An eye redness state analysis method and apparatus, a computing device, and a storage medium

Abstract

The application discloses an eye red state analysis method and device, a computing device and a storage medium. The method comprises the following steps: acquiring an eye image to be detected; performing regional segmentation on the eye image based on a segmentation network to extract a conjunctival region and an eyelash region; performing edge detection and edge-out mask segmentation on the conjunctival region to obtain a limbal region; performing red threshold segmentation on the eye image to obtain an eye red region after removing the eyelash region; calculating an eye red proportion in the conjunctival region and an eye red proportion in the limbal region to obtain an eye red state analysis result. The scheme can improve the eye red proportion recognition accuracy and efficiency, and provide reliable data support for various eye diseases.

Description

Technical Field

[0001] This invention relates to the field of eye image processing technology, specifically to a method, apparatus, computing device, and storage medium for analyzing redness in the eyes. Background Technology

[0002] With the increasing frequency of use of electronic devices such as mobile phones and computers, more and more people are suffering from dry eye syndrome. Some patients with dry eye syndrome may experience symptoms such as redness, foreign body sensation, and bloodshot eyes. These symptoms can be relieved with medication, such as artificial tears and anti-inflammatory drugs. Physical therapy, such as meibomian gland steaming, massage, and photothermal therapy, can also help alleviate the inflammatory response.

[0003] In clinical practice, timely and accurate detection of red eyes, along with quantitative assessment of the area and proportion of conjunctival and ciliary body hyperemia, plays a crucial role in the diagnosis and treatment of dry eye syndrome.

[0004] Currently, most analyses of redness in the eyes employ traditional image processing methods. These methods use thresholding to extract the conjunctiva, limbus, and redness regions from the original image, and then calculate the redness ratio based on the area of ​​these three regions. However, thresholding is easily affected by lighting conditions and changes in the eyeball, which can impact the detection rate.

[0005] Chinese patent document CN115035103A discloses a method for processing red-eye image data and a device for analyzing red-eye levels. This method involves inputting red-eye image data into a pre-trained first neural network model, outputting a red-eye analysis region, extracting color channels from the red-eye image data, calculating the red-eye lesion region based on the color channels and the red-eye analysis region, calculating the ratio between the red-eye analysis region and the red-eye lesion region, and determining the red-eye level of the red-eye image data. Compared with traditional digital image processing techniques, this method is less affected by external factors such as lighting when analyzing the proportion of red areas, and has high robustness and accuracy. However, this scheme only segments the entire conjunctival region and requires two neural network models to obtain the red-eye lesion region, resulting in a large computational load, which is not conducive to real-time analysis of red-eye data. Summary of the Invention

[0006] To improve the accuracy and efficiency of redness ratio recognition, this scheme proposes a redness state analysis method. Based on a segmentation network, the conjunctival and eyelash regions are segmented. For the conjunctival region, edge detection algorithms and morphological manipulation methods are used to extract the limbal region. This method combines deep learning algorithms with traditional segmentation algorithms, improving the accuracy and efficiency of segmentation of each region of interest, reducing interference from irrelevant regions, and decreasing computational load. Finally, the redness ratios within the conjunctival and limbal regions are calculated separately, providing important reference for the diagnosis of various eye diseases such as dry eye, conjunctivitis, and scleritis, and offering more refined redness state analysis results.

[0007] According to a first aspect of the present invention, a method for analyzing redness of the eyes is provided, comprising:

[0008] Acquire an eye image to be detected; perform region segmentation on the eye image based on a segmentation network to extract the conjunctival region and the eyelash region; perform edge detection and edge outward masking segmentation on the conjunctival region to obtain the limbal region; perform red thresholding segmentation on the eye image to obtain the red eye region after removing the eyelash region; calculate the red eye ratio in the conjunctival region and the red eye ratio in the limbal region to obtain the red eye state analysis result.

[0009] Optionally, in the red eye state analysis method provided by the present invention, the segmentation network can adopt any one of the network architectures of U-Net, FCN, XceptionA, Transformer, and DFAnet. The backbone network of the segmentation network is an encoder-decoder structure, which is used to output the background prediction probability, conjunctival prediction probability, and eyelash prediction probability, respectively.

