Chicken anomaly prediction method and device based on egg image and computer

Abstract

The invention relates to a chicken anomaly prediction method and device based on an egg image and a computer. The method comprises the following steps: performing eggshell optical feature recognition on multi-source eggshell acquisition data of an egg to be analyzed to obtain eggshell image feature data; according to the eggshell image feature data, determining an egg with an abnormal eggshell from the to-be-analyzed eggs, and selecting abnormal eggshell feature data corresponding to the egg with the abnormal eggshell from the eggshell image feature data; according to the abnormal eggshell feature data, performing disease dynamics simulation on a suspicious chicken corresponding to the egg with the abnormal eggshell to obtain chicken disease dynamics data corresponding to the suspicious chicken; and performing disease risk prediction on the suspicious chicken according to the chicken disease dynamics data to obtain disease risk prediction data of the suspicious chicken. By adopting the method, the abnormity of the chicken can be effectively predicted through the egg image.

Description

Technical Field

[0001] The present application relates to the field of computer technology, and in particular to a method, device and computer for predicting chicken abnormalities based on egg images. Background Art

[0002] In traditional technologies, egg image recognition technology mainly focuses on egg quality detection, crack identification and freshness assessment. Although computer vision and deep learning have been applied in livestock and poultry monitoring, there is still a large technical gap in the related field of combining egg images with livestock and poultry monitoring, resulting in the inability to effectively predict chicken abnormalities through egg images. Summary of the invention

[0003] Based on this, it is necessary to provide a chicken abnormality prediction method, device and computer based on egg images, which can effectively predict chicken abnormalities through egg images, in order to solve the above technical problems.

[0004] In a first aspect, the present application provides a chicken abnormality prediction method based on egg images, comprising: Performing eggshell optical feature recognition on the multi-source eggshell collected data of the egg to be analyzed to obtain eggshell image feature data; Determine eggs with abnormal eggshells from the eggs to be analyzed according to the eggshell image feature data, and select abnormal eggshell feature data corresponding to the eggs with abnormal eggshells from the eggshell image feature data; According to the abnormal eggshell characteristic data, a disease dynamics simulation is performed on the suspicious chickens corresponding to the eggs with abnormal eggshells to obtain the chicken disease dynamics data corresponding to the suspicious chickens; According to the chicken disease dynamics data, disease risk prediction is performed on the suspicious chicken to obtain disease risk prediction data of the suspicious chicken.

[0005] In a second aspect, the present application also provides a chicken abnormality prediction device based on egg images, comprising: An eggshell data recognition module is used to perform eggshell optical feature recognition on the multi-source eggshell collected data of the eggs to be analyzed to obtain eggshell image feature data; An abnormal eggshell selection module is used to determine eggs with abnormal eggshells from the eggs to be analyzed according to the eggshell image feature data, and to select abnormal eggshell feature data corresponding to the eggs with abnormal eggshells from the eggshell image feature data; An abnormal eggshell analysis module is used to perform disease dynamics simulation on suspicious chickens corresponding to the eggs with abnormal eggshells according to the abnormal eggshell characteristic data, so as to obtain disease dynamics data of the chickens corresponding to the suspicious chickens; The chicken disease prediction module is used to predict the disease risk of the suspicious chickens according to the chicken disease dynamics data to obtain the disease risk prediction data of the suspicious chickens.

[0006] In a third aspect, the present application further provides a computer, comprising a memory and a processor, wherein the memory stores a computer program, and when the processor executes the computer program, it implements any step of a method for predicting chicken abnormalities based on egg images.

[0007] The above-mentioned chicken abnormality prediction method, device and computer based on egg images, through optical feature recognition of multi-source eggshell collection data of the eggs to be analyzed, accurately extracts eggshell image feature data, and realizes intelligent screening of eggs with abnormal eggshells. Further, by analyzing the abnormal eggshell feature data, the disease dynamics simulation is performed after associating the suspicious chickens, and the chicken disease dynamics data of the suspicious chickens is obtained; the disease risk prediction of the suspicious chickens is performed using the chicken disease dynamics data, and the obtained disease risk prediction data can effectively predict the abnormality of the chickens through egg images, thereby realizing early warning in the breeding process, which is helpful for the accurate prevention and control of poultry diseases, reducing breeding risks, improving breeding efficiency and economic benefits, and reducing the impact of disease transmission on the health of poultry flocks. BRIEF DESCRIPTION OF THE DRAWINGS

[0008] In order to more clearly illustrate the technical solutions in the embodiments of the present application or the related technologies, the drawings required for use in the embodiments or the related technical descriptions are briefly introduced below. Obviously, the drawings described below are only some embodiments of the present application. For ordinary technicians in this field, other drawings can be obtained based on these drawings without paying creative work.

[0009] Figure 1 A diagram showing an application environment of a chicken abnormality prediction method based on egg images in one embodiment; Figure 2 A schematic diagram of a flow chart of a chicken abnormality prediction method based on egg images in one embodiment; Figure 3 The figure is a flow chart of a method for obtaining disease dynamics data of chickens in a first embodiment; Figure 4 A schematic diagram of a process for obtaining macroscopic disease analysis data of chickens in one embodiment; Figure 5 A schematic diagram of a process for obtaining chicken microscopic disease analysis data in one embodiment; Figure 6 A schematic diagram of a flow chart of a second chicken disease dynamics data method in one embodiment; Figure 7A schematic diagram of a process for obtaining disease risk prediction data in one embodiment; Figure 8 A schematic flow chart of a method for obtaining eggshell image feature data in one embodiment; Fig. 9 A schematic flow chart of a method for obtaining shallow layer characteristic data of eggshells in one embodiment; Fig.10 is a structural block diagram of a chicken abnormality prediction device based on egg images in one embodiment; Fig.11 FIG. 4 is a diagram showing the internal structure of a computer in one embodiment. DETAILED DESCRIPTION

[0010] In order to make the purpose, technical solution and advantages of the present application more clearly understood, the present application is further described in detail below in conjunction with the accompanying drawings and embodiments. It should be understood that the specific embodiments described herein are only used to explain the present application and are not used to limit the present application.

[0011] The chicken abnormality prediction method based on egg images provided in the embodiment of the present application can be applied to Figure 1 In the application environment shown. Among them, the terminal 102 communicates with the server 104 through the network. The data storage system can store the data that the server 104 needs to process. The data storage system can be integrated on the server 104, or it can be placed on the cloud or other network servers. Among them, the server 104 can be implemented with an independent server or a server cluster composed of multiple servers.

[0012] In an exemplary embodiment, Figure 2 As shown in FIG. 1 , a chicken abnormality prediction method based on egg images is provided, and the method is applied to Figure 1 The server in the example is used to illustrate, including the following steps 202 to 208. Among them: Step 202 , performing eggshell optical feature recognition on the multi-source eggshell collected data of the egg to be analyzed to obtain eggshell image feature data.

[0013] Among them, the eggs to be analyzed can be eggs selected for testing in the farm or production process. These eggs may come from different batches or breeding environments, and their health status has not yet been determined. Further analysis is needed to determine whether there are any abnormalities, such as eggshell defects, color abnormalities or pathological characteristics.

[0014] Among them, multi-source eggshell collection data can be eggshell image data collected by different types of detection equipment (such as hyperspectral imaging, visible light cameras, infrared scanning, etc.) and multi-angle shooting technology, including optical and physical characteristics of the eggshell such as color, smoothness, cracks, thickness, etc.

[0015] Among them, eggshell optical feature recognition can be a process of analyzing eggshell image data and extracting its optical features through computer vision, spectral analysis or deep learning.

[0016] Among them, the eggshell image feature data can be structured data obtained after optical feature recognition, including eggshell color value, texture parameters, spectral reflectivity, edge detection information, etc.

[0017] Specifically, a high-resolution camera, a multi-spectral or hyperspectral imaging device is used to perform optical inspection on the eggs to be analyzed, that is, multi-source eggshell data is collected on the eggs to be analyzed to obtain multi-source eggshell collection data; the multi-source eggshell collection data is input into a computer vision model and / or an image processing model, and the multi-source eggshell collection data is preprocessed, including denoising, image enhancement and edge detection, and further color space conversion, texture analysis, wavelet transform and other methods are used to extract optical characteristics of the eggshell such as color, brightness, surface roughness, crack distribution, etc. from the preprocessed multi-source eggshell collection data, and feature dimensionality reduction technology (such as principal component analysis PCA) is used to optimize the data dimension to obtain structured eggshell image feature data.

[0018] Step 204, determining eggs with abnormal eggshells from the eggs to be analyzed based on the eggshell image feature data, and selecting abnormal eggshell feature data corresponding to the eggs with abnormal eggshells from the eggshell image feature data.

[0019] Among them, eggs with abnormal eggshells can be eggs that are detected to have abnormal eggshells after optical feature recognition, such as abnormal eggshell color, uneven thickness, cracks or other defects, indicating that the health of the laying hens is abnormal.

[0020] Among them, the abnormal eggshell feature data can be feature data corresponding to eggs with abnormal eggshells extracted from the eggshell image feature data, specifically including the abnormality type (such as color abnormality, cracks, gloss abnormality, etc.), severity and spatial distribution information.

[0021] Specifically, the eggshell image feature data is input into the trained eggshell abnormality recognition model for comparison and analysis, and the machine learning algorithm or deep learning algorithm (such as support vector machine SVM, convolutional neural network CNN) in the trained eggshell abnormality recognition model is used for classification to automatically identify eggs with abnormal eggshells and eggs with normal eggshells. In the recognition process, the trained eggshell abnormality recognition model mainly detects features such as abnormal eggshell color (such as too dark color, whitish), cracks (micro cracks on the surface or through cracks), uneven eggshell thickness, and abnormal glossiness; once an abnormal eggshell is detected through the above-mentioned detection criteria, the egg is marked as an eggshell abnormal egg, and the abnormal feature data corresponding to the eggshell abnormal egg is extracted from the eggshell image feature data and stored.

[0022] Step 206, based on the abnormal eggshell characteristic data, disease dynamics simulation is performed on the suspicious chickens corresponding to the eggs with abnormal eggshells to obtain chicken disease dynamics data corresponding to the suspicious chickens.

[0023] Among them, suspicious chickens can be laying hens that are suspected of having health problems or potential infection risks after tracing back the abnormal eggshell characteristic data. Suspicious chickens may have a certain disease or be in the incubation period of the disease.

[0024] Among them, disease dynamics simulation can be a simulation analysis of the disease transmission process of suspicious chickens and their flocks through mathematical models or computer simulation, taking into account the disease's incubation period, transmission route, infection rate and recovery rate to predict the development trend of the disease in the flock.

[0025] Among them, the chicken disease dynamics data can be quantitative data obtained through disease dynamics simulation, including the infection probability of chickens, disease transmission path, incubation period, herd immunity status, etc.

[0026] Specifically, through the traceability analysis of eggs with abnormal eggshells, the laying hens corresponding to the abnormal eggshells are marked as suspicious individuals, and the possible range of abnormal chicken flocks is determined by combining breeding records, egg-laying time, biological information and other data. Based on the above-mentioned suspicious individuals and the range of abnormal chicken flocks as guiding information, the abnormal eggshell feature data is input into the poultry disease dynamics model (such as SIR model, SEIR model or individual-based simulation model). The poultry disease dynamics model will simulate the disease transmission pattern in the chicken flock. During the simulation process, the poultry disease dynamics model will comprehensively consider factors including the health status of the chickens, immunization history, feeding environment (such as temperature and humidity, ventilation conditions), contact network and other factors, calculate the possible disease infection of the suspicious chickens, and finally generate chicken disease dynamics data, including incubation period estimation, disease transmission probability, possible sources of infection and diffusion trends.

[0027] Step 208: predicting the disease risk of the suspicious chickens based on the disease dynamics data of the chickens, and obtaining disease risk prediction data of the suspicious chickens.

[0028] Among them, disease risk prediction can be based on chicken disease dynamics data, combined with statistical models or artificial intelligence algorithms, to predict the disease development trends of suspicious chickens in the short, medium and long term, assess infection risks, etc.

[0029] Among them, disease risk prediction data can be the specific output results of disease risk prediction, including the infection risk index, disease transmission possibility, health status assessment and warning level of each suspected chicken. These data can be used to guide breeding management decisions, such as early intervention, isolation or optimization of epidemic prevention measures.

[0030] Specifically, based on the disease dynamics data of chickens and the time series analysis algorithm, the short-term, medium-term and long-term disease risks of suspicious chickens are predicted. Among them, the short-term risk prediction mainly combines the current abnormal eggshell characteristics and the disease dynamics data of chickens to evaluate the probability of infection and the development trend of early symptoms in the next few days; the medium-term prediction sets the parameters of the disease evolution model according to the disease dynamics data of chickens, and combines the contact network, feeding environment and immune status of the flock to analyze the disease progression of suspicious chickens in a few weeks, including the evolution of the incubation period, the aggravation of symptoms and the possible secondary transmission risk; the long-term prediction relies on the disease dynamics data of chickens for nonlinear modeling, and considers the breeding cycle, immune tolerance and historical epidemic data to evaluate the health stability of chickens throughout the production cycle, the risk of chronic diseases and the possibility of overall disease outbreaks in the farm. Finally, the short-term, medium-term and long-term prediction data are spliced ​​to obtain the disease risk prediction data. Among them, the disease risk prediction data is used to guide the short-term high-risk individuals to focus on monitoring or isolation, the medium-term risk chickens can strengthen immunity or adjust the feeding strategy, and the long-term high-risk individuals need to be combined with breeding optimization or early elimination programs.

[0031] In the above-mentioned chicken abnormality prediction method based on egg images, optical feature recognition is performed on the multi-source eggshell collection data of the eggs to be analyzed, and the eggshell image feature data is accurately extracted to realize intelligent screening of eggs with abnormal eggshells. Further, by analyzing the abnormal eggshell feature data, the disease dynamics simulation is performed after associating the suspicious chickens to obtain the chicken disease dynamics data of the suspicious chickens; the disease risk prediction data of the suspicious chickens is predicted by using the chicken disease dynamics data, and the obtained disease risk prediction data can effectively predict the abnormality of the chickens through egg images, thereby realizing early warning in the breeding process, which is helpful for the accurate prevention and control of poultry diseases, reducing breeding risks, improving breeding efficiency and economic benefits, and reducing the impact of disease transmission on the health of poultry flocks.

[0032] In an exemplary embodiment, Figure 3 As shown, according to the abnormal eggshell characteristic data, the disease dynamics simulation is performed on the suspicious chickens corresponding to the eggs with abnormal eggshells to obtain the chicken disease dynamics data corresponding to the suspicious chickens, including steps 302 to 306. Among them: Step 302, based on the abnormal eggshell feature data, the disease transmission evolution analysis is performed on the suspicious chickens to obtain the macroscopic disease analysis data of the suspicious chickens.

[0033] Among them, the evolution analysis of disease transmission can be based on the system dynamics model to conduct macro-level mathematical modeling and simulation analysis of the disease transmission process of suspicious chickens and their flocks.

[0034] Among them, the macro-disease analysis data of chickens can be the group-level disease transmission characteristic data calculated by the system dynamics model, including key information such as the infection rate change trend of the chicken flock, the disease transmission speed, the group immunity level, and the disease outbreak inflection point.

[0035] Specifically, the health status, breeding environment and historical epidemic data of suspicious chickens are obtained to determine the possible disease types and disease parameters corresponding to the disease types, including the contact pattern of the flock, vaccination status, incubation period of the disease, transmission rate, recovery rate, etc.; the above data are combined with abnormal eggshell characteristic data to construct a system dynamics (SD) model, based on SEIR (susceptible-S, latent-E, infected-I, recovered-R) or its extended model, and differential equations are used to describe the evolution of the health status of chickens over time. By setting the conversion rate of different states of the system dynamics model, the spread trend of the disease in the chicken flock is simulated, and the impact of key variables in the system dynamics model (such as the basic reproduction number R0, onset time, and herd immunity ratio) on the spread of the disease is analyzed. The system dynamics model is further solved using numerical simulation methods to simulate the disease spread under different scenarios, and predict the infection probability and disease development trajectory of suspicious chickens in the short, medium and long term. Finally, the analysis generates macro disease analysis data for chickens, including the overall infection trend of the flock, the inflection point of disease transmission, and the demand for immunization coverage, etc. Step 304, based on the abnormal eggshell characteristic data, a spatiotemporal analysis of disease transmission of the suspicious chickens is performed to obtain microscopic disease analysis data of the suspicious chickens.

[0036] Among them, the spatiotemporal analysis of disease transmission can be based on a multi-agent model to simulate the disease transmission pattern of suspicious chickens at the individual level, focusing on how the disease is transmitted between different chickens and spreads in the breeding environment.

[0037] Among them, the chicken microscopic disease analysis data can be the individual-level disease transmission information generated by multi-agent model simulation, including the infection probability of each chicken, potential transmission chains, local high-risk areas, super-spreading individual identification and other data.

[0038] Specifically, while the system dynamics model analyzes the disease transmission trend at the macro level, a multi-agent model is used to simulate the spatiotemporal characteristics of disease transmission of suspicious chickens at the individual level. In the process of multi-agent model analysis, based on abnormal eggshell feature data and breeding environment information, the chicken house is divided into different functional areas (such as feeding area, drinking area, and roosting area), and each chicken is defined as an agent, with an independent health status (susceptible, latent, infected or recovered) and behavior pattern (activity range, contact frequency, movement path, etc.). Then, the interaction rules between agents are set to simulate the activity trajectory of chickens in the breeding environment and the disease transmission process caused by contact transmission, while considering the impact of external factors (such as temperature and humidity changes, air flow) on virus transmission; using multiple rounds of simulation calculations, the disease transmission path, spatial diffusion pattern and high-risk transmission nodes between individuals are analyzed, the disease outbreak risk within the flock is quantified, and susceptible individuals and possible super spreaders are identified. The final analysis result is the microscopic disease analysis data of chickens, including individual infection probability, local high-risk areas, potential transmission chains, etc.

[0039] Step 306, coupling the chicken macroscopic disease analysis data and the chicken microscopic disease analysis data to obtain chicken disease dynamics data.

[0040] Specifically, the disease transmission trends at the group level are extracted from the macroscopic disease analysis data of chickens generated by the system dynamics model, such as changes in infection rate over time, group immunity level, and disease peak prediction. At the same time, the microscopic disease analysis data output by the multi-agent model is analyzed to identify individual infection paths, spatial transmission patterns, high-risk areas, and key transmission nodes; and the mapping relationship between the macroscopic and microscopic levels is established by using data alignment and feature matching methods, so that individual infectious behaviors are embedded in the group transmission model, indicating that the overall spread trend of the disease can be dynamically adjusted to reflect the local infection situation. Fusion simulation calculations are used to simulate the disease evolution process under different scenarios, such as adjustment of prevention and control measures, optimization of individual isolation strategies, etc., to evaluate their impact on the overall health of the chicken flock, and obtain the coupled chicken disease dynamics data.

[0041] In this embodiment, at the macro level, by analyzing the disease transmission evolution of abnormal eggshell feature data based on the system dynamics model (SD), it is possible to simulate the evolution trend of the disease in the whole chicken flock, predict the infection rate, changes in immune level and the risk of epidemic outbreak, and provide a scientific basis for the group prevention and control strategy. At the micro level, by analyzing the disease transmission spatiotemporal analysis of abnormal eggshell feature data through a multi-agent model (Multi-Agent), the disease transmission path between individuals, super-spreading individual identification and high-risk area distribution are accurately portrayed, thereby optimizing local prevention and control measures. Finally, the macro disease analysis data and the micro disease analysis data are coupled to generate chicken disease dynamics data, and a comprehensive disease prediction system that takes into account individuals and groups, short-term and long-term, space and time is constructed. Not only can the health status of suspicious chickens be accurately identified, but also the allocation of epidemic prevention resources can be optimized, and the disease early warning ability and prevention and control efficiency of the farm can be improved, thereby reducing the risk of epidemic outbreaks, ensuring the health of the chicken flock, and improving the efficiency of breeding production.

[0042] In an exemplary embodiment, Figure 4 As shown, according to the abnormal eggshell characteristic data, the disease transmission evolution analysis is performed on the suspicious chickens to obtain the macroscopic disease analysis data of the suspicious chickens, including steps 402 to 410. Among them: Step 402, setting the initial inventory of the disease transmission evolution model according to the actual flock size and actual health status corresponding to the suspicious chickens.

[0043] Among them, the actual flock size can be the actual number of chickens in the current farm, including the total number of all chickens, the individual distribution in different breeding areas, and the stocking density of the flock.

[0044] The actual health status can be the health status of different individuals in the flock, including the number distribution of healthy individuals, latently infected individuals, confirmed infected individuals and recovered individuals.

[0045] The initial inventory can be the initial state variable in the system dynamics (SD) model at the beginning of the disease transmission simulation, which represents the number or proportion of individuals in the flock in various health states (such as susceptible, latent, infected, and recovered).

[0046] Specifically, according to the actual flock size (such as the total number of chickens, stocking density) and actual health status (including the number of healthy, latently infected, confirmed infected and recovered individuals) of the suspected chickens, the initial state variables of the disease transmission evolution model (system dynamics model) are set. That is, under the SEIR model framework, the proportions of chicken populations corresponding to different health statuses are set, for example: susceptible individuals (S), latent individuals (E), infected individuals (I), and recovered individuals (R); in addition, key parameters such as the basic reproduction number (R0), disease incubation period, and transmission cycle need to be introduced to obtain the initial inventory.

[0047] Step 404, inputting the abnormal eggshell characteristic data into the disease transmission evolution model to obtain the calculation flow rate of the disease transmission evolution model.

[0048] The calculation flow rate may be the conversion rate between different health states in the system dynamics model, including the rate at which susceptible individuals become infected, the rate at which chickens become ill during the incubation period, the rate at which infected individuals recover or die, etc.

[0049] Specifically, the abnormal eggshell characteristic data is input into the disease transmission evolution model, and the disease transmission evolution model analyzes the abnormal eggshell characteristic data, that is, converts the disease risk-related features in the abnormal eggshell characteristic data, such as abnormal eggshell color, thickness deviation, crack conditions, etc., into health impact parameters that can be identified by the disease transmission evolution model. For example, abnormal eggshells may reflect the decreased immunity of chickens or the potential infection risk of specific diseases. These data are then input into the disease transmission evolution model. The disease transmission evolution model (system dynamics model) describes the dynamic changes in the health status of the flock through differential equations, including: the flow rate of susceptible individuals (S) turning into latent individuals (E), which is affected by the frequency of contact between the flock, the degree of health damage corresponding to abnormal eggshells, and environmental factors (such as stocking density and air quality); the flow rate of latent individuals (E) turning into infected individuals (I), which is determined by the incubation period of the disease and the virus replication rate; the flow rate of infected individuals (I) turning into recovered individuals (R), which depends on the natural recovery cycle of the disease or the effectiveness of vaccination, that is, the flow rate of disease transmission, that is, the conversion rate between different health states. It is worth noting that in the process of calculating the flow rate, the disease transmission evolution model will combine the flock density, historical epidemic data and external intervention measures to dynamically adjust the disease transmission parameters.

[0050] Step 406, updating the initial inventory according to the calculated flow rate to obtain an updated inventory of the disease transmission evolution model.

[0051] The updated inventory may be the updated value of the flock health status variable after calculating the flow rate in each cycle, for example, while susceptible individuals decrease, latent individuals increase, or infected individuals are transformed into recovered individuals.

[0052] Specifically, the flow rate is used to calculate the change in each health state. For example, according to the infection rate of susceptible individuals (S) turning into latent individuals (E), the number of latently infected chickens added in this time step is calculated, and the number of susceptible individuals is reduced accordingly. Similarly, latent individuals (E) will turn into infected individuals (I) at the rate at the end of the incubation period, and infected individuals (I) will enter the recovery state (R) according to the recovery rate. The calculation process is controlled by the differential equation in the disease transmission evolution model, and combined with environmental factors (such as temperature and humidity, feed nutrition, and ventilation conditions), the relevant model parameters are adjusted synchronously during the inventory update process to ensure that the disease transmission process conforms to biological laws and breeding environment characteristics. The updated health state data is then stored and used as the initial inventory for the next time step, that is, the updated inventory.

[0053] Step 408, uses the updated inventory as the initial inventory, and uses the abnormal eggshell feature data at the next moment as the abnormal eggshell feature data, and returns to execute the step of inputting the abnormal eggshell feature data into the disease transmission evolution model to obtain the calculation flow rate of the disease transmission evolution model, until the actual number of cycles of the disease transmission evolution model triggers the preset number of cycles, and obtains the iterative disease analysis data of the chickens.

[0054] The actual number of cycles may be the number of cycles actually performed during the disease transmission simulation process.

[0055] The preset number of cycles may be the maximum number of cycles set during the disease transmission simulation process.

[0056] Among them, the iterative disease analysis data of chickens can be the health status change data gradually accumulated over time during the entire disease transmission simulation process, including the number of susceptible, latent, infected and recovered individuals at different times, as well as disease transmission rate, infection peak prediction and other information.

[0057] Specifically, in each actual cycle, the updated inventory is used as the new initial inventory of the disease transmission evolution model, and the abnormal eggshell feature data at the next moment is used as the new abnormal eggshell feature data of the disease transmission evolution model, that is, the chicken population density, contact pattern, immune status and other factors at the next moment are used as the disease transmission evolution model to recalculate the disease transmission flow rate, including the infection rate of susceptible individuals, the conversion rate of the incubation period, the recovery rate, etc. Further, the health state variables are adjusted according to the calculated flow rate to simulate the continuous spread trend of the disease, and the changes in the number of healthy individuals, latent individuals, infected individuals and recovered individuals at different time points are recorded to form preliminary iterative disease analysis data. In this process, the disease transmission evolution model will dynamically correct the disease transmission parameters according to the actual farm's epidemic prevention measures (such as vaccination, isolation strategy) to more accurately simulate the epidemic spread trend. The iterative process of the disease transmission evolution model will continue to execute until the actual number of cycles triggers the preset number of cycles, and finally the actual number of cycles is output to trigger the preset number of cycles.

[0058] Step 410, using the chicken house environment information corresponding to the abnormal chickens, perform scenario intervention correction on the chicken iterative disease analysis data to obtain the chicken macro disease analysis data.

[0059] Among them, the chicken house environmental information can be the farm environmental factors that affect the health of the chickens and the spread of diseases, including temperature and humidity, air quality, light, stocking density, chicken activity range, vaccination status, feed supply, etc.

[0060] Among them, scenario intervention correction can be to adjust the simulation data after the disease transmission simulation is completed, combined with the actual chicken house environment information, to improve the accuracy of the prediction results. For example, if the chicken house ventilation system is good, the disease transmission rate may be lower than the value calculated by simulation, so the flow rate parameters need to be adjusted; if the vaccination rate is high, the proportion of infected individuals can be appropriately reduced.

[0061] Specifically, the key environmental variables related to the chicken house are identified from the chicken house environmental information corresponding to the abnormal chickens, including temperature and humidity, air circulation, stocking density, lighting conditions, nutrition supply status and vaccination records, and these environmental factors are further quantified and set based on historical data to correct weights. For example, a higher stocking density may increase the contact transmission rate, and a good ventilation system may reduce the probability of airborne disease transmission. The above-mentioned weighted data are used to dynamically adjust the key variables of the iterative disease analysis data of the chickens, such as the disease transmission rate and health state conversion probability, to simulate the actual spread of the disease in a real breeding environment. The corrected macroscopic disease analysis data of the chickens is obtained.

[0062] In this embodiment, the initial inventory of the disease transmission evolution model is set by the actual chicken flock size and actual health status to ensure that the disease transmission simulation matches the actual chicken flock status and enhance the authenticity of the prediction; the disease transmission flow rate is dynamically calculated using abnormal eggshell feature data to achieve real-time update of individual health status (susceptible, latent, infected, recovered), ensuring that the disease transmission model can reflect the epidemic development trend in different time steps. The inventory is further updated by cyclic iterative calculation, so that the model can track the evolution trajectory of the disease in the chicken flock and accurately predict the future epidemic change trend. Finally, the scene intervention correction is combined with the chicken house environmental information to optimize the disease transmission rate and simulate the impact of different environmental variables on the spread of the epidemic, thereby generating more realistic chicken macro disease analysis data, and can predict the development trend of the chicken flock epidemic through automated dynamic simulation, optimize the allocation of prevention and control resources, improve disease early warning capabilities, and help farms adopt precise epidemic prevention strategies (such as isolation, immune intervention, and optimized ventilation conditions) to minimize the risk of epidemic spread and improve the health level of chickens and the economic benefits of breeding.

[0063] In an exemplary embodiment, Figure 5 As shown, according to the abnormal eggshell characteristic data, the disease transmission time-space analysis of the suspicious chickens is performed to obtain the microscopic disease analysis data of the suspicious chickens, including steps 502 to 508. Among them: Step 502: construct a spatiotemporal model of disease transmission based on the actual flock size and actual health status of the suspected chickens.

[0064] Among them, the disease transmission spatiotemporal model can be a model based on multi-agent modeling, simulating the disease transmission process within the chicken flock, and combining spatial factors to analyze the disease diffusion pattern.

[0065] The initial infection probability may be the initial probability that each chicken may have been infected with the disease without additional intervention at the beginning of the disease transmission simulation.

[0066] The initial latent state may be a state in which some chickens may have been infected but have not yet shown clinical symptoms at the initial stage of simulation.

[0067] Specifically, according to the actual flock size of the suspected chicken, the chicken house is divided into multiple spatial units (such as feeding area, drinking area, roosting area, etc.), and each chicken is regarded as an intelligent agent, and is given an independent health status (such as susceptible, latent, infected, recovered) and behavior pattern (such as activity range, contact frequency, movement path). In the initialization stage of the disease transmission spatiotemporal model (multi-agent model), based on the delineation of each chicken (each intelligent agent) in each spatial unit, combined with the historical epidemic data and actual health status of the chicken flock, the initial infection probability (that is, the possibility of a chicken being infected at the initial moment) and the initial latent state (that is, whether it is in the disease incubation period but has not shown symptoms) of the disease transmission spatiotemporal model are set.

[0068] Step 504, optimizing the initial infection probability and the initial latent state according to the abnormal eggshell characteristic data to obtain the real-time infection probability and the real-time latent state.

[0069] The real-time infection probability can be the individual infection risk adjusted dynamically according to the spread of the disease during the simulation. The real-time infection probability will be continuously updated with the contact frequency between chickens, environmental changes, and the enhancement or attenuation of individual immune capacity.

[0070] The real-time latent state can be a dynamic state of whether an individual is in the latent period at a certain time step during the disease transmission simulation. This state evolves over time. For example, an initially latent individual may enter the infectious period after multiple time steps, while a susceptible individual may newly enter the latent state after contacting the infection source.

[0071] Specifically, the abnormal eggshell feature data is input into the disease transmission spatiotemporal model, and the health influencing factors corresponding to the disease transmission spatiotemporal model (such as decreased immunity, nutritional deficiencies, and possible pathogen exposure) are used to dynamically optimize the initial infection probability and initial latent state in the model. For example, if the eggshells produced by a chicken are highly abnormal, the probability of infection increases accordingly, and the latent state may be adjusted to a higher risk level. At the same time, the disease transmission spatiotemporal model also combines the health status of neighboring chickens to dynamically adjust the individual infection risk to make it more consistent with the actual disease transmission situation, and obtain the real-time infection probability and real-time latent state.

[0072] Step 506, analyzing the interactive evolution of diseases of each of the suspicious chickens according to the disease transmission rules and proximity rules of the disease transmission spatiotemporal model, and obtaining chicken transmission disease analysis data.

[0073] Among them, the disease transmission rules can be mathematical or logical rules used to simulate how the disease spreads from one individual to another in a disease transmission spatiotemporal model.

[0074] Among them, the proximity rule can be a rule that the spatial distance between individuals affects the probability of disease transmission during the disease transmission simulation process.

[0075] Among them, disease interaction evolution can be a process in which the health status of individuals changes continuously due to mutual contact or environmental influences during the simulation process.

[0076] Among them, the chicken infectious disease analysis data can be the individual-level disease transmission data calculated by the disease transmission spatiotemporal model, including the infection probability of each chicken, the evolution of the latent state, the disease transmission chain, the contact frequency, the possible super-spreading individuals and the distribution of high-risk areas and other information.

[0077] Specifically, according to the disease transmission rules of the spatiotemporal model of disease transmission (such as the calculation of infection probability based on contact transmission or air transmission) and the proximity rule (i.e. the impact of the distance between chickens on the risk of infection), during the simulation of the disease interaction evolution of each suspected chicken, each chicken moves in the chicken house according to its daily behavior pattern and interacts with other chickens in different areas to simulate the disease transmission path. For example, in the feeding area or drinking area, a high contact frequency may lead to a higher probability of infection, while less active individuals may have a lower risk of infection. Through multiple rounds of simulation of disease interaction evolution, the disease transmission at different time points is analyzed, high-risk individuals, super-spreading individuals and potential transmission chains are identified, and the chicken infectious disease analysis data is obtained, including individual infection probability, disease spread trend in the flock and high risk levels in different areas.

[0078] Step 508, based on the chicken house environment information corresponding to the abnormal chickens, the chicken infectious disease analysis data is subjected to scene intervention correction to obtain the chicken microscopic disease analysis data.

[0079] Specifically, the chicken house environmental information corresponding to the abnormal chickens includes information such as chicken house temperature and humidity, air quality, ventilation system, stocking density, lighting conditions, disinfection frequency, vaccination coverage, and feed supply, and analyzes the direct impact of these factors on the disease transmission rate and individual infection probability; for example, high-density breeding may increase individual contact opportunities and increase the diffusion rate of contact-transmitted diseases, while a good ventilation system may reduce the infection probability of airborne diseases. After further quantification of these environmental factors, correction weights are set based on historical data. For example, in a high humidity environment, some pathogens have a longer survival time, so the virus transmission ability of infected individuals will be appropriately increased; if the farm takes strict disinfection measures, the transmission probability between chickens can be reduced. The above-mentioned weighted data are used to dynamically adjust the disease transmission rate, individual infection probability, and transmission chain of the chicken infectious disease analysis data to adjust the identification results of high-risk areas and super-spreading individuals in the chicken flock, and obtain the corrected chicken microscopic disease analysis data.

[0080] In this embodiment, the initial infection probability and initial latent state of the disease transmission spatiotemporal model are set by the actual flock size and actual health status corresponding to the suspicious chickens to ensure the accuracy of the simulation data; the infection probability and latent state are dynamically optimized in combination with abnormal eggshell feature data, so that the model can reflect the health risks of suspicious chickens in real time and improve the prediction accuracy. On this basis, the interactive evolution of diseases between chickens is simulated by disease transmission rules and proximity rules, and individual infection paths, high-risk transmission areas and super-spreading individuals are accurately portrayed to generate chicken infectious disease analysis data. Finally, the scene intervention correction, the model further combines the chicken house environmental information (such as temperature and humidity, ventilation conditions, density distribution, etc.), optimizes the disease transmission rate and individual infection risk, so that the prediction results are closer to the actual situation, and the obtained chicken microscopic disease analysis data can provide a scientific basis for precise epidemic prevention measures, such as isolation of high-risk individuals, adjustment of feeding layout, optimization of ventilation system or dynamic adjustment of immunization strategy, thereby reducing the risk of disease spread, improving breeding management efficiency, and reducing the impact of epidemics on breeding economy.

[0081] In an exemplary embodiment, Figure 6 As shown, the chicken macroscopic disease analysis data and the chicken microscopic disease analysis data are coupled to obtain the chicken disease dynamics data, including steps 602 to 608. Among them: Step 602, using the chicken macroscopic disease analysis data, optimize the interactive evolution trend of the chicken microscopic disease analysis data to obtain optimized microscopic disease analysis data.

[0082] Among them, the interactive evolution trend can be the law that during the disease transmission process, individuals influence each other due to factors such as contact, activity patterns, and immune status, resulting in the dynamic changes in the disease transmission mode and speed over time.

[0083] Among them, the optimized microscopic disease analysis data can be the data obtained by correcting and optimizing the individual transmission situation using macroscopic disease data when simulating the transmission of individual diseases in chickens at the microscopic level.

[0084] Specifically, since the macroscopic disease analysis data of chickens include the overall disease transmission speed in the flock, the infection rate change trend, the group immunity level and the disease peak prediction, while the microscopic disease analysis data of chickens focuses on the disease interaction at the individual level, the local transmission path and the influence relationship between neighboring chickens. In the optimization process, the individual infection rate of the microscopic disease analysis data of chickens is first compared with the overall infection rate of the macroscopic disease analysis data of chickens. If the individual infection rate is too low or too high, the contact probability, the incubation period and the inter-individual transmission coefficient are adjusted to make the individual transmission behavior more in line with the macroscopic epidemiological law. Further combined with the immunization coverage of the macroscopic disease analysis data of chickens, the individual recovery rate and secondary infection probability in the microscopic disease analysis data of chickens are optimized to more realistically simulate how chickens affect the spread of the disease after immunization, and obtain optimized microscopic disease analysis data.

[0085] Step 604, integrating the chicken microscopic disease analysis data to construct global microscopic disease analysis data corresponding to the suspicious chickens.

[0086] Among them, the global microscopic disease analysis data can be the overall microscopic disease transmission information formed by integrating the disease transmission data of all individuals.

[0087] Specifically, the health status change data of all individual chickens at different time steps, including key indicators such as infection probability, latent state evolution, disease transmission chain, contact frequency, and distribution of high-risk areas, are used to obtain a microscopic disease analysis data set for chickens. In order to improve the parseability and consistency of the data, the data set for microscopic disease analysis of chickens is standardized, and the disease states of different time steps and different individuals in the microscopic disease analysis data set of chickens are converted into a unified structured data set to ensure that the disease transmission relationship between different individuals can be clearly represented. Spatial modeling and social network analysis methods are further used to mine the disease transmission path within the flock, identify high-risk transmission areas, super-spreading individuals (i.e., individuals with high contact rates and high infection probability), and possible disease transmission chains. For example, by analyzing the movement trajectory and contact pattern of chickens in the chicken house, it is possible to identify which areas are the core hotspots of disease transmission and which chickens play a key role in the transmission hub in the group. In order to improve the reliability of the data, data smoothing is also performed on abnormal data points (such as a sudden increase in infection rate in a short period of time or abnormally high transmission probability of some individuals) to avoid extreme values ​​affecting the overall analysis results and obtain global microscopic disease analysis data.

[0088] Step 606, using the global microscopic disease analysis data, correct the chicken macroscopic disease analysis data to obtain corrected macroscopic disease analysis data.

[0089] Among them, the correction of macro disease analysis data can be to use the global micro disease analysis data to adjust the epidemic spread trend at the macro level, so that the macro prediction is more consistent with the data obtained from the actual disease spread situation in the chicken flock.

[0090] Specifically, the individual infection rate, latent state evolution, super-spreading individual influence and other indicators in the global micro-disease analysis data are matched with the changes in the group infection rate, disease transmission speed, and immunization coverage in the macro-disease analysis data of chickens. If the overall infection rate of the macro model is found to be abnormal, its key parameters such as the basic reproduction number (R0), disease transmission rate, and recovery rate are adjusted to make it more consistent with the actual transmission path reflected by the micro data. The local high-risk areas and transmission chains of the macro-disease analysis data of chickens are further analyzed to identify factors that are not fully considered by the macro-analysis at the micro level, such as the high contact rate in certain specific areas leading to a high local transmission rate, or the influence of some super-spreading individuals is underestimated. In view of these situations, when correcting the macro-disease analysis data of chickens, the disease transmission weights of different regions will be dynamically adjusted to optimize the model's prediction ability for local disease spread; and in order to ensure the stability of the macro-disease analysis data, a time series smoothing algorithm will be introduced to avoid the instability of the macro-data prediction results due to excessive fluctuations in individual data, and obtain the corrected macro-disease analysis data.

[0091] Step 608, embedding the optimized microscopic disease analysis data into the modified macroscopic disease analysis data to obtain the chicken disease dynamics data.

[0092] Specifically, the optimized microscopic disease analysis data at the individual level is fused with the modified macroscopic disease analysis data at the group level, so that the fused data can not only reflect the evolution of the disease at the macro level, but also accurately characterize the individual transmission behavior. For example, in the final dynamic data, not only the disease development curve of the chicken group as a whole is included, but also microscopic details such as individual transmission paths, spatial diffusion patterns, and high-risk area identification are embedded to obtain the disease dynamics data of chickens.

[0093] In this embodiment, the disease evolution trend at the group level is provided by macro disease data, and the interactive evolution trend of macro data optimizing micro data is optimized, which can improve the accuracy of micro disease analysis and match individual disease transmission behavior with the overall epidemic development. Through the global integration of micro disease data, a more detailed individual transmission network can be constructed, high-risk transmission areas and super-spreading individuals can be identified, and a basis for precise intervention can be provided. On this basis, the macro data is corrected using global micro disease data, which can optimize the group disease transmission rate, infection peak prediction and immune coverage analysis, so that the macro disease prediction is closer to the actual situation. Finally, the optimized micro data is nested into the corrected macro data to form a complete chicken disease dynamics data, which can simultaneously reflect the individual disease evolution trend and the group epidemic development law, and is used for precise prevention and control strategy formulation, such as dynamic isolation, intelligent immunization planning and disease spread inhibition, thereby effectively reducing the risk of disease transmission and improving the health management level and production efficiency of farms.

[0094] In an exemplary embodiment, Figure 7 As shown, according to the disease dynamics data of the chickens, the disease risk prediction of the suspicious chickens is performed to obtain the disease risk prediction data of the suspicious chickens, including steps 702 to 710. Among them: Step 702, setting a disease prediction period corresponding to a disease risk prediction according to the chicken disease dynamics data.

[0095] Among them, the disease prediction period can be a disease risk prediction range of different time scales set based on chicken disease dynamics data, including short-term prediction, medium-term prediction and long-term prediction.

[0096] Specifically, a time series analysis of the disease dynamics data of chickens is conducted to analyze the overall trend of disease transmission within the flock, the evolution path of individual infection, and the impact of environmental factors on disease spread, and to identify key nodes such as disease transmission rate, infection peak, and immune formation cycle, so as to determine the reasonable range of short-term, medium-term, and long-term predictions. Short-term predictions (a few days to a week) mainly focus on the rapid spread of the disease, and evaluate the short-term outbreak risk of the epidemic through high-frequency data updates (such as daily infection change rate); medium-term predictions (several weeks to a month) combine the immune response of chickens and environmental control measures (such as vaccination, feeding adjustments) to simulate the spread and evolution of the disease; long-term predictions (several months or the entire breeding cycle) use nonlinear modeling, combined with the establishment of immune barriers, long-term environmental changes, and breeding cycle factors, to predict the evolution trend of the overall health status of the flock. Finally, based on these analysis results, the time dimension of disease risk prediction is set, that is, it is divided into disease prediction periods.

[0097] Step 704, using the chicken disease discrete dynamics algorithm, based on the chicken disease dynamics data, short-term disease prediction is performed on the suspicious chickens to obtain short-term disease risk data.

[0098] Among them, the chicken disease discrete dynamic algorithm can be a computational method for short-term disease risk prediction. It discretizes the health status of the chicken flock (such as susceptible, latent, infected, and recovered) and iterates the calculation with a smaller time step (such as daily or hourly) to simulate the rapid spread of the disease.

[0099] Specifically, in the prediction of short-term risks, a discrete dynamic algorithm for chicken diseases is used to simulate the rapid evolution of the health status of suspected chickens based on chicken disease dynamics data to predict their infection risk in the next few days to a week. During the simulation of the discrete dynamic algorithm for chicken diseases, it runs in a discrete calculation mode with a small time step. Each time step is iteratively calculated based on the disease transmission rate, individual health status changes (susceptible, latent, infected, recovered), and environmental factors (such as stocking density and ventilation conditions). Short-term predictions are mainly used to identify individuals who may become ill in a short period of time, detect whether the disease will spread rapidly within the flock, and provide early intervention recommendations, such as emergency isolation of high-risk individuals or feed / drug adjustment strategies, to generate short-term disease risk data.

[0100] Step 706, using the chicken disease intervention evolutionary algorithm, based on the chicken disease dynamics data, mid-term disease prediction is performed on the suspicious chickens to obtain mid-term disease risk data.

[0101] Among them, the chicken disease intervention evolutionary algorithm can be a modeling method for medium-term disease risk prediction. It not only simulates the natural spread of the disease, but also dynamically evaluates the impact of different epidemic prevention intervention measures (such as vaccination, isolation, and breeding environment optimization) on the spread of the disease.

[0102] Specifically, in the prediction of medium-term risks, the chicken disease intervention evolutionary algorithm is used, combined with chicken disease dynamics data, to simulate the disease evolution of suspicious chickens in the next few weeks to a month. In the process of simulating the chicken disease intervention evolutionary algorithm, the disease development path, individual immune response, environmental influencing factors and the intervention effect of disease prevention and control measures are comprehensively considered to evaluate the disease transmission trend of chickens in the medium term. Unlike short-term predictions, medium-term predictions not only focus on the natural spread of the disease, but also dynamically simulate the impact of different epidemic prevention measures (such as vaccination, stocking density adjustment, ventilation improvement) on the spread of the disease, analyze whether these intervention strategies can effectively reduce the infection rate or shorten the duration of the epidemic, and generate medium-term disease risk data.

[0103] Step 708, using the chicken disease nonlinear algorithm, based on the chicken disease dynamics data, long-term disease prediction is performed on the suspicious chickens to obtain long-term disease risk data.

[0104] Among them, the chicken disease nonlinear algorithm can be an advanced computational method for long-term disease risk prediction, specifically used to analyze complex disease transmission patterns and long-term health trends.

[0105] Specifically, in the prediction of long-term risks, the nonlinear algorithm for chicken diseases is used to simulate the health trends of chickens in the next few months or the entire breeding cycle based on historical disease dynamics data. In the process of simulating the nonlinear algorithm for chicken diseases, since the long-term disease transmission is affected by many complex factors, such as individual immune tolerance, long-term changes in the chicken house environment, and the cycle of epidemics during the breeding cycle, the nonlinear algorithm for chicken diseases uses nonlinear modeling methods (such as neural networks, autoregressive prediction, and chaos analysis) to analyze the evolution pattern of the disease in the long term and predict the possible risk of chronic diseases, secondary infection risks, or the possibility of new outbreaks. Long-term predictions are mainly used for breeding planning at the strategic level, such as evaluating the effectiveness of long-term immunization strategies, predicting the cycle of epidemic outbreaks, optimizing population health management plans, etc., to generate long-term disease risk data.

[0106] Step 710, concatenate the short-term disease risk data, the medium-term disease risk data, and the long-term disease risk data to obtain disease risk prediction data.

[0107] Specifically, the short-term disease risk data, medium-term disease risk data, and long-term disease risk data of different time scales are aligned in time series to ensure that the high-precision trend analysis of short-term predictions can naturally transition to trend changes in medium-term and long-term predictions. The aligned data are further smoothed to avoid excessive fluctuations in predictions between different time scales and ensure the stability and interpretability of disease risk assessment. Finally, long-term disease risk data are generated for multi-level disease prevention and control decisions. Short-term predictions can be used for rapid response and emergency isolation, medium-term predictions can be used to optimize disease control strategies, and long-term predictions can help with macro-planning of aquaculture health management.

[0108] In this embodiment, by setting a reasonable disease prediction period, targeted analysis can be performed on the characteristics of disease transmission within different time ranges; for short-term prediction, a chicken disease discrete dynamic algorithm is used to quickly identify high-risk individuals in the next few days to a week through high-frequency data updates, providing data support for early intervention and emergency isolation; for medium-term prediction, a chicken disease intervention evolution algorithm is used, combined with disease control measures (such as vaccination, environmental regulation) to simulate the disease spread trend within a few weeks to a month, so as to evaluate the effectiveness of prevention and control measures, optimize immunization strategies and breeding environment management; for long-term prediction, a chicken disease nonlinear algorithm is used, and a complex disease transmission model is used to analyze the long-term evolution pattern of the disease, the formation of immune barriers and the potential outbreak cycle, providing a scientific basis for the long-term health management of the farm and the adjustment of the disease prevention and control strategy. By splicing short-term, medium-term and long-term disease risk data to form complete disease risk prediction data, it can provide all-round decision support from short-term emergency prevention and control to long-term immune optimization, improve the intelligent level of disease management in farms, reduce disease losses, and improve the health and production efficiency of chickens.

[0109] In an exemplary embodiment, Figure 8 As shown, the eggshell optical feature recognition is performed on the multi-source eggshell collected data of the egg to be analyzed to obtain eggshell image feature data, including steps 802 to 806. Among them: Step 802, performing eggshell three-dimensional visible light feature recognition on the multi-source eggshell collected data of the egg to be analyzed to obