A smart wound image analysis method for WeChat Mini Programs

Abstract

This invention discloses an intelligent wound image analysis method for a WeChat mini-program, relating to the field of image recognition technology. It includes an image acquisition port and an auxiliary terminal set by the mini-program. By setting the image acquisition port and auxiliary terminal, this invention constructs image diffusion models under different networks. The output image influence analysis value can quantify the degree of influence of network factors on image analysis. Combined with a preset image influence virtual mechanism, it rationally supplements and restores real-time image data. After initially extracting required features from two-dimensional image data, it constructs a three-dimensional dynamic model of the wound, performs dynamic matching based on a standard human body image, and analyzes and judges the healing status in future stages. The predicted wound recovery volume value and the actual wound image data can obtain the degree of wound healing. It also performs false analysis and judgment on the wound healing process, helping to identify possible artifacts during wound healing and improving the quality of wound recognition.

Description

Technical Field

[0001] This invention relates to the field of image recognition technology, and in particular to a method for intelligent analysis of wound images via a mini-program. Background Technology

[0002] With the rapid development of mobile internet technology, mini-programs have been widely used in the healthcare field due to their advantages such as no download or installation required and ease of use. Simultaneously, significant breakthroughs in image processing and artificial intelligence technologies have made it possible to intelligently assess wound conditions using image analysis. This has led to the development of intelligent wound image analysis methods on mini-programs, aiming to provide more convenient, efficient, and accurate support for wound care and treatment. Through lightweight deep learning models and mobile image processing technology, real-time and precise analysis of wound features is achieved, making it suitable for telemedicine, home care, and primary healthcare scenarios. Its core principle is to leverage the local computing power of mobile devices, combined with medical image analysis and intelligent algorithms, to provide users with convenient, real-time, and private wound management services.

[0003] In existing technologies, current systems rely on single-shot images. During the process of uploading images to the cloud for analysis, the stability and speed of network transmission can affect the integrity of data and the efficiency of analysis. Image transmission may be interrupted or some data may be lost, which in turn affects the accuracy of the analysis results. For some complex wound images, there are still shortcomings in feature extraction and classification recognition. It is difficult to accurately identify the type, depth, and healing stage of the wound, and it is impossible to capture key dynamic indicators in the wound healing process. Early signs of infection may be missed, and temporary erythema during the healing process may be misclassified as infection deterioration, leading to overtreatment.

[0004] To address the aforementioned technical shortcomings, a solution is proposed. Summary of the Invention

[0005] The purpose of this invention is to: overcome the influence of network rate by constructing a network dataset and an image diffusion model to achieve data supplementation and recovery; to construct a three-dimensional dynamic model using image analysis; to analyze the wound healing process and future situation by combining standard human images; and to generate a predicted recovery volume value. Based on the predicted value and actual data, the recovery integrity is calculated to perform a false analysis of wound abnormalities.

[0006] To achieve the above objectives, the present invention adopts the following technical solution: a method for intelligent analysis of wound images on a mini-program, comprising the following steps:

[0007] Step 1: Based on the image acquisition port and auxiliary terminal set in the mini-program, acquire wound image data, preprocess the acquired historical wound image data, and construct a network dataset based on the network rate of the monitoring terminal when transmitting wound image data.

[0008] Step 2: Based on the wound image data and the network dataset, establish an image diffusion model for image analysis under different networks to output image impact analysis values;

[0009] Step 3: Obtain the wound image data in real time, perform network rate analysis according to the image diffusion model to generate image impact analysis values, and perform supplementary recovery of image data according to the preset image impact virtual mechanism to generate real-time wound image data in a normal state;

[0010] Step 4: Perform image feature analysis based on the real-time wound image data in a normal state, construct a three-dimensional dynamic model of the wound in combination with the location of the wound to generate a real-time three-dimensional dynamic model of the wound, and perform dynamic matching based on the human standard diagram according to the real-time constructed three-dimensional dynamic model of the wound to perform healing analysis and judgment at the future stage time to generate a predicted wound recovery volume value;

[0011] Step 5: Obtain the predicted wound recovery volume value, combine it with the actually obtained wound image data, calculate the wound recovery integrity, and generate a wound recovery analysis dataset by comparing the difference between the current wound state and the predicted healed state;

[0012] Step 6: Obtain the wound recovery analysis dataset, perform a false wound analysis and judgment on the state during the wound healing process, and generate an identification report of the wound image;

[0013] Perform a false wound analysis and judgment to generate an identification report of the wound image, which specifically includes the following:

[0014] S500. Obtain the two-dimensional image of the real-time wound image data in a normal state, extract the real-time wound two-dimensional color features according to the two-dimensional image, and make a joint judgment with the healing degree η of the three-dimensional dynamic model. According to the saturation value of the i-th pixel in the wound area and the saturation value of the i-th pixel in the healthy skin reference area, comprehensively obtain the average saturation value of the corresponding area, and calculate the variance according to the average saturation value of the corresponding area to obtain the wound abnormal comparison variance set as S;

[0015] S501. The identification report of the wound image includes that if η>1, it means healing delay, and S<Y, where Y is the set variance standard threshold, then it is determined as a false abnormal phenomenon;

[0016] If η<0.8, it represents normal healing, and S<Y, then it is determined as a false abnormal phenomenon.

[0017] Furthermore, in the above Step 1, the methods for obtaining the wound image data and constructing the network dataset are as follows:

[0018] A standardized image acquisition port is preset on the mini-program, and an image acquisition module that conforms to the medical image transmission protocol is set up to capture multi-dimensional image data of the wound area in real time in the injury monitoring scenario, and simultaneously record the ambient light parameters at the time of shooting.

[0019] Access medical image databases, extract historical wound image datasets and their corresponding healing process records, and construct a reference sample library with temporal characteristics;

[0020] A preliminary dynamic resolution adjustment mechanism is implemented for historical wound image data to adapt to different network environments, and the acquired historical wound image data is preprocessed and then stored.

[0021] It also includes a network monitoring module, which monitors and acquires network characteristic parameters when collecting image data. These network characteristic parameters include instantaneous effective throughput and jitter index, forming a network dataset.

[0022] Furthermore, image diffusion models for image analysis under different networks are established, specifically including the following:

[0023] S100. Based on the collected network dataset, analyze the network feature parameters under different network states to obtain the range of network dynamic parameters. Collect historical wound image data according to the image acquisition port. Perform specific integration and statistics on the historical wound image data according to the range of network dynamic parameters. Set the range of network dynamic parameters for three stages to obtain historical wound image data for three stages of network state. Perform quality assessment analysis and statistics on the historical wound image data for each stage to obtain the local prediction distortion value dataset of the original image and the output image after transmission.

[0024] S101. Based on the local prediction distortion value dataset, set the physical constraints of the initial network model based on the fully connected neural network architecture, preprocess the historical wound image dataset of the three stages of the network state to meet the requirements of model training, divide the preprocessed historical wound image data into training set, validation set and test set, and obtain the image diffusion model after training.

[0025] S103. Network feature parameters of historical wound image data are used as input layer conditions for the image diffusion model.

[0026] The estimated image distortion value is used as the output layer condition of the image diffusion model.

[0027] Furthermore, based on a preset image impact virtual mechanism, image data is supplemented and restored, specifically including the following:

[0028] S200. Obtain wound image data in real time, obtain real-time network parameter features of the image data, input the real-time network parameter features into the image diffusion model, and perform network rate analysis to obtain the estimated image distortion value in order to generate image impact analysis value.

[0029] S201. Based on the image impact analysis value and the preset image impact virtual mechanism, a matching degree analysis is performed to obtain the target image impact virtual mechanism, so as to repair the image to different degrees. The preset image impact virtual mechanism is to set the repair degree for images at different network state stages.

[0030] S202. Based on the repair level setting, supplement and restore the wound image data in the real-time state to generate real-time wound image data in the normal state.

