First-aid system for eye trauma in battlefield environment

Abstract

The invention relates to a first-aid system for eye trauma in a battlefield environment. The first-aid system comprises a first-aid eyeshade and a judgment device, the first-aid eyeshade comprises an eyeshade body and an elastic band, wherein the eyeshade body comprises a rigid eyeshade layer and a flexible eyeshade layer which are A plurality of through holes are formed in the eyeshade body, pressure reducing units are arranged in the through holes, when the pressure intensity in the eyeshade body exceeds 30 mmHg, the pressure reducing units are automatically opened to release pressure outwards, the pressure intensity recovers to 30 mmHg, and meanwhile the pressure reducing units cannot be opened reversely; the judging device comprises a camera and an edge processor; the processor is used for receiving images collected by the camera. The device is simple and portable, can provide safety protection for injured eyes in a field or battlefield environment, realizes complete coverage of the injured eyes, and avoids pollution and prolapse / herniation of contents in the eyes; with the progress of injury, the tissues have bleeding and edema, and the pressure in the flexible eyeshade is increased to effectively relieve the pressure, so that the eye tissue avascular necrosis and orbital apex syndrome caused by over-high pressure or over-tight binding are avoided; meanwhile, serious injury conditions are judged, and follow-up treatment is facilitated.

Description

Technical Field

[0001] The present invention belongs to the field of medical devices, and in particular, relates to a first aid system for eye injuries. Background Art

[0002] On-site first aid refers to the initial rescue activities carried out on the wounded at the scene, also known as pre-hospital first aid, which includes rescue by medical personnel and self-rescue and mutual rescue by combat personnel. It is the starting point of graded treatment. Proper on-site first aid plays an important role in reducing the mortality and disability rates and maintaining the combat effectiveness of the troops.

[0003] In particular, eyeball rupture is the focus and difficulty of emergency treatment of eye injuries. It occurs more frequently in daily traffic, accidents or wars. When the eyeball wall tissue ruptures, severe vision loss may occur, seriously affecting vision and even leading to blindness.

[0004] When eye injury occurs, the first principle is to maintain the integrity of the eyeball and preserve the contents of the eye as much as possible. The injured eye must never be exposed to the external environment for a long time, so as to provide conditions for further diagnosis and treatment by specialists and suturing and trimming of the wound. Otherwise, it will cause further loss of eyeball contents, intraocular infection, sympathetic ophthalmia, etc.

[0005] At present, the common method used both at home and abroad, including by our army and foreign armies, is to cover the injured eye with a convex piece of hard material, such as plastic or metal, to provide hard protection for the damaged and exposed contents of the eyeball and to provide conditions for specialized treatment after evacuation.

[0006] But the equipment has the following outstanding problems: 1. Single size.

[0007] Eye patches are of a uniform standard, which not only cannot reflect individual differences according to different facial parts, but also cannot completely cover the eye when a large amount of eye contents are prolapsed and bleeding and swelling are severe.

[0008] 2. The pressure is fixed and cannot adapt to changes in the injury.

[0009] The intensity of the eye patch bandage is based on the initial condition of the wounded, and it may take a long time to be transported to an emergency station or even a specialized hospital after being in the field or on the battlefield. During this process, the body will experience reactive edema or bleeding, resulting in increased tissue pressure. This fixed bandage has little effect on general body tissues and can even stop bleeding. However, eye tissues, especially the uvea, retina and optic nerve, are highly sensitive to ischemia and hypoxia. According to previous battlefield treatment experience at home and abroad, the rate of eye removal due to tissue necrosis is extremely high.

[0010] If an eye patch that is too large is used to reserve space or the bandage pressure is reduced, the eye patch may become loose and affect the field of vision of the healthy eye. More importantly, when the injured person is moving or being carried, the contents of the eye will not be properly fixed and will shake drastically, aggravating the injury.

[0011] 3. Tissue herniation.

[0012] In order to avoid the aforementioned iatrogenic injury caused by the secondary increase in tissue pressure, holes are reserved on the eye mask to relieve pressure.

[0013] However, because these holes are made directly into the hard eye patch, tissue can be squeezed and herniated through the holes.

[0014] On the one hand, small-diameter herniation not only fails to reduce pressure, but aggravates the ischemia and hypoxia of the herniated tissue; large-diameter herniation does not protect the eyes from damage.

[0015] On the other hand, herniation is directly or indirectly exposed to the air, which aggravates tissue dryness and necrosis, and is not conducive to later treatment; at the same time, a large number of open holes will greatly increase the risk of infection in the field or battlefield environment.

[0016] In addition, different eye injuries require different treatments. If the eyeball is ruptured, it is urgent to maintain the integrity of the eyeball and preserve the contents of the eye as much as possible to avoid direct and prolonged exposure of the injured eye to the external environment, so as to gain time and conditions for subsequent treatment.

[0017] Therefore, in a battlefield environment, it is necessary to quickly determine whether an eyeball rupture has occurred.

[0018] However, in a battlefield environment, it is sometimes difficult for doctors and other professionals to arrive at the scene in the first time, which will affect the rescue of the wounded.

[0019] If non-professionals can use simple and easy-to-operate means to assess and bandage injuries, it will be very beneficial to the rescue and subsequent rehabilitation of the injured.

[0020] The existing technology proposes to use neural networks to judge injuries, but this requires complex algorithms and high requirements on equipment computing power, configuration, and resources, and is not suitable for battlefield environments. In addition, it is not specifically aimed at judging eye rupture, so the accuracy is also relatively limited, making it impossible for non-professionals to operate it.

[0021] Therefore, how to quickly determine if the eyeball has ruptured and how to use a convenient device to quickly handle it is an urgent problem to be solved. Summary of the invention

[0022] In order to solve one or more of the above technical problems, it is proposed: A first aid system for eye injuries in a battlefield environment, comprising a first aid eye mask and a discrimination device; The first aid eye mask comprises an eye mask body and an elastic band, wherein the eye mask body comprises a rigid eye mask layer and a flexible eye mask layer which are tightly connected; the eye mask body has a plurality of through holes, and a decompression unit is arranged in the through holes. When the pressure inside the eye mask body exceeds 30 mmHg, the decompression unit automatically opens to release the pressure outward, and the pressure is restored to 30 mmHg, and the decompression unit will not open in the reverse direction; The identification device includes a camera and an edge processor; the processor is used to receive the image captured by the camera and process the image as follows: (1) Observation correction of eyeball image: Implement edge detection to obtain edge images , set two initial thresholds , ,and , respectively, using the above two thresholds to Perform binary segmentation to obtain the corresponding image after segmentation , , for the image , Perform edge response calculations to obtain the corresponding edge responses , ; Calculate the local feature convolution map: ;in Indicates The position offset of the center, , Represents the side length of the convolution window, and the local feature convolution map is used as the original image Observation correction of (2) Calculating a feature map based on the local feature convolution map: defining a pixel encoding vector as an eight-dimensional vector, The initial value of each dimension of the encoding vector at is 0; take the local feature convolution map as is the center and the offset is An element at , taking this element as the reference element, calculate the absolute value of the pixel difference between it and the 8 adjacent coding elements around it, take the coding element with the largest absolute value, and add 1 to the value of the corresponding coding vector dimension; when traversing through After all elements in the center neighborhood are obtained The encoding vector at the position is obtained, and then the encoding vector at all positions is obtained, which is recorded as the feature map ;in, represents coordinates, represents the dimension of the encoding vector; (3) Establishing an eyeball rupture discrimination model for discrimination: The eyeball rupture discrimination model adopts a layered neural network model, and the output is a feature map , the output layer outputs the discrimination result.

[0023] The rigid eye mask layer is located outside the flexible eye mask layer and connected to the flexible eye mask layer, and the rigid eye mask layer is smaller in size than the flexible eye mask layer.

[0024] There are 4 to 6 decompression units, which communicate one-way from the inside of the flexible eye mask layer to the outside of the rigid eye mask layer.

