayscale images for remote transmission, thereby improving the transmission effect.

[0047] See also Figure 1 , which shows a flow chart of an image transmission method for ophthalmic remote detection provided by an embodiment of the present invention, specifically comprising:

[0048] Step S1, acquiring at least two consecutive frames of fundus grayscale images with a preset exposure time.

[0049] In one embodiment of the present invention, a professional fundus camera is first used to continuously capture at least two frames of fundus images with a preset exposure time, so as to subsequently compare and analyze the detail clarity and grayscale distribution differences of adjacent frames of fundus images; wherein the median of the interval of the preset standard exposure time range is used as the preset exposure time; and then each frame of fundus grayscale image is grayscaled to highlight the fundus details and contrast for subsequent analysis. Please refer to Figure 2 , which shows two consecutive frames of fundus grayscale images provided by an embodiment of the present invention.

[0050] It should be noted that the preset standard exposure time range refers to the professional guidance range of the fundus camera when collecting fundus grayscale images, which is usually between 1 / 4 second and 1 / 60 second, that is, the preset exposure time is 7 / 30; the professional guidance range of different models of fundus cameras may not be consistent. Implementers can set the preset standard exposure time range according to ophthalmologists or professional materials to determine the preset exposure time.

[0051] It should be noted that obtaining fundus images and performing grayscale processing is an existing technology well known to those skilled in the art and will not be elaborated here; when obtaining fundus grayscale images, it should be done in a dark room as much as possible to ensure that the pupil is large enough; the fundus images taken by some fundus cameras are already grayscale images and do not require additional grayscale processing.

[0052] Step S2, obtaining the target area and the background area in each frame of fundus grayscale image, where the target area is the fundus structure area; and evaluating the suitability of the ambient light when obtaining all fundus grayscale images based on the grayscale distribution difference between the target area and the background area in each frame of fundus grayscale image.

[0053] In order to evaluate the clarity of the detailed structure in the fundus grayscale image, the embodiment of the present invention first obtains the target area and background area in each frame of the fundus grayscale image. The target area refers to the retinal blood vessels, optic disc, macula, and abnormal pathological structures such as hemorrhage or pigmentation and other important detailed structures in the fundus grayscale image; the background area refers to other relatively uniform areas in the fundus grayscale image that lack obvious structures or textures except the target area.

[0054] Preferably, in one embodiment of the present invention, considering that the convolutional neural network can automatically learn image features to capture key information, each frame of the fundus grayscale image is segmented based on the convolutional neural network to obtain the target area and the background area.

[0055] It should be noted that the training and application of convolutional neural networks are already existing technologies well known to those skilled in the art and will not be elaborated here. In other embodiments of the present invention, implementers may also use other deep learning models such as U-Net, Mask R-CNN, etc. to segment the fundus grayscale image and obtain the target area and background area, which will not be elaborated here.

[0056] In another embodiment of the present invention, the implementer may also use an edge detection algorithm to obtain the detailed structure of the fundus grayscale image, and use all detected edge corresponding areas as target areas, and the remaining non-edge areas as background areas. Edge detection is an existing technology well known to those skilled in the art and will not be described in detail here.

[0057] In other embodiments of the present invention, the implementer may also analyze the grayscale of the pixel points of the detail structure and the background part, and set a segmentation threshold to obtain the target area and the background area, which will not be described in detail here.

[0058] Taking into account that the more obvious the grayscale difference between the target area and the background area in the fundus grayscale image, the higher the clarity or recognizability of the detail texture or structure corresponding to the target area, and the higher the quality of the collected fundus grayscale image, which indirectly reflects that the ambient light when collecting the fundus grayscale image is more suitable; therefore, after obtaining the target area and the background area in each frame of the fundus grayscale image, the embodiment of the present invention evaluates the suitability of the ambient light when obtaining all the fundus grayscale images according to the grayscale distribution difference between the target area and the background area in each frame of the fundus grayscale image. Suitable ambient light can make the detail texture presented in the fundus image clearer and more accurate, which is convenient for the subsequent transmission of high-quality fundus grayscale images and the evaluation of abnormal lesions.

[0059] Preferably, in one embodiment of the present invention, the method for obtaining the suitability of ambient light includes:

[0060] See also Figure 3 , which shows a flow chart of a method for obtaining the suitability of ambient light provided by an embodiment of the present invention.

[0061] Step S301 : obtaining a first parameter according to the difference between the grayscale distributions of the target area and the background area in each frame of the fundus grayscale image.

[0062] Considering that the greater the difference in the distribution of gray levels corresponding to the target area and the background area, the higher the contrast between the target area and the background area and the recognizability of the texture details corresponding to the target area; the first parameter uses the difference in the overall gray level distribution to evaluate the clarity of the fundus grayscale image, reflecting the difficulty of distinguishing between the target area and the background area, which facilitates the subsequent evaluation of the suitability of ambient light.

[0063] Preferably, in one embodiment of the present invention, the method for obtaining the first parameter includes:

[0064] In each frame of fundus grayscale image, a grayscale set of the target area and a grayscale set of the background area are obtained respectively, and the Jaccard similarity coefficient between the two grayscale sets is used as the first parameter.

[0065] As an example, the union and intersection of the grayscales corresponding to the target area and the background area are obtained respectively; the ratio between the number of grayscales corresponding to the union and the intersection is used as the Jaccard index, i.e., the first parameter. The smaller the intersection, the greater the difference in the grayscales corresponding to the two, and the less interference the background area causes to the target area; the larger the union, the wider the overall grayscale range of the fundus grayscale image, and the more clearly the detailed structure in the fundus grayscale image can be expressed; the greater the difference between the grayscale distributions corresponding to the target area and the background area, the smaller the corresponding ratio, and the smaller the first parameter.

[0066] In another embodiment of the present invention, the grayscale histograms of all pixels corresponding to the target area and the background area are obtained respectively, and the area differences of the columns corresponding to the same grayscale level in the two grayscale histograms are compared. After accumulating all area differences, negative correlation mapping is performed to obtain the first parameter; the larger the accumulated area, the greater the grayscale distribution difference, and the smaller the corresponding first parameter. The negative correlation mapping method used is specifically to use the accumulated sum of the area differences as an exponential function e with the natural constant e as the base. -x The index x in , to achieve negative correlation mapping, can also be used by other means, which will not be described in detail.

[0067] Step S302, obtaining a second parameter according to the difference between the number of corresponding pixels at each same gray level between the target area and the background area.

[0068] Considering that a greater total number of the target area and the background area corresponding to the same gray level does not necessarily mean a smaller difference in gray level distribution, it is also necessary to evaluate in combination with the number of corresponding pixels at each same gray level; when the total number of corresponding pixels in the background area is much smaller than the total number of corresponding pixels in the target area at the same gray level, it means that the interference effect of the background area on the target area is smaller; the second parameter further combines the difference in the number of pixels at the same gray level to evaluate the clarity of the fundus gray image, reflecting the difficulty of distinguishing between the target area and the background area, which facilitates the subsequent evaluation of the suitability of ambient light.

[0069] In one embodiment of the present invention, the calculation formula of the second parameter is specifically expressed as:

[0070] Where D2 is the second parameter; A is the symbol of the target area; B is the symbol of the background area; m is the number of the target area and the background area with the same gray level; M(A∩B) is the total number of the target area and the background area with the same gray level; q(B m ) is the total number of pixels in the background area at the mth gray level; q(A m ) is the total number of pixels in the target area at the mth gray level.

[0071] The smaller the second parameter is, the easier it is to distinguish between the target area and the background area, and the clearer the fundus grayscale image is.

[0072] Step S303, obtaining the difficulty coefficient of distinguishing between the target area and the background area in each frame of fundus grayscale image according to the first parameter and the second parameter; the first parameter and the second parameter are both positively correlated with the difficulty coefficient of distinguishing.

[0073] In one embodiment of the present invention, the first parameter and the second parameter are multiplied and combined to obtain the difficulty coefficient of distinguishing between the target area and the background area in each frame of the fundus grayscale image. The smaller the first parameter or the second parameter is, the smaller the difficulty coefficient of distinguishing is, and the higher the clarity of the corresponding fundus image is, which indirectly reflects that the ambient light of the currently collected fundus grayscale image is more suitable.

[0074] In other embodiments of the present invention, the implementer may also use basic mathematical operations such as addition or weighted summation or related mapping means to combine the two to obtain the distinction difficulty coefficient, which will not be repeated here.

[0075] Step S304 , negatively normalize the minimum value of the difficulty coefficients of distinguishing all fundus grayscale images to obtain the suitability of the ambient light for all fundus grayscale images.

[0076] Considering that each frame of fundus grayscale image can be approximately considered to be continuously acquired at the same time and scene, the corresponding differentiation difficulty coefficient of each frame of fundus grayscale image reflects the suitability of the ambient light at the time of current acquisition to a certain extent; but considering that when collecting fundus grayscale images, the eye movement of patients with eye diseases may cause the clarity of the fundus grayscale image, that is, the corresponding differentiation difficulty coefficient, to decrease. Therefore, in one embodiment of the present invention, a minimum value is selected from the corresponding differentiation difficulty coefficients of all fundus grayscale images, and the minimum differentiation difficulty coefficient is negatively normalized to obtain the suitability of the ambient light for all fundus grayscale images. The specific formula is expressed as: Q = 1-norm[min(D)]; where Q is the suitability of ambient light; min() is the minimum value function; D is the differentiation difficulty coefficient; min(D) is the minimum differentiation difficulty coefficient; norm[] is the standard normalization function.

[0077] In the ambient light suitability formula, the minimum distinction difficulty coefficient is linearly normalized, and the normalized value is mapped to a function expression of the form 1-x for negative correlation normalization. The logic is adjusted so that the smaller the minimum distinction difficulty coefficient is, the higher the ambient light suitability is when the fundus grayscale image is collected.

[0078] In other embodiments of the present invention, the implementer may also adopt other negative correlation normalization means, such as mapping to an exponential function e with the natural constant e as the base. -x Or do inverse operations, etc., which will not be repeated here.

[0079] At this point, the suitability of ambient light when collecting all fundus grayscale images has been obtained. The higher the suitability of ambient light, the higher the clarity of the fundus grayscale images collected under the preset exposure time, and the lower the need to adjust the exposure time; on the contrary, it means that the collection clarity of the fundus grayscale images of the current ambient light images is poor, and the exposure time needs to be adjusted to obtain clear images, thereby improving the subsequent transmission effect.

[0080] Step S3, between adjacent frames of fundus grayscale images, based on the similarity of grayscale differences between all pixels at the same position, the degree of micro-movement of the eyeball when all fundus grayscale images are acquired is evaluated.

[0081] When continuously collecting fundus grayscale images, the eye movement of patients with eye diseases may cause the clarity of the collected fundus grayscale images to decrease, and it is necessary to adjust to a lower exposure time to reduce motion blur; by evaluating the grayscale difference between pixels at the same position between adjacent frames of fundus grayscale images, the changes in pixels at the same position between adjacent frames of fundus grayscale images can be judged, and then the degree of micro-movement of the eyeball when collecting adjacent frames of fundus grayscale images can be evaluated, and then the preset exposure time can be adjusted based on the degree of micro-movement to reduce motion blur.

[0082] Preferably, in one embodiment of the present invention, the method for obtaining the micro-motion degree includes:

[0083] See also Figure 4 , which shows a flow chart of a method for obtaining the degree of micro-motion provided by an embodiment of the present invention.

[0084] Step S401, obtaining a difference image between adjacent frames of fundus grayscale images.

[0085] In one embodiment of the present invention, the difference image between the fundus grayscale images of adjacent frames is first acquired to prepare for the subsequent analysis of the change of the pixel points at the same position and then the evaluation of the degree of micro-movement of the eyeball.