[0031] Furthermore, the construction of a three-dimensional dynamic model of the wound specifically includes the following:

[0032] S300. Based on real-time wound image data under normal conditions, perform image feature analysis to extract the edge features, shape, and depth of the wound, and perform preliminary extraction of required features from the two-dimensional image of real-time wound image data under normal conditions.

[0033] S301. Based on the initially extracted requirements features and the location of the wound, a three-dimensional dynamic model of the wound is constructed with the center of the wound as the coordinate point. The verification point is then marked on the three-dimensional dynamic model of the wound. The center feature points of the standard human body part model are matched with the verification point. The model is then synchronously corrected according to the matching results to obtain the three-dimensional dynamic model of the wound.

[0034] S302. Based on the real-time wound image data under normal conditions presented by the three-dimensional dynamic model of the wound, the real-time wound area is calculated according to the SFM algorithm statistical model to obtain the current actual wound volume. .

[0035] Furthermore, by comparing the current wound condition with the predicted healing condition, a wound recovery analysis dataset is generated, which includes the following:

[0036] S400: Through a three-dimensional dynamic model of the wound, dynamic matching is performed based on the standard human body diagram to locate the skin mechanical parameters of the wound location, set the predicted recovery and healing time period, and predict the wound based on the set healing time period.

[0037] Predicted wound healing volume for:

[0038] ;

[0039] A patient-specific healing factor, The standard skin tension coefficient, Predict the recovery and healing timeframe;

[0040] S401. Based on the predicted wound recovery volume and the current actual wound area, perform a healing degree η analysis to obtain... Generate the current wound recovery analysis dataset based on the healing degree η.

[0041] Furthermore, it also includes a feedback verification mechanism, which verifies the image repair of real-time wound image data under normal conditions;

[0042] The real-time wound image data under normal conditions is compared with the wound image data before repair. The sharpness between the two is analyzed based on the edge calculation algorithm to obtain the sharpness between the two.

[0043] The repair rate of the wound image data before repair is determined based on the clarity. It is then compared with the standard repair rate. If the standard repair rate is not reached, the wound image data before repair is triggered again until the preset standard repair rate is reached.

[0044] In summary, due to the adoption of the above technical solution, the beneficial effects of the present invention are:

[0045] This mini-program-based intelligent wound image analysis method, by setting up an image acquisition port and an auxiliary port, can comprehensively collect wound image data, analyze network environment factors during image data transmission, construct image diffusion models under different networks, and output image impact analysis values ​​that can quantify the degree of influence of network factors on image analysis. Combined with a preset image impact virtual mechanism, it can reasonably supplement and restore real-time image data. After initially extracting the required features from the two-dimensional image data, it constructs a three-dimensional dynamic model of the wound, performs dynamic matching based on the human body standard image, and analyzes and judges the healing status in the future stage. The predicted wound recovery volume value and the actual wound image data can obtain the degree of wound healing. It can also perform false analysis and judgment on the wound healing process, which helps to identify possible artifacts in the wound healing process and improve the quality of wound recognition. Attached Figure Description

[0046] Figure 1 A schematic diagram of the process flow of the method steps of the present invention is shown. Detailed Implementation

[0047] The technical solutions of the embodiments of the present invention will be clearly and completely described below with reference to the accompanying drawings. Obviously, the described embodiments are only some embodiments of the present invention, and not all embodiments. Based on the embodiments of the present invention, all other embodiments obtained by those skilled in the art without creative effort are within the scope of protection of the present invention.

[0048] Example 1:

[0049] like Figure 1 As shown, a method for intelligent wound image analysis via a mini-program includes the following steps:

[0050] Step 1: Based on the image acquisition port and auxiliary terminal set in the mini-program, acquire wound image data, preprocess the acquired historical wound image data, and construct a network dataset based on the network rate of the monitoring terminal when transmitting wound image data.

[0051] Step 2: Based on the wound image data and network dataset, establish an image diffusion model for image analysis under different networks to output image impact analysis values;

[0052] Step 3: Acquire wound image data in real time, perform network rate analysis based on the image diffusion model, generate image impact analysis values, and supplement and restore image data according to the preset image impact virtual mechanism to generate real-time wound image data in normal state.

[0053] Step 4: Perform image feature analysis based on real-time wound image data under normal conditions, construct a three-dimensional dynamic model of the wound in combination with the location of the wound, generate a real-time three-dimensional dynamic model of the wound, perform dynamic matching based on the constructed real-time three-dimensional dynamic model of the wound and the healing analysis and judgment based on the future time stage, and generate the predicted wound recovery volume value.

[0054] Step 5: Obtain the predicted wound healing volume value, combine it with the actual acquired wound image data, calculate the wound healing integrity, and generate a wound healing analysis dataset by comparing the difference between the current wound state and the predicted healing state.

[0055] Step 6: Obtain the wound recovery analysis dataset, perform wound falsehood analysis on the state of the wound healing process, and generate a wound image recognition report;

[0056] Perform wound false positive analysis and generate a wound image recognition report, specifically including the following:

[0057] S500. Obtain a two-dimensional image of the real-time wound image data in the normal state, extract the real-time wound two-dimensional color features based on the two-dimensional image, and perform a combined judgment with the healing degree η of the three-dimensional dynamic model. According to the saturation value of the i-th pixel in the wound area and the saturation value of the i-th pixel in the healthy skin reference area, comprehensively obtain the saturation mean value of the corresponding area, calculate the variance based on the saturation mean value of the corresponding area, and obtain the wound abnormality comparison variance set as S;

[0058] S501. The recognition report of the wound image, including if η > 1, indicating delayed healing, and S < Y, where Y is the set variance standard threshold, then it is determined as a false abnormal phenomenon;

[0059] If η < 0.8, it represents normal healing, and S < Y, then it is determined as a false abnormal phenomenon.

[0060] In this solution, by setting the image acquisition port and the auxiliary end, the wound image data can be comprehensively collected, the network environment factors during image data transmission can be analyzed, an image diffusion model under different networks can be constructed, and the output image impact analysis value can quantify the impact degree of network factors on image analysis. Combining with the preset image impact virtual mechanism, the real-time image data can be rationally supplemented and restored, avoiding wound changes caused by excessive repair. After initially extracting the required features from the two-dimensional image data, a three-dimensional dynamic model of the wound is constructed, dynamically matched according to the human standard map, and the healing situation in the future stage is analyzed and judged. The predicted wound recovery volume value and the actual wound image data can obtain the healing degree between the wounds, and a false analysis and judgment of the wound healing process is carried out, which helps to identify the possible false appearances during the wound healing process and improve the quality of wound recognition.

[0061] In the first step, the wound image data acquisition method and the construction method of the network data set are as follows:

[0062] Preset a standardized image acquisition port on the mini-program side, and set an image acquisition module that conforms to the medical imaging transmission protocol. In the injury monitoring scenario, capture the multi-dimensional image data of the wound area in real time, and synchronously record the environmental light parameters at the time of shooting;

[0063] Access the medical imaging database, extract the historical wound image data set and its corresponding healing process records, and construct a reference sample library with time series characteristics;

[0064] And perform a preliminary dynamic resolution adjustment mechanism on the historical wound image data to adapt to different network environments, and preprocess and store the obtained historical wound image data;

[0065] The auxiliary device will encode the injury parameters based on the location of the injury, the time of the injury, and the patient's age. By mapping the injury parameters to the corresponding image data in a spatiotemporal manner, it will construct injury data tuples with multi-dimensional attribute labels and generate standardized injury feature labels based on the real-time acquired patient parameter information.

[0066] It also includes a network monitoring module, which monitors and acquires network characteristic parameters when collecting image data. These network characteristic parameters include instantaneous effective throughput and jitter index, forming a network dataset.

[0067] Image diffusion models for image analysis under different networks are established, specifically including the following:

[0068] S100. Based on the collected network dataset, analyze the network feature parameters under different network states to obtain the range of network dynamic parameters. Collect historical wound image data according to the image acquisition port. Perform specific integration and statistics on the historical wound image data according to the range of network dynamic parameters. Set the range of network dynamic parameters for three stages to obtain historical wound image data for three stages of network state. Perform quality assessment analysis and statistics on the historical wound image data for each stage to obtain the local prediction distortion value dataset of the original image and the output image after transmission.