[0025] The eye mask body is hemispherical.

[0026] The identification device also includes a communication unit for transmitting the identification result to the medical terminal.

[0027] The medical terminal is a medical individual processing system.

[0028] The communication unit also transmits medical requests, preliminary medical condition information, and location coordinate information.

[0029] The above discriminant model is trained and the cost function used is ; in, is the category of the model output, is the category of training samples.

[0030] A method for operating the first aid system in a battlefield environment, using a discrimination device to take a photo of the wounded, after the discrimination device identifies that the eyeball is ruptured, instructs the use of a first aid eye mask, and sends a medical request, preliminary condition information, and location coordinate information to a medical individual processing system.

[0031] The discrimination device performs observation correction of the eyeball image, characteristic map calculation, and discrimination algorithm through the discrimination model.

[0032] The invention and technical effects of the present invention are as follows: 1. The present invention adopts a decompression unit setting by combining rigid and flexible structures, and utilizes the characteristics of fluidity and viscosity of eye ointment to release pressure outwards through the decompression unit for tissue bleeding and edema that may occur after eye trauma, so that the eye mask can be kept fixed and airtight while the bandage pressure can be dynamically adjusted to maintain a preset stable level; at the same time, since the contact part with the eye is a flexible structure, the present invention can also achieve individualized close application.

[0033] 2. The present invention realizes accurate extraction of data features through preprocessing algorithms, reduces the amount of data, and ensures that the neural network model can achieve a relatively lightweight design to achieve the technical effect of accurately identifying eyeball rupture. At the same time, a corresponding dedicated neural network model is designed to achieve accurate identification with less computing power support, which is suitable for use in environments with limited field resources, so that non-professionals can accurately identify through simple operations. BRIEF DESCRIPTION OF THE DRAWINGS

[0034] Figure 1 It is a perspective view of an eye mask for first aid of eye injuries of the present invention.

[0035] Figure 2 It is a front view of the decompression unit of the present invention. DETAILED DESCRIPTION

[0036] The technical solutions in the embodiments of the present invention will be described clearly and completely below in conjunction with the drawings in the embodiments of the present invention. Obviously, the described embodiments are only some embodiments of the present invention, rather than all embodiments.

[0037] Based on the embodiments of the present invention, all other embodiments obtained by ordinary technicians in the field without making any creative work shall fall within the scope of protection of the present invention.

[0038] The utility model discloses a first aid system for eye injuries in battlefield environment, which mainly comprises a distinguishing device and a first aid eye mask.

[0039] The eye mask for first aid of eye injury comprises an eye mask body and an elastic band 4.

[0040] The eye mask body comprises two layers that are tightly connected, the outer layer is a rigid eye mask layer 1, and the inner layer is a flexible eye mask layer 2.

[0041] The eye mask body is hemispherical and has a plurality of through holes thereon, and the through holes pass through the rigid eye mask layer 1 and the flexible eye mask layer 2.

[0042] A decompression unit 3 is sealed and installed in the through hole, and the inside of the eye mask body is connected to the outside in one direction through the decompression unit 3.

[0043] The rigid eye mask layer is located outside the flexible eye mask layer and connected to the flexible eye mask layer, and the rigid eye mask layer is smaller in size than the flexible eye mask layer.

[0044] The elastic bands 4 are fixed on both sides of the eye mask body, so as to facilitate wearing on the head.

[0045] When in use, apply eye ointment inside the flexible eye mask and fix it to the head with an elastic band. Regardless of whether the patient is in a standing or lying position, when the pressure of the covered content exceeds 30 mmHg, the gel-like eye ointment will be squeezed out of the eye mask body through the decompression unit. At this time, the decompression unit will automatically open due to the increase in eye ointment pressure to release pressure outward.

[0046] When the excess eye ointment is squeezed out of the eye mask body, the pressure inside the eye mask body is automatically reduced.

[0047] When the pressure returns to 30 mmHg, the eye ointment is no longer squeezed out of the eye mask body through the decompression unit, and at this time the decompression unit automatically closes and seals due to the reduction in eye ointment pressure, thereby preventing outside air and debris from entering the eye mask body through the decompression unit, i.e. ensuring that the decompression unit will not open in the reverse direction.

[0048] As a preferred embodiment, since the present invention adopts a double-layer structure, the inner layer is a flexible eye mask layer, and the size is larger than the outer rigid eye mask layer, so that after the eye ointment has overflowed, when the pressure is reduced, the unique elasticity of the inner layer flexible material and the resulting tight bonding ability with the eye ointment ensure that no "hole" will occur between the eye ointment and the inner layer, thereby avoiding excessive overflow of the eye ointment, or insufficient amount of eye ointment in the eye mask body when the pressure is reduced, thereby affecting the protective effect.

[0049] The number of decompression units is 4-6, and they communicate one-way from the inside of the eye mask body to the outside of the eye mask body.

[0050] As an example, the decompression unit may include a cylinder 3-2, a cover 3-1, an elastic member 3-3, and a connecting shaft 3-4.

[0051] The connecting shaft is located in the middle of one end of the cylinder, and the connecting shaft is movably connected to the left and right cover bodies 3-1.

[0052] The connecting shaft is also connected to one end of the elastic member 3-3, and the other end of the elastic member 3-3 is connected to the inner side of the cover body 3-1.

[0053] When the internal pressure increases, the left and right covers 3-1 can be opened outwards under the push of the internal pressure, and the elastic member generates an inward pulling force, such as Figure 2 The middle arrow is a schematic diagram of the eye ointment overflow path; when the internal pressure decreases to a level less than the tension generated by the elastic member, the left and right covers 3-1 are pulled back to the ends of the cylinder by the elastic member, thereby achieving sealing.

[0054] The radii of the left and right covers 3 - 1 are both larger than the outer radius of the cylinder.

[0055] The elastic force of the control elastic member can be set so that the cover opens when the pressure is greater than 30 mmHg, and the cover closes under the action of the elastic force when the pressure is less than 30 mmHg.

[0056] Usage example: In the wild or battlefield environment, when a casualty suffers an eye injury and there are no suitable rescue or medical conditions, nearby personnel can take photos with the recorder they are wearing, and the discrimination algorithm can quickly identify whether the eyeball is ruptured (the specific method will be described in detail below).

[0057] If the system indicates that the eyeball has ruptured, nearby personnel can remove the eye patch, apply an excess of tobramycin dexamethasone eye ointment to the flexible eye patch (no need to consider the dosage carefully), then cover the eye, use an elastic band to wrap around the head, and quickly put on the eye patch.

[0058] This allows complete coverage of the injured eye, avoiding contamination and prolapse / herniation of eye contents.

[0059] If the eye ointment is applied excessively, the pressure inside the eye mask will increase when it is worn, so that the excess eye ointment will be automatically discharged from the decompression unit. Therefore, there is no need to carefully consider the dosage. It is suitable for non-professionals to operate in an emergency without too much learning.

[0060] As the injury progresses, the tissue may bleed and swell, and the pressure inside the flexible eye mask increases. When the pressure exceeds 30 mmHg, the eye ointment is released outward through the decompression unit to relieve the pressure, avoiding ischemic necrosis of eye tissue and orbital apex syndrome caused by excessive pressure or tight bandage. Although the pressure of 30 mmHg is higher than the normal intraocular pressure of 21 mmHg, it does not cause obvious damage to the eye in a short period of time. On the contrary, the relatively high pressure can play a role in compressing bleeding and preventing edema to a certain extent.

[0061] As a preferred embodiment, the first aid system may further include a distinguishing device, so as to assist in judging whether the eyeball is ruptured and whether it is necessary to wear a first aid goggles immediately.

[0062] The identification device includes a camera and an edge processor.

[0063] The camera is used to capture eye trauma and transmit the images to the processor.

[0064] The processor is used to receive images, and pre-process and identify the images to determine whether the eye trauma has caused a rupture of the eyeball.