[0086] It should be noted that the method for obtaining the difference image is an existing technology well known to those skilled in the art and will not be described in detail here.

[0087] Step S402, obtaining the overall micro-motion coefficient according to the discrete degree of the grayscale values ​​corresponding to all the pixels in the difference image.

[0088] Considering the grayscale value of the pixel at each position in the difference image, it is the grayscale change of the pixel at the corresponding position between the fundus grayscale images of adjacent frames. If the change of the corresponding grayscale change of all pixels in the difference image is more consistent, that is, the corresponding grayscale value discreteness is smaller, then it indirectly indicates that the micro-movement of the eyeball between the adjacent frames of fundus grayscale images is smaller, or the eyeball moves as a whole but the image clarity of the fundus grayscale image is low, and the corresponding overall micro-movement coefficient is smaller.

[0089] In one embodiment of the present invention, the standard deviation of the grayscale values ​​corresponding to all pixels in the difference image is normalized to obtain the overall micro-motion coefficient; in other embodiments of the present invention, the variance may be used instead of the standard deviation, which will not be described in detail.

[0090] Step S403, in the difference image, all target points are screened out from all pixels according to the grayscale distribution of each pixel within a preset range; and the local micro-motion coefficient is obtained according to the grayscale difference between each target point and other pixels within the corresponding preset range.

[0091] Considering that the slight movement of the eyeball may cause blurring of local details, analyzing the slight movement of the eyeball from the overall perspective of the difference image may easily overlook these local details; therefore, in one embodiment of the present invention, firstly, all target points are screened out from all pixels based on the grayscale distribution within a preset range of each pixel point; and the local grayscale of each target point is further analyzed to obtain the local slight movement coefficient.

[0092] It should be noted that the preset range is a preset neighborhood constructed with each pixel in the difference image as the center, so as to facilitate local analysis of the degree of micro-motion; the size of the preset neighborhood is 11×11, that is, a 11×11 local window is constructed with each pixel as the center; in other embodiments of the present invention, the implementer can also set the preset neighborhood size and shape by himself.

[0093] Preferably, in one embodiment of the present invention, the method for acquiring the target point includes:

[0094] All pixels in the difference image are sorted in descending order according to the grayscale value, and the first preset number of pixels in the sequence are selected as preliminary screening points; at least two target points are obtained from the preliminary screening points, wherein the distance between any two target points is greater than a preset length value.

[0095] Among them, the initial screening points are the pixels with larger grayscale values ​​in the difference image, that is, the pixels at this position have large grayscale changes between the fundus grayscale images of adjacent frames and are more likely to have local blur; all target points are further screened from the initial screening points to avoid concentrated analysis of the grayscale changes of pixels in a certain local area and improve analysis efficiency.

[0096] It should be noted that the preset number is the total number of pixels corresponding to the first 10% in the descending sorted sequence of the grayscale values ​​of all pixels in the difference image; since the size of the preset neighborhood is 11×11, the preset length value is at least 5; in other embodiments of the present invention, the implementer may also set the preset number by himself, set the preset length value to other values ​​greater than or equal to 5, or adjust the preset length value according to the preset neighborhood.

[0097] In another embodiment of the present invention, a method for acquiring target points includes: sorting all pixel points in the difference image in descending order according to grayscale values, and selecting the first preset number of pixel points in the sequence as preliminary screening points; obtaining the Euclidean distance between different preliminary screening points based on the coordinate position of the preliminary screening points in the difference image, clustering all preliminary screening points according to the Euclidean distance, obtaining all clusters, and taking the cluster center of each cluster as the target point, thereby obtaining all target points.

[0098] It should be noted that the method for obtaining the preliminary screening points is consistent with the method for obtaining the points in the above embodiment and will not be described in detail here; all the preliminary screening points are clustered using the DBSCAN clustering algorithm; in other embodiments of the present invention, other clustering algorithms such as the K-means clustering algorithm may also be used, which are all existing technologies well known to those skilled in the art and will not be described in detail here.

[0099] After all target points are screened out, the local micro-motion coefficient is obtained according to the grayscale difference between each target point and other pixel points in the corresponding preset range.

[0100] As an example, the local micromotion coefficient is calculated as:

[0101] Where L is the local micro-motion coefficient; g is the serial number of the target point in the difference image; G is the total number of target points in the difference image; f g is the serial number of the pixel point in the preset neighborhood of the g-th target point in the difference image; F g is the total number of pixels in the preset neighborhood of the g-th target point in the difference image; is the fth target point in the preset neighborhood of the gth target point in the difference image. g Gray value of pixel; Δh g is the gray value of the g-th target point in the difference image.

[0102] In the calculation formula of the local micro-motion coefficient, since the grayscale values ​​of other pixels in the preset neighborhood cannot be greater than the target point, Always less than or equal to 1; The larger it is, the larger the grayscale value in the local range of the preset neighborhood corresponding to the target point is, that is, the greater the possibility of motion blur at this local position between the fundus grayscale images of adjacent frames. Conversely, the more consistent and smaller the grayscale change is, the smaller the possibility of motion blur is. Further, the local micro-motion coefficient of the difference image is evaluated comprehensively for all target points. The larger the local micro-motion coefficient is, the greater the possibility of motion blur is.

[0103] In other embodiments of the present invention, the implementer may also use the mode instead of the mean to obtain the local micro-motion coefficient, which will not be elaborated here.

[0104] Step S404 , obtaining the micromotion degree according to the overall micromotion coefficient and the local micromotion coefficient, wherein both the overall micromotion coefficient and the local micromotion coefficient are positively correlated with the micromotion degree.

[0105] In one embodiment of the present invention, the overall micro-motion coefficient and the local micro-motion coefficient are multiplied and combined to obtain the micro-motion degree of the eyeball when all fundus grayscale images are acquired. The smaller the overall micro-motion coefficient or the local micro-motion coefficient, the smaller the micro-motion degree, and the lower the possibility of motion blur. In other embodiments of the present invention, the implementer may also combine the two by basic mathematical operations such as addition or weighted summation or related mapping means, which will not be repeated here.

[0106] Step S4, adjusting the preset exposure time according to the suitability of the ambient light and the degree of micro-movement to obtain the optimal exposure time; using the optimal exposure time to collect the fundus grayscale image to be transmitted, and remotely transmitting it.

[0107] Considering that the fundus grayscale image is acquired in a dark room, the possibility of overexposure is low. If the fundus grayscale image is not clear enough and the acquisition quality is poor, it may be due to the dark ambient light. Therefore, the preset exposure time should be appropriately increased to capture more light, thereby improving the image clarity and acquisition quality; and considering that the greater the degree of eye movement, the preset exposure time needs to be appropriately reduced to reduce motion blur. Therefore, the embodiment of the present invention first adjusts the preset exposure time according to the suitability of the ambient light and the degree of micro-movement to obtain the optimal exposure time; and then uses the optimal exposure time to acquire high-quality fundus grayscale images to be transmitted, and transmits them remotely.

[0108] Preferably, in one embodiment of the present invention, the method for obtaining the optimal exposure time includes:

[0109] Obtaining the total length of the intervals within the preset standard exposure time range, and using the total length of the intervals with a preset ratio as an adjustment reference value;

[0110] The negative correlation of the ambient light suitability is normalized as the weight of the adjustment reference value to obtain the upward adjustment range of the preset exposure time;

[0111] Using the micro-motion degree as the weight of the adjustment reference value, obtaining the downward adjustment amplitude of the preset exposure time;

[0112] The optimal exposure time is obtained according to the upward adjustment range and the downward adjustment range of the preset exposure time. The upward adjustment range is positively correlated with the optimal exposure time, and the downward adjustment range is negatively correlated with the optimal exposure time.

[0113] As an example, the adjustment reference value is half of the interval length of the preset standard exposure time range. In other embodiments of the present invention, the implementer may also adjust it to one third or one quarter of the interval length, etc., but it is necessary to ensure that the adjustment reference value is always less than or equal to half of the interval length to avoid the obtained optimal exposure time exceeding the standard range.

[0114] The calculation formula for the optimal exposure time is: T=t+(1-Q)×T0-W×T0; where T is the optimal exposure time; t is the preset exposure time; Q is the suitability of ambient light; T0 is the adjustment reference value; W is the degree of micro-movement; (1-Q)×T0 is the upward adjustment range; W×T0 is the downward adjustment range.

[0115] In the calculation formula of the optimal exposure time, the ambient light suitability is negatively normalized in the form of 1-Q as the weight of the adjustment reference value to obtain the upward adjustment range. The higher the ambient light suitability, the smaller the upward adjustment range. In other embodiments of the present invention, the implementer may also adopt other negative correlation normalization means, such as taking the ambient light suitability as an exponential function e with the natural constant e as the base. -x The exponent x in the formula or the inverse operation can be performed; the normalized value of the minimum differentiation difficulty coefficient can also be directly used as the weight of the adjustment reference value to obtain the upward adjustment range of the preset exposure time; the micro-motion degree can be used as the weight of the adjustment reference value to obtain the downward adjustment range, and the greater the micro-motion degree, the greater the downward adjustment range.

[0116] At this point, the optimal exposure time is obtained, and the optimal exposure time is used to re-acquire a high-quality fundus grayscale image, obtain the fundus grayscale image to be transmitted, and then transmit it remotely. In a preferred embodiment of the present invention, a secure transmission channel is established through VPN technology, and the fundus grayscale image to be transmitted is encrypted and remotely transmitted using the SSL encryption protocol, thereby ensuring the quality of the fundus grayscale image to be transmitted while ensuring the security and privacy protection of the transmission.

[0117] In summary, the embodiment of the present invention first acquires continuous frames of fundus grayscale images with a preset exposure time; further, based on the grayscale distribution difference between the target area and the background area in each frame of the fundus grayscale image, evaluates the suitability of the ambient light when acquiring all fundus grayscale images; further, between adjacent frames of fundus grayscale images, based on the similarity of the grayscale difference between all pixels at the same position, evaluates the degree of micro-movement of the eyeball when acquiring all fundus grayscale images; and then adjusts the preset exposure time, acquires the fundus grayscale image to be transmitted, and transmits it remotely. The present invention compares and analyzes the detail clarity and grayscale distribution difference of adjacent frames of fundus grayscale images acquired continuously with a preset exposure time, evaluates the influence of ambient light during acquisition and eyeball movement on fundus grayscale image acquisition, and thereby adjusts the preset exposure time to acquire high-quality fundus grayscale images for remote transmission, thereby improving the transmission effect.

[0118] It should be noted that the sequence of the above embodiments of the present invention is only for description and does not represent the advantages and disadvantages of the embodiments. The processes depicted in the accompanying drawings do not necessarily require the specific order or continuous order shown to achieve the desired results. In some embodiments, multitasking and parallel processing are also possible or may be advantageous.

[0119] The various embodiments in this specification are described in a progressive manner, and the same or similar parts between the various embodiments can be referenced to each other, and each embodiment focuses on the differences from other embodiments.

Claims

1. An image transmission method for remote ophthalmic detection, characterized in that: The method comprises: Acquire at least two consecutive frames of fundus grayscale images with a preset exposure time; Acquire a target area and a background area in each frame of the fundus grayscale image, wherein the target area is a fundus structure area; and evaluate the suitability of ambient light when acquiring all the fundus grayscale images according to the grayscale distribution difference between the target area and the background area in each frame of the fundus grayscale image; Between the fundus grayscale images of adjacent frames, the degree of micro-movement of the eyeball when all the fundus grayscale images are acquired is evaluated according to the similarity of the grayscale differences between all pixels at the same position; The preset exposure time is adjusted according to the suitability of the ambient light and the degree of micro-movement to obtain an optimal exposure time; the fundus grayscale image to be transmitted is collected using the optimal exposure time and transmitted remotely.

2. The image transmission method for ophthalmic remote detection according to claim 1, characterized in that: The method for obtaining the suitability of ambient light includes: In each frame of the fundus grayscale image, obtaining a first parameter according to a difference between corresponding grayscale distributions of the target area and the background area; Acquire a second parameter according to the difference between the number of corresponding pixels at each same gray level between the target area and the background area; According to the first parameter and the second parameter, obtaining a distinction difficulty coefficient between the target area and the background area in each frame of the fundus grayscale image, wherein the first parameter and the second parameter are both positively correlated with the distinction difficulty coefficient; All the fundus grayscale images are negatively normalized with respect to the minimum value of the differentiation difficulty coefficient, so as to obtain the suitability of the ambient light when all the fundus grayscale images are obtained.

3. The image transmission method for ophthalmic remote detection according to claim 2, characterized in that: The method for obtaining the first parameter includes: In each frame of the fundus grayscale image, a grayscale set of the target area and a grayscale set of the background area are respectively obtained, and the Jaccard similarity coefficient between the two grayscale sets is used as the first parameter.

4. The image transmission method for ophthalmic remote detection according to claim 1, characterized in that: The method for obtaining the micro-motion degree includes: Acquire a difference image between fundus grayscale images of adjacent frames; Obtaining the overall micro-motion coefficient according to the discrete degree of the grayscale values ​​corresponding to all pixels in the difference image; In the difference image, all target points are screened out from all pixels according to the grayscale distribution of each pixel within a preset range; local micro-motion coefficients are obtained according to the grayscale difference between each target point and other pixels within the corresponding preset range; The micromotion degree is obtained according to the overall micromotion coefficient and the local micromotion coefficient, and both the overall micromotion coefficient and the local micromotion coefficient are positively correlated with the micromotion degree.

5. The image transmission method for ophthalmic remote detection according to claim 4, characterized in that: The method for acquiring the target point includes: Sort all the pixels in the difference image in descending order according to the grayscale value, and select the first preset number of pixels in the sequence as preliminary screening points; At least two target points are obtained from the preliminary screening points, wherein the distance between any two target points is greater than a preset length value.

6. The image transmission method for ophthalmic remote detection according to claim 1, characterized in that: The method for obtaining the preset exposure time includes: The median value of the preset standard exposure time range is used as the preset exposure time.

7. The image transmission method for ophthalmic remote detection according to claim 6, characterized in that: The method for obtaining the optimal exposure time includes: Obtaining a total length of an interval within a preset standard exposure time range, and using a preset proportion of the total length of the interval as an adjustment reference value; The negative correlation of the ambient light suitability is normalized as the weight of the adjustment reference value to obtain an upward adjustment range of the preset exposure time; Taking the micro-motion degree as the weight of the adjustment reference value, obtaining a downward adjustment amplitude of the preset exposure time; The optimal exposure time is obtained according to the upward adjustment range and the downward adjustment range of the preset exposure time, wherein the upward adjustment range is positively correlated with the optimal exposure time, and the downward adjustment range is negatively correlated with the optimal exposure time.

8. The image transmission method for ophthalmic remote detection according to claim 1, characterized in that: The method for obtaining the target area and the background area in each frame of the fundus grayscale image comprises: Each frame of the fundus grayscale image is segmented based on a convolutional neural network to obtain the target area and the background area.

9. The image transmission method for ophthalmic remote detection according to claim 1, characterized in that: The remote transmission method of the fundus grayscale image to be transmitted comprises: A secure transmission channel is established through VPN technology, and the SSL encryption protocol is used to encrypt the fundus grayscale image to be transmitted and transmit it remotely.

10. The image transmission method for ophthalmic remote detection according to claim 7, characterized in that: The preset ratio is one-half.

Images