[0069] S101. Based on the local prediction distortion value dataset, set the physical constraints of the initial network model based on the fully connected neural network architecture, preprocess the historical wound image dataset of the three stages of the network state to meet the requirements of model training, divide the preprocessed historical wound image data into training set, validation set and test set, and obtain the image diffusion model after training.

[0070] S103. Network feature parameters of historical wound image data are used as input layer conditions for the image diffusion model.

[0071] The estimated image distortion value is used as the output layer condition of the image diffusion model.

[0072] Based on a preset virtual image impact mechanism, image data is supplemented and restored, specifically including the following:

[0073] S200. Obtain wound image data in real time, obtain real-time network parameter features of the image data, input the real-time network parameter features into the image diffusion model, and perform network rate analysis to obtain the estimated image distortion value in order to generate image impact analysis value.

[0074] S201. Based on the image impact analysis value and the preset image impact virtual mechanism, a matching degree analysis is performed to obtain the target image impact virtual mechanism, so as to repair the image to different degrees. The preset image impact virtual mechanism is to set the repair degree for images at different network state stages.

[0075] S202. Based on the repair level setting, supplement and restore the wound image data in the real-time state to generate real-time wound image data in the normal state.

[0076] The construction of a three-dimensional dynamic model of the wound includes the following:

[0077] S300. Based on real-time wound image data under normal conditions, perform image feature analysis to extract the edge features, shape, and depth of the wound, and perform preliminary extraction of required features from the two-dimensional image of real-time wound image data under normal conditions.

[0078] S301. Based on the initially extracted requirements features and the location of the wound, a three-dimensional dynamic model of the wound is constructed with the center of the wound as the coordinate point. The verification point is then marked on the three-dimensional dynamic model of the wound. The center feature points of the standard human body part model are matched with the verification point. The model is then synchronously corrected according to the matching results to obtain the three-dimensional dynamic model of the wound.

[0079] S302. Based on the real-time wound image data under normal conditions presented by the three-dimensional dynamic model of the wound, the real-time wound area is calculated according to the SFM algorithm statistical model to obtain the current actual wound volume. .

[0080] By comparing the current wound condition with the predicted healing condition, a wound recovery analysis dataset is generated, which includes the following:

[0081] S400: Through a three-dimensional dynamic model of the wound, dynamic matching is performed based on the standard human body diagram to locate the skin mechanical parameters of the wound location, set the predicted recovery and healing time period, and predict the wound based on the set healing time period.

[0082] Predicted wound healing volume for:

[0083] ;

[0084] A patient-specific healing factor, The standard skin tension coefficient, Predict the recovery and healing timeframe;

[0085] S401. Based on the predicted wound recovery volume and the current actual wound area, perform a healing degree η analysis to obtain... Generate the current wound recovery analysis dataset based on the healing degree η.

[0086] It also includes a feedback verification mechanism, which performs image restoration verification on real-time wound image data under normal conditions;

[0087] The real-time wound image data under normal conditions is compared with the wound image data before repair. The sharpness between the two is analyzed based on the edge calculation algorithm to obtain the sharpness between the two.

[0088] The repair rate of the wound image data before repair is determined based on the clarity. It is then compared with the standard repair rate. If the standard repair rate is not reached, the wound image data before repair is triggered again until the preset standard repair rate is reached.

[0089] The size of the interval and threshold is set to facilitate comparison. The size of the threshold depends on the amount of sample data and the number of bases set by those skilled in the art for each set of sample data; as long as it does not affect the ratio between the parameter and the quantized value.

[0090] The above formulas are all dimensionless calculations. The formulas are derived from software simulations based on a large amount of collected data to obtain the most recent real-world results. The preset parameters in the formulas are set by those skilled in the art according to the actual situation.

[0091] In the two embodiments provided in this application, it should be understood that the disclosed apparatus and system can be implemented in other ways; for example, the apparatus embodiments described above are merely illustrative, and the division of modules is merely a logical functional division. In actual implementation, there may be other division methods, such as multiple modules or components can be combined or integrated into another system, or some features can be ignored or not executed; furthermore, the coupling or direct coupling or communication connection between the shown or discussed mutuals can be through some interfaces, and the indirect coupling or communication connection between the apparatus or modules can be electrical, mechanical or other forms.

[0092] The above description is only a preferred embodiment of the present invention, but the scope of protection of the present invention is not limited thereto. Any equivalent substitutions or modifications made by those skilled in the art within the scope of the technology disclosed in the present invention, based on the technical solution and inventive concept of the present invention, should be covered within the scope of protection of the present invention.

Claims

1. A method for intelligent analysis of wound images via a WeChat mini-program, characterized in that, It includes the following steps: Step 1: According to the image acquisition port and the auxiliary end set on the applet side, acquire wound image data, preprocess the acquired wound image data, and acquire the network rate during the transmission of the wound image data by the monitoring end to construct a network dataset; Step 2: Based on the wound image data and the network dataset, establish an image diffusion model for image analysis under different networks to output an image impact analysis value; Step 3: Acquire the wound image data in real time, perform network rate analysis according to the image diffusion model, generate an image impact analysis value, and perform image data supplementation and restoration according to the preset image impact virtual mechanism to generate real-time wound image data in a normal state; Step 4: Perform image feature analysis on the real-time wound image data in a normal state, combine the wound location to construct a wound three-dimensional dynamic model, generate a real-time wound three-dimensional dynamic model, and perform dynamic matching based on the human standard map according to the real-time constructed wound three-dimensional dynamic model, and perform healing analysis and judgment at the future stage time to generate a predicted wound recovery volume value; Step 5: Acquire the predicted wound recovery volume value, combine it with the actually acquired wound image data, calculate the wound recovery integrity, and generate a wound recovery analysis dataset by comparing the difference between the current wound state and the predicted healed state; Step 6: Acquire the wound recovery analysis dataset, perform a false wound analysis and judgment on the state during the wound healing process, and generate an identification report of the wound image; Perform a false wound analysis and judgment, and generate an identification report of the wound image, specifically including the following: S500. Obtain the two-dimensional image of the real-time wound image data in a normal state, extract the real-time wound two-dimensional color features according to the two-dimensional image, and perform a joint judgment with the healing degree η of the three-dimensional dynamic model. According to the saturation value of the i-th pixel in the wound area and the saturation value of the i-th pixel in the healthy skin reference area, comprehensively obtain the average saturation value of the corresponding area, and calculate the variance according to the average saturation value of the corresponding area to obtain the wound abnormal comparison variance, denoted as S; S501. The identification report of the wound image includes that if η>1, indicating a healing delay, and S<Y, where Y is the set variance standard threshold, then it is determined as a false abnormal phenomenon; If η<0.8, it represents normal healing, and S<Y, then it is determined as a false abnormal phenomenon.

2. The intelligent wound image analysis method for mini-program terminals according to claim 1, characterized in that, In the above Step 1, the method for acquiring wound image data and constructing the network dataset is as follows: Preset a standardized image acquisition port on the applet side, and set an image acquisition module that conforms to the medical image transmission protocol. In the injury monitoring scenario, capture multi-dimensional image data of the wound area in real time, and synchronously record the environmental light parameters at the time of shooting; Access the medical image database, extract the historical wound image dataset and its corresponding healing process records, and construct a reference sample library with time series characteristics; And perform a preliminary dynamic resolution adjustment mechanism on the historical wound image data to adapt to different network environments, and preprocess and store the acquired historical wound image data; It also includes a network monitoring module, which monitors and acquires network characteristic parameters when collecting image data. These network characteristic parameters include instantaneous effective throughput and jitter index, forming a network dataset.

3. The intelligent wound image analysis method for mini-program terminals according to claim 2, characterized in that, Image diffusion models for image analysis under different networks are established, specifically including the following: S100. Based on the collected network dataset, analyze the network feature parameters under different network states to obtain the range of network dynamic parameters. Collect historical wound image data according to the image acquisition port. Perform specific integration and statistics on the historical wound image data according to the range of network dynamic parameters. Set the range of network dynamic parameters for three stages to obtain historical wound image data for three stages of network state. Perform quality assessment analysis and statistics on the historical wound image data for each stage to obtain the local prediction distortion value dataset of the original image and the output image after transmission. S101. Based on the local prediction distortion value dataset, set the physical constraints of the initial network model based on the fully connected neural network architecture, preprocess the historical wound image dataset of the three stages of the network state to meet the requirements of model training, divide the preprocessed historical wound image data into training set, validation set and test set, and obtain the image diffusion model after training. S103. Network feature parameters of historical wound image data are used as input layer conditions for the image diffusion model. The estimated image distortion value is used as the output layer condition of the image diffusion model, and the image diffusion model outputs the image influence analysis value.

4. The intelligent wound image analysis method for mini-program terminals according to claim 3, characterized in that, Based on a preset virtual image impact mechanism, image data is supplemented and restored, specifically including the following: S200. Obtain wound image data in real time, obtain real-time network parameter features of the image data, input the real-time network parameter features into the image diffusion model, and perform network rate analysis to obtain the estimated image distortion value in order to generate image impact analysis value. S201. Based on the image impact analysis value and the preset image impact virtual mechanism, a matching degree analysis is performed to obtain the target image impact virtual mechanism, so as to repair the image to different degrees. The preset image impact virtual mechanism is to set the repair degree for images at different network state stages. S202. Based on the repair level setting, supplement and restore the wound image data in the real-time state to generate real-time wound image data in the normal state.

5. The intelligent wound image analysis method for mini-program terminals according to claim 1, characterized in that, The construction of a three-dimensional dynamic model of the wound includes the following: S300. Based on real-time wound image data under normal conditions, perform image feature analysis to extract the edge features, shape, and depth of the wound, and perform preliminary extraction of required features from the two-dimensional image of real-time wound image data under normal conditions. S301. Based on the initially extracted requirements features and the location of the wound, a three-dimensional dynamic model of the wound is constructed with the center of the wound as the coordinate point. The verification point is then marked on the three-dimensional dynamic model of the wound. The center feature points of the standard human body part model are matched with the verification point. The model is then synchronously corrected according to the matching results to obtain the three-dimensional dynamic model of the wound. S302. Based on the real-time wound image data under normal conditions presented by the three-dimensional dynamic model of the wound, the real-time wound area is calculated according to the SFM algorithm statistical model to obtain the current actual wound volume. .

6. The intelligent wound image analysis method for mini-program terminals according to claim 5, characterized in that, By comparing the current wound condition with the predicted healing condition, a wound recovery analysis dataset is generated, which includes the following: S400: Through a three-dimensional dynamic model of the wound, dynamic matching is performed based on the standard human body diagram to locate the skin mechanical parameters of the wound location, set the predicted recovery and healing time period, and predict the wound based on the set healing time period. Predicted wound healing volume for: ； A patient-specific healing factor, The standard skin tension coefficient, Predict the recovery and healing timeframe; S401. Based on the predicted wound recovery volume and the current actual wound area, perform a healing degree η analysis to obtain... Generate the current wound recovery analysis dataset based on the healing degree η.

7. The intelligent wound image analysis method for mini-program terminals according to claim 1, characterized in that, It also includes a feedback verification mechanism, which performs image restoration verification on real-time wound image data under normal conditions; The real-time wound image data under normal conditions is compared with the wound image data before repair. The sharpness between the two is analyzed based on the edge calculation algorithm to obtain the sharpness between the two. The repair rate of the wound image data before repair is determined based on the clarity. It is then compared with the standard repair rate. If the standard repair rate is not reached, the wound image data before repair is triggered again until the preset standard repair rate is reached.

Images