[0065] The identification device can exist independently, such as a handheld terminal, a mobile phone, etc.; the functions of the camera and the processor can also be integrated into existing equipment, such as a recorder on a helmet, that is, the existing equipment has a camera and a processor to collect images and perform identification to achieve the above functions.

[0066] As a preferred embodiment, the identification device further includes a communication unit, which can transmit the identification result to a medical terminal, for example, to a medical individual soldier processing system, so as to send a medical request, preliminary condition information, location coordinate information, etc. to a nearby medical individual soldier, and guide the medical individual soldier to come for processing. In this way, after the emergency processing by the surrounding non-professionals, further professional treatment can be carried out.

[0067] As a further preference, when transmitting, the communication unit first determines the nearest medical soldier from past communication processes with multiple nearby medical individual processing systems, and preferentially sends the above request and information to the nearest medical individual processing system.

[0068] As a preferred embodiment, the identification device also includes a communication unit, which can transmit the identification result to a remote server, such as a command center system, so that the command center can understand the injury situation and facilitate the deployment of medical resources for rescue and subsequent treatment.

[0069] The processor implements the following determination method: Step 1: Observation correction and feature extraction method of eyeball image.

[0070] A high-definition image of the eye area to be diagnosed is acquired, with a resolution of 1024*1024 or above, and the acquired original image is input into the observation correction system of step 1 for processing.

[0071] The above-mentioned observation correction system corrects the original image according to the apparent observation of the image, so as to make the subsequent feature extraction process more accurate and improve the accuracy of the diagnosis of eyeball rupture.

[0072] For the original image Implement edge detection to obtain edge images .

[0073] Set two initial thresholds , ,and , respectively, using the above two thresholds to Perform binary segmentation to obtain the corresponding image after segmentation , .

[0074] Taking a 256-grayscale image as an example, the optimal value is set through experiments , .

[0075] , are binary images. , Test edge response.

[0076] by For example, the edge response test process is as follows.

[0077] Define the edge response function: ; In the above formula , , is the marginal parameter, agreed And take the integer, The value is less than half of the image's unidirectional size and is rounded to an integer. , are the pixel coordinates of the image; is the response parameter.

[0078] Test the edge response of each pixel in a binary image based on the given edge parameters.

[0079] That is, for pixels with a value of 1 in a binary image, their coordinates , Substitute into equation 1 and determine whether equation 1 is satisfied. If so, it means that the pixel has a response. Otherwise, it means that the pixel has no response.

[0080] For the current given parameters, record the total number of pixels that respond.

[0081] For all parameter combinations in the edge parameter value space, the total number of response pixels is calculated according to the above steps, and a set of parameters with the largest total number of response pixels is taken as the edge response parameters of the binary image.

[0082] At the same time, the corresponding pixels with a value of 1 in the binary image are marked as the edge response of the binary image, denoted as .

[0083] Similarly, obtaining The corresponding edge response .

[0084] Edge Response , They are also binary images.

[0085] According to the edge response , , calculate the original image feature map.

[0086] First, calculate the local feature convolution map: ; in Indicates The position offset of the center.

[0087] , Represents the side length of the convolution window.

[0088] Local feature convolution map as the original image Observation correction.

[0089] Secondly, the feature map is calculated based on the local feature convolution map.

[0090] Original Image The feature map of is defined as an image matrix with the same size as the original image, and the element at each coordinate is a vector associated with the neighborhood at the corresponding coordinate in the original image.

[0091] The neighborhood side length described above is equivalent to the convolution window defined in Formula 2.

[0092] Define the pixel encoding vector as an eight-dimensional vector. The initial value of each dimension of the encoding vector at is 0.

[0093] Each dimension of the encoding vector represents the position of the encoding element relative to the reference element, and the encoding element is adjacent to the reference element.

[0094] Assume the coordinates of the reference element are ,So: If the encoding element coordinates are , corresponds to the first dimension of the encoded vector.

[0095] If the encoding element coordinates are , corresponds to the second dimension of the encoded vector.

[0096] If the encoding element coordinates are , then it corresponds to the third dimension of the encoding vector.

[0097] If the encoding element coordinates are , corresponds to the fourth dimension of the encoding vector.

[0098] If the encoding element coordinates are , corresponds to the fifth dimension of the encoding vector.

[0099] If the encoding element coordinates are , corresponds to the sixth dimension of the encoding vector.

[0100] If the encoding element coordinates are , corresponds to the seventh dimension of the encoding vector.

[0101] If the encoding element coordinates are , corresponds to the eighth dimension of the encoding vector.

[0102] Take the convolutional graph is the center and the offset is An element at , taking this element as the reference element, calculate the absolute value of the pixel difference with the 8 adjacent coding elements around it, take the coding element with the largest absolute value, and add one to the value of the corresponding coding vector dimension.

[0103] When traversing After all elements in the center neighborhood are obtained The encoding vector at .

[0104] Furthermore, the encoding vectors at all positions are obtained and recorded as feature maps .

[0105] in, represents coordinates, Represents the dimension of the encoding vector.

[0106] Step 2: Establish an eyeball rupture discrimination model for discrimination.

[0107] The discriminant model inputs the feature map obtained in step 1 to obtain the diagnosis result (rupture 1 / non-rupture-1).

[0108] The discriminant model is trained to obtain the output of effective diagnostic results.

[0109] The discriminant model adopts a layered neural network model, and its hierarchical structure and calculation method are as follows.

[0110] First layer: ; in, is a two-dimensional convolution kernel, Represents the relative offset coordinates of the convolution kernel, represents the coordinates in the image, is the linear bias, is the dimension of the aforementioned encoding vector.

[0111] is the activation function of the neural network. It is defined as follows: ; in, It is a control parameter used to balance the convergence speed and classification performance. .

[0112] is the circumference constant, represents the natural exponential function.

[0113] Compared with the classic sigmoid function, the above activation function can improve the computational performance and classification performance of the model.

[0114] Second layer: ; in, ; In the above formula, , is a two-dimensional convolution kernel, but compared to Reduce the number of parameters by half, that is: ; Represents the relative offset coordinates of the convolution kernel, represents the coordinates in the image, is the linear bias, is the dimension of the aforementioned encoding vector.

[0115] is the activation function of the neural network (same as above).

[0116] The second layer introduces the aforementioned edge response to suppress the parameter amount of the non-edge part, highlighting the model's ability to model the features of the edge part and improving the classification performance of eyeball rupture observation.

[0117] Third layer: ; This layer is a pooling layer, which enables the network to model multi-scale features and improve classification performance.

[0118] Fourth floor: ;in, ; In the above formula, , is a two-dimensional convolution kernel, but compared to Reduce the number of parameters by half, that is: ; Represents the relative offset coordinates of the convolution kernel, represents the coordinates in the image, is the linear bias, is the dimension of the aforementioned encoding vector.

[0119] is the activation function of the neural network (same as above).

[0120] The fourth layer is similar to the second layer. Considering the scale after pooling, the coordinates of the corresponding scaled convolution map are used when the edge is prominent.

[0121] Fifth floor: ; This layer is a fully connected layer. represents the linear coefficient of full connection, is the corresponding linear offset.

[0122] After the fully connected layer, it is mapped to the output vector , define the vector as a 64-dimensional vector.

[0123] is the activation function (same as above).

[0124] The sixth layer (output layer): ; is the fully connected linear coefficient of the output layer, is the corresponding linear offset.

[0125] is the activation function (same as above); outputs the result of eyeball rupture diagnosis. During sample training, Corresponding to the sample where no rupture occurred, Corresponding to the sample where rupture occurred.

[0126] When identifying Indicates that no rupture occurred. Indicates that a rupture has occurred.

[0127] Compared with the existing methods, the optimized model parameters of the method of the present invention are small, which is sufficient to complete learning and recognition on embedded and edge computing devices. It does not need to exchange data and samples with the background server. It is suitable for stand-alone use in the field, has high computing efficiency, and has a considerable recognition accuracy.

[0128] The cost function used to train the above model is ; in, is the category of the model output, is the category of training samples.

[0129] Experimental results show that the behavior recognition rate is comparable to that of classical comparable methods, and it can be deployed in a single-machine edge computing environment with high computational efficiency, making it convenient for immediate use in the field.

[0130] The above is only a preferred embodiment of the present invention. It should be pointed out that for ordinary technicians in this technical field, several improvements and modifications can be made without departing from the principle of the present invention. These improvements and modifications should also be regarded as the scope of protection of the present invention.

Claims

1. A first aid system for eye injuries in battlefield environments, characterized by: Includes identification device and emergency goggles; The first aid eye mask comprises an eye mask body and an elastic band, wherein the eye mask body comprises a rigid eye mask layer and a flexible eye mask layer which are tightly connected; the eye mask body has a plurality of through holes, and a decompression unit is arranged in the through holes. When the pressure inside the eye mask body exceeds 30 mmHg, the decompression unit automatically opens to release pressure outwards, and when the pressure is restored to 30 mmHg, the decompression unit will not open in the reverse direction; The identification device includes a camera and an edge processor; the edge processor is used to receive an image captured by the camera and process the image as follows: (1) Observation correction of eyeball image: Implement edge detection to obtain edge images , set two initial thresholds , ,and , respectively, using the above two thresholds to Perform binary segmentation to obtain the corresponding image after segmentation , , for the image , Perform edge response calculations to obtain the corresponding edge responses , ; Calculate the local feature convolution map: ;in Indicates The center position offset, , Represents the side length of the convolution window, and the local feature convolution map is used as the original image Observation correction of (2) Calculating a feature map based on the local feature convolution map: defining a pixel encoding vector as an eight-dimensional vector, The initial value of each dimension of the encoding vector at is 0; take the local feature convolution map as is the center and the offset is An element at , taking this element as the reference element, calculate the absolute value of the pixel difference between it and the 8 adjacent coding elements around it, take the coding element with the largest absolute value, and add 1 to the value of the corresponding coding vector dimension; when traversing through After all elements in the center neighborhood are The encoding vector at the position is obtained, and then the encoding vector at all positions is obtained, which is recorded as the feature map ;in, represents coordinates, represents the dimension of the encoding vector; (3) Establishing an eyeball rupture discrimination model for discrimination: The eyeball rupture discrimination model adopts a layered neural network model, and the output is a feature map , the output layer outputs the discrimination result.

2. The emergency rescue system according to claim 1, characterized in that: The rigid eye mask layer is located outside the flexible eye mask layer and connected thereto, and the size of the rigid eye mask layer is smaller than that of the flexible eye mask layer.

3. The emergency rescue system according to claim 1, characterized in that: The number of the decompression units is 4-6, and they communicate one-way from the inner side of the flexible eye mask layer to the outer side of the rigid eye mask layer.

4. The emergency rescue system according to claim 1, characterized in that: The eye mask body is hemispherical.

5. The emergency rescue system according to claim 1, characterized in that: The identification device further includes a communication unit, and the communication unit is used to transmit the identification result to the medical terminal.

6. The emergency system according to claim 5, characterized in that: The medical terminal is a medical individual processing system.

7. The emergency rescue system according to claim 5, characterized in that: The communication unit also transmits medical requests, preliminary medical condition information, and location coordinate information.

8. The emergency rescue system according to claim 1, characterized in that: The cost function used to train the above discriminant model is: ; in, is the category of the model output, is the category of training samples.

9. A method for operating the intelligent emergency system according to claims 1-8 in a battlefield environment, characterized in that: The identification device is used to take a photo of the injured person. After the identification device recognizes that the eyeball is ruptured, it instructs the use of an emergency goggles and sends the medical request, preliminary medical information, and location coordinate information to the medical individual processing system.

10. The method according to claim 9, characterized in that: The discrimination device performs observation correction of the eyeball image, characteristic map calculation, and discrimination algorithm through the discrimination model.

Images