[0010] Optionally, in the redness analysis method provided by the present invention, the inner edge of the conjunctival region is detected based on an edge detection algorithm; the inner edge is expanded based on a dilation operation, and the expanded edge image is converted into a binary mask to obtain a binary outward expansion mask; the limbal region is obtained based on the intersection of the binary outward expansion mask and the conjunctival region.

[0011] Optionally, in the red eye state analysis method provided by the present invention, the eye image is converted from RGB color space to HSV color space to obtain an HSV image; based on the red channel of the HSV image and a preset effective red range value, the red eye region is thresholded to obtain a threshold segmentation result; after removing the eyelash region from the threshold segmentation result, the intersection with the conjunctival region is taken to obtain the red eye region.

[0012] Optionally, in the redness state analysis method provided by the present invention, if the pixel color value of the HSV image tone channel is within the preset red effective range, the gray value of the pixel is set to 0; if the pixel color value of the HSV image tone channel is not within the preset red effective range, the gray value of the pixel is set to 255.

[0013] Optionally, in the red eye state analysis method provided by the present invention, regional connectivity analysis can also be performed on the red eye region to filter out the final red eye region based on the area of ​​the red region and a preset area threshold.

[0014] By removing the eyelash region and performing region connectivity analysis on the red eye region obtained from threshold segmentation, noise interference can be reduced and the accuracy of red eye region detection can be improved.

[0015] Optionally, in the redness state analysis method provided by the present invention, the number of first pixels in the conjunctival region and the number of second pixels of redness pixels in the conjunctival region are counted; the redness ratio in the conjunctival region is calculated based on the ratio of the number of second pixels to the number of first pixels.

[0016] The number of third pixels and the number of fourth pixels in the red-eyed region within the limbal region are counted; the red-eyed ratio in the limbal region is calculated based on the ratio of the number of fourth pixels to the number of third pixels.

[0017] According to a second aspect of the present invention, an eye redness state analysis device is provided, comprising an acquisition module, a conjunctival region extraction module, a limbal region extraction module, an eye redness region extraction module, and a statistical analysis module.

[0018] The acquisition module is used to acquire the eye image to be detected.

[0019] The conjunctival region extraction module is used to perform region segmentation on the eye image based on a segmentation network, and extract the conjunctival region and the eyelash region.

[0020] The limbal region extraction module is used to perform edge detection and edge outward masking segmentation on the conjunctival region to obtain the limbal region;

[0021] The red eye region extraction module is used to perform red threshold segmentation on the eye image to obtain the red eye region after removing the eyelash area; the statistical analysis module is used to calculate the proportion of red eye in the conjunctival region and the proportion of red eye in the limbal region to obtain the red eye status analysis results.

[0022] According to a third aspect of the present invention, a computing device is provided, comprising: at least one processor; and a memory storing program instructions, wherein the program instructions are configured to be executed by the at least one processor, the program instructions including instructions for performing the above-described red-eye state analysis method.

[0023] According to a fourth aspect of the present invention, a readable storage medium storing program instructions is provided, which, when read and executed by a computing device, causes the computing device to perform the above-described red-eye state analysis method.

[0024] According to the solution provided by this invention, the conjunctival region and eyelash region are segmented based on a segmentation network. For the conjunctival region, edge detection algorithms and morphological manipulation methods are used to extract the limbal region. This combines deep learning algorithms with traditional segmentation algorithms, avoiding the influence of adverse factors such as illumination, noise, and occlusion on region segmentation, reducing interference from irrelevant regions, and improving the accuracy and efficiency of segmentation for each region of interest. It eliminates the need for multiple deep learning network segmentations, quickly and accurately determining the proportion of redness in the conjunctival region and the limbal region. This provides important reference data for the diagnosis of various eye diseases such as dry eye, conjunctivitis, and scleritis, and offers more refined analysis results of redness status.

[0025] The above description is merely an overview of the technical solution of the present invention. In order to better understand the technical means of the present invention and to implement it in accordance with the contents of the specification, and in order to make the above and other objects, features and advantages of the present invention more apparent and understandable, specific embodiments of the present invention are described below. Attached Figure Description

[0026] Various other advantages and benefits will become apparent to those skilled in the art upon reading the following detailed description of preferred embodiments. The accompanying drawings are for illustrative purposes only and are not intended to limit the invention. Furthermore, the same reference numerals denote the same parts throughout the drawings. In the drawings:

[0027] Figure 1 A structural diagram of a computing device 100 according to an embodiment of the present invention is shown;

[0028] Figure 2 A flowchart illustrating a red-eye state analysis method 200 according to an embodiment of the present invention is shown;

[0029] Figure 3 A schematic diagram of a segmentation network model structure according to an embodiment of the present invention is shown;

[0030] Figure 4 A schematic diagram of the limbal region according to an embodiment of the present invention is shown;

[0031] Figure 5 A schematic diagram of the red eye region extraction result according to an embodiment of the present invention is shown;

[0032] Figure 6A schematic diagram of the structure of an eye redness state analysis device 600 according to an embodiment of the present invention is shown. Detailed Implementation

[0033] Redness of the eyes is an eye disease characterized primarily by conjunctival vasodilation and congestion, manifesting as conjunctival hyperemia, ciliary hyperemia, and mixed hyperemia. Redness of the eyes can be associated with various diseases, such as dry eye syndrome, allergic conjunctivitis, bacterial conjunctivitis, scleritis, and iritis, and the distribution of redness varies considerably among these conditions. Therefore, accurate detection of the proportion of redness in the eyes is beneficial for the timely detection and differentiation of various diseases such as dry eye, conjunctivitis, and iritis.

[0034] This solution proposes a method for analyzing redness of the eye, which combines deep learning algorithms with traditional threshold segmentation algorithms. This improves the accuracy and efficiency of segmenting regions of interest, reduces interference from irrelevant regions, reduces computational load, and provides more refined redness analysis results. This provides important reference for the diagnosis of various eye diseases such as dry eye, conjunctivitis, and scleritis.

[0035] Exemplary embodiments of the present disclosure will now be described in more detail with reference to the accompanying drawings. While exemplary embodiments of the present disclosure are shown in the drawings, it should be understood that the present disclosure may be implemented in various forms and should not be limited to the embodiments set forth herein. Rather, these embodiments are provided so that this disclosure will be thorough and complete, and will fully convey the scope of the disclosure to those skilled in the art.

[0036] Figure 1 A structural diagram of a computing device 100 according to an embodiment of the present invention is shown. Figure 1 As shown, in the basic configuration 102, the computing device 100 typically includes system memory 106 and one or more processors 104. Memory bus 108 can be used for communication between processor 104 and system memory 106.

[0037] Depending on the desired configuration, processor 104 can be any type of processor, including but not limited to: microprocessors (μP), microcontrollers (μC), digital information processors (DSPs), or any combination thereof. Processor 104 may include one or more levels of cache such as L1 cache 110 and L2 cache 112, processor core 114, and registers 116. Example processor core 114 may include an arithmetic logic unit (ALU), a floating-point unit (FPU), a digital signal processing core (DSP core), or any combination thereof. Example memory controller 118 may be used with processor 104, or in some implementations, memory controller 118 may be an internal part of processor 104.

[0038] Depending on the desired configuration, system memory 106 can be any type of memory, including but not limited to: volatile memory (such as RAM), non-volatile memory (such as ROM, flash memory, etc.), or any combination thereof. Physical memory in a computing device typically refers to volatile RAM, and data on a disk needs to be loaded into physical memory before it can be read by processor 104. System memory 106 may include operating system 120, one or more applications 122, and program data 124. In some embodiments, application 122 may be configured to execute instructions on the operating system using program data 124 by one or more processors 104. Operating system 120 may be, for example, Linux, Windows, etc., and includes program instructions for handling basic system services and performing hardware-dependent tasks. Application 122 includes program instructions for implementing various user-desired functions, and application 122 may be, for example, a browser, instant messaging software, software development tools (such as integrated development environments IDEs, compilers, etc.), but is not limited to these. When application 122 is installed in computing device 100, driver modules may be added to operating system 120.

[0039] When computing device 100 starts up, processor 104 reads and executes program instructions from memory 106 of operating system 120. Application 122 runs on operating system 120, utilizing interfaces provided by operating system 120 and underlying hardware to implement various user-expected functions. When user starts application 122, application 122 is loaded into memory 106, and processor 104 reads and executes program instructions from memory 106 of application 122.

[0040] The computing device 100 also includes a storage device 132, which includes a removable storage device 136 and a non-removable storage device 138, both of which are connected to a storage interface bus 134.

[0041] The computing device 100 may also include an interface bus 140 that facilitates communication from various interface devices (e.g., output devices 142, peripheral interfaces 144, and communication devices 146) to the basic configuration 102 via a bus / interface controller 130. Example output devices 142 include a graphics processing unit 148 and an audio processing unit 150. They may be configured to facilitate communication with various external devices such as displays or speakers via one or more A / V ports 152. Example peripheral interfaces 144 may include a serial interface controller 154 and a parallel interface controller 156, which may be configured to facilitate communication with external devices such as input devices (e.g., keyboards, mice, pens, voice input devices, touch input devices) or other peripherals (e.g., printers, scanners, etc.) via one or more I / O ports 158. Example communication devices 146 may include a network controller 160, which may be arranged to facilitate communication with one or more other computing devices 162 via a network communication link through one or more communication ports 164.

[0042] A network communication link can be an example of a communication medium. A communication medium can typically be embodied in computer-readable instructions, data structures, program modules in a modulated data signal such as a carrier wave or other transmission mechanism, and can include any information delivery medium. A “modulated data signal” can be a signal whose data set, or whose modifications, can be encoded with information in the signal. As a non-limiting example, a communication medium can include wired media such as wired networks or leased lines, and various wireless media such as sound, radio frequency (RF), microwave, infrared (IR), or other wireless media. The term computer-readable medium as used herein can include both storage media and communication media. In the computing device 100 according to the invention, application 122 includes instructions for performing the red-eye state analysis method 200 of the invention.

[0043] Figure 2 A flowchart illustrating an eye redness state analysis method 200 according to an embodiment of the present invention is shown. Figure 2 As shown, the method 200 begins with step S210, which acquires an image of the eye to be detected.

[0044] High-definition cameras can be used to capture unobstructed images of the eyes from the front. Alternatively, medical devices such as ocular surface analyzers can be used to acquire images. Or, images can be obtained from existing ocular image databases.

[0045] Then, step S220 is executed, which performs region segmentation on the eye image based on the segmentation network to extract the conjunctival region and the eyelash region.

[0046] The segmentation network can adopt any network architecture such as U-Net, FCN, XceptionA, Transformer, DFAnet, etc., and this scheme does not impose any restrictions on it. The backbone network of the segmentation network is an encoder-decoder structure, which is used to output the background prediction probability, conjunctival prediction probability, and eyelash prediction probability, respectively.

[0047] Before using the constructed segmentation network for conjunctival region segmentation, it needs to be trained. The weights of the segmentation network trained on the Cityscapes dataset can be used as initial weights.

[0048] The Cityscapes dataset contains various stereo video sequences recorded from street views of 50 different cities, providing 5,000 finely annotated images. This set of 5,000 finely annotated images is usually used for training and evaluation.

[0049] Then, a certain number of labeled eye images are obtained, and the data is augmented through preprocessing operations such as scaling, grayscale stretching, random blurring, flipping, rotating, and translating to obtain a dataset for model training.

[0050] Subsequently, the dataset can be divided into a training set and a validation set. The training set is input into the segmentation network for iterative training, and the segmentation accuracy of the conjunctival region and the eyelash region is supervised based on the cross-entropy loss value to determine the model weight parameters and obtain the trained segmentation network.

[0051] Figure 3 A schematic diagram of a segmentation network model structure according to an embodiment of the present invention is shown. Figure 3 As shown, taking the DFANet segmentation network as an example, enc refers to the encoder convolutional layer block, fc attention refers to the attention module that acquires semantic and category information, C refers to the concatenation layer by channel, and xN refers to upsampling, which is an upsampling operation of multiples of N.

[0052] The encoder contains multiple branches, each containing multiple encoder modules and an full-c attention module. The output of each module is fused with the input of the next branch. The decoder receives the output of the first encoder module and the output of the full-c attention module of each encoder branch, and after upsampling, outputs the background prediction probability, conjunctival prediction probability, and eyelash prediction probability, respectively.

[0053] like Figure 3As shown, the Encoder structure is divided into three branches. Each branch contains three encoding modules and an fully connected (fc) attention module that implements channel attention. The output of each module is fused with the input of the next branch. After the outputs and inputs of the modules in the three branches are fused together, the enc2 output and fcattention output of each branch are then connected to the Decoder structure.

[0054] In the Decoder structure, the outputs of the enc2 module and the full-c attention module from each encoder branch are received. The outputs of the three enc2 modules are summed, convolved, and then summed again with the outputs of the three full-c attention modules. After upsampling, the background prediction probability, conjunctival prediction probability, and eyelash prediction probability are output respectively, resulting in a segmentation probability map. Based on the segmentation probability map, binary segmented images of the conjunctival and eyelash regions can be obtained.

[0055] The eye image to be detected is input into the trained segmentation network for conjunctival region prediction, which outputs a binary segmentation image containing the conjunctival region and the background region.

[0056] Next, step S230 is executed to perform edge detection and edge outward masking segmentation on the conjunctival region to obtain the limbal region.

[0057] The limbus is a grayish-white transition zone between the cornea and conjunctiva. This area is richly vascularized and prone to bleeding. Therefore, it is necessary to assess the degree of redness in this area.

[0058] Specifically, the inner edge of the conjunctival region can be extracted first using edge detection algorithms, such as Canny edge detection or the Sobel operator. Then, the inner edge is expanded using a dilation operation, for example, by expanding the inner edge outward by 30 pixels to both sides of the image. The expanded edge image is then converted into a binary mask to obtain a binary outward expansion mask. In this mask, the edge region is represented by 1, and other regions are represented by 0.

[0059] Finally, the limbal region can be obtained by intersecting the original eye image with the binarized extended mask (bitwise AND operation).

[0060] Figure 4 A schematic diagram of the limbal region according to an embodiment of the present invention is shown. Figure 4 As shown, the highlighted area is the extracted limbal region, i.e., the limbal region.

[0061] Then, step S240 is executed to perform red threshold segmentation on the eye image to obtain the red area of ​​the eye after removing the eyelash area.

[0062] The main characteristic of red and congested eyes is distinguished by color information: red represents the red area of ​​the eye, and white represents the conjunctival area. You can first convert the eye image from the RGB color space to the HSV color space to obtain an HSV image.

[0063] Then, based on the red channel of the HSV image and the preset effective red range value, threshold segmentation of the red-eyed region is performed to obtain the threshold segmentation result. If the pixel color value of the HSV image tone channel is within the preset effective red range, the gray value of the pixel is set to 0; if the pixel color value of the HSV image tone channel is not within the preset effective red range, the gray value of the pixel is set to 255.

[0064] Finally, after removing the eyelash region from the threshold segmentation result, the intersection with the conjunctival region is taken to obtain the red eye region.

[0065] Due to the influence of light, some eyelash areas may be mistaken for the red eye area. Therefore, it is necessary to extract the eyelash area obtained in step S220 and then intersect it with the conjunctival area to obtain the red eye area.

[0066] To improve the accuracy of red eye region extraction, regional connectivity analysis can be performed on the red eye region to identify connected red regions. Based on the size of the red region, an area threshold can be set to filter out the red eye regions that meet the requirements, thus obtaining the final red eye region.

[0067] Figure 5 A schematic diagram illustrating the result of red eye region extraction according to an embodiment of the present invention is shown. Figure 5 As shown, the black lines within the conjunctival area represent the red area of ​​the eye.

[0068] Finally, step S250 is executed to calculate the proportion of redness in the conjunctival region and the proportion of redness in the limbal region, thus obtaining the redness status analysis results.

[0069] After extracting the image ranges of the conjunctival region, limbal region, and red eye region respectively, the proportion of red eye in the conjunctival region and the proportion of red eye in the limbal region can be calculated by counting the number of pixels in each region.

[0070] Specifically, the number of first pixels S0 within the conjunctival region and the number of second pixels S1 within the conjunctival region that are red in the eyes can be counted; the proportion of red in the eyes within the conjunctival region can be calculated based on the ratio of the number of second pixels S1 to the number of first pixels S0.

[0071] The number of third pixels (S2) and the number of fourth pixels (S3) of red-eyed pixels within the limbal region are counted. The red-eyed ratio within the limbal region is calculated based on the ratio of the number of fourth pixels (S3) to the number of third pixels (S2).

[0072] Figure 6 A schematic diagram of an eye redness state analysis device 600 according to an embodiment of the present invention is shown. Figure 6 As shown, the redness state analysis device 600 may include an acquisition module 610, a conjunctival region extraction module 620, a limbal region extraction module 630, a redness region extraction module 640, and a statistical analysis module 650.

[0073] The acquisition module 610 acquires the eye image to be detected. The conjunctival region extraction module 620 can perform region segmentation on the eye image based on a segmentation network to extract the conjunctival region and the eyelash region.

[0074] The limbal region extraction module 630 can perform edge detection and edge outward masking segmentation on the conjunctival region to obtain the limbal region.

[0075] The red area extraction module 640 can perform red threshold segmentation on the eye image to obtain the red area of ​​the eye after removing the eyelashes.

[0076] The statistical analysis module 650 can calculate the proportion of redness in the conjunctival region and the proportion of redness in the limbal region, and obtain the results of the redness status analysis.

[0077] The redness analysis method and apparatus provided by this invention segment the conjunctival region and eyelash region using a segmentation network. For the conjunctival region, edge detection algorithms and morphological manipulation methods are used to extract the limbal region. This method combines deep learning algorithms with traditional segmentation algorithms, improving the accuracy and efficiency of segmentation of each region of interest, reducing interference from irrelevant regions, and decreasing computational load. Finally, the redness ratio in the conjunctival region and the limbal region are calculated separately, providing important reference for the diagnosis of various eye diseases such as dry eye, conjunctivitis, and scleritis, and providing more refined redness analysis results.

[0078] Numerous specific details are set forth in the specification provided herein. However, it will be understood that embodiments of the invention may be practiced without these specific details. In some instances, well-known methods, structures, and techniques have not been shown in detail so as not to obscure the understanding of this specification.

[0079] Similarly, it should be understood that, in order to streamline this disclosure and aid in understanding one or more of the various aspects of the invention, in the above description of exemplary embodiments of the invention, various features of the invention are sometimes grouped together in a single embodiment, figure, or description thereof. However, this method of disclosure should not be interpreted as reflecting an intention that the claimed invention requires more features than are expressly recited in each claim. Rather, as reflected in the following claims, inventive aspects lie in fewer than all features of a single foregoing disclosed embodiment. Therefore, the claims following the detailed description are hereby expressly incorporated into this detailed description, wherein each claim itself is a separate embodiment of the invention.

[0080] Those skilled in the art will understand that modules, units, or components of the devices disclosed in the examples herein can be arranged in the devices described in this embodiment, or alternatively, can be located in one or more devices different from the devices in this example. The modules in the foregoing examples can be combined into a single module or, in addition, can be divided into multiple sub-modules.

[0081] Those skilled in the art will understand that modules in the device of the embodiments can be adaptively changed and placed in one or more devices different from that embodiment. Modules, units, or components in the embodiments can be combined into a single module, unit, or component, and further, they can be divided into multiple sub-modules, sub-units, or sub-components. Except where at least some of such features and / or processes or units are mutually exclusive, any combination can be used to combine all features disclosed in this specification (including the accompanying claims, abstract, and drawings) and all processes or units of any method or device so disclosed. Unless expressly stated otherwise, each feature disclosed in this specification (including the accompanying claims, abstract, and drawings) may be replaced by an alternative feature that serves the same, equivalent, or similar purpose.

[0082] Furthermore, those skilled in the art will understand that although some embodiments described herein include certain features but not others included in other embodiments, combinations of features from different embodiments are intended to be within the scope of the invention and form different embodiments. For example, in the following claims, any of the claimed embodiments can be used in any combination.

[0083] Furthermore, some of the embodiments described herein are methods or combinations of method elements that can be implemented by a processor of a computer system or by other means of performing the functions. Therefore, a processor having the necessary instructions for implementing the methods or method elements forms means for implementing the methods or method elements. Furthermore, the elements described herein in the apparatus embodiments are examples of means for implementing the functions performed by elements for the purposes of carrying out the invention.

[0084] As used herein, unless otherwise specified, the use of ordinal numbers such as “first,” “second,” “third,” etc., to describe ordinary objects merely indicates different instances of similar objects and is not intended to imply that the objects being described must have a given order in time, space, ordering, or any other manner.

[0085] Although the invention has been described with respect to a limited number of embodiments, those skilled in the art will understand from the foregoing description that other embodiments are conceivable within the scope of the invention described herein. Furthermore, it should be noted that the language used in this specification has been chosen primarily for readability and edibility purposes, and not for the purpose of interpreting or limiting the subject matter of the invention. Therefore, many modifications and variations will be apparent to those skilled in the art without departing from the scope and spirit of the appended claims. The disclosure of the invention is illustrative rather than restrictive, and the scope of the invention is defined by the appended claims.

Claims

1. An ocular redness state analysis method characterized by, The method comprises the following steps: Obtaining an eye image to be detected; Segmenting the eye image based on a segmentation network to extract a conjunctival region and an eyelash region; Detecting the inner edge of the conjunctival region based on an edge detection algorithm; Extending the inner edge based on an expansion operation, converting the extended edge image into a binary mask to obtain a binary outer expansion mask; Obtaining a limbal region based on the intersection of the binary outer expansion mask and the conjunctival region; Converting the eye image from an RGB color space to an HSV color space to obtain an HSV image; Performing threshold segmentation on the eye red region based on the red channel of the HSV image and a preset red effective range value to obtain a threshold segmentation result; Obtaining an eye red region by removing the eyelash region from the threshold segmentation result and taking the intersection with the conjunctival region; Calculating the eye red proportion in the conjunctival region and the eye red proportion in the limbal region to obtain an eye red state analysis result.

2. The eye redness state analysis method according to claim 1, characterized by, The segmentation network adopts any one of U-Net, FCN, XceptionA, Transformer and DFAnet network architectures, and the backbone network of the segmentation network is an encoder-decoder structure, for outputting background prediction probability, conjunctival prediction probability and eyelash prediction probability, respectively.

3. The eye redness state analysis method according to claim 1, characterized by, The step of performing threshold segmentation on the eye red region based on the hue channel of the HSV image and a preset red effective range value to obtain a threshold segmentation result comprises: If the pixel color value of the hue channel of the HSV image is within the preset red effective range, the gray value of the pixel is set to 0; If the pixel color value of the hue channel of the HSV image is not within the preset red effective range, the gray value of the pixel is set to 255.

4. The eye redness state analysis method according to claim 1, characterized by, The step of performing red threshold segmentation on the eye image to obtain an eye red region without the eyelash region further comprises: Performing regional connectivity analysis on the eye red region, and filtering out the final eye red region according to the red area and a preset area threshold.

5. The eye redness state analysis method according to claim 1, characterized by, The step of calculating the eye red proportion in the conjunctival region and the eye red proportion in the limbal region comprises: Counting the first pixel number in the conjunctival region and the second pixel number of eye red pixels in the conjunctival region; Calculating the eye red proportion in the conjunctival region based on the ratio of the second pixel number to the first pixel number; Counting the third pixel number in the limbal region and the fourth pixel number of eye red pixels in the limbal region; Calculating the eye red proportion of the limbal region based on the ratio of the fourth pixel number to the third pixel number.

6. An ocular redness state analysis device, characterized by, The method comprises the following steps: An acquisition module is configured to acquire an eye image to be detected; A conjunctival region extraction module is configured to segment the eye image based on a segmentation network to extract a conjunctival region and an eyelash region; A limbal region extraction module is configured to detect the inner edge of the conjunctival region based on an edge detection algorithm; An expansion operation is used to extend the inner edge, and the extended edge image is converted into a binary mask to obtain a binary outer expansion mask. The limbal region is obtained based on the intersection of the binary outer expansion mask and the conjunctival region. The eye red area extraction module is configured to convert the eye image from an RGB color space to an HSV color space to obtain an HSV image; perform eye red area threshold segmentation based on a red channel of the HSV image and a preset red effective range value to obtain a threshold segmentation result; and obtain an eye red area by performing an intersection operation between the threshold segmentation result after removing the eyelash area and the conjunctival area. The statistical analysis module is configured to calculate an eye red proportion in the conjunctival area and an eye red proportion in the limbal area to obtain an eye red state analysis result. 7.A computing device comprising: at least one processor; and a memory having program instructions stored therein, wherein the program instructions are configured to be adapted for execution by the at least one processor, and the program instructions comprise instructions for performing the eye red state analysis method according to any one of claims 1-5. 8.A readable storage medium having program instructions stored therein, when the program instructions are read and executed by a computing device, causing the computing device to perform the eye red state analysis method according to any one of claims 1-5.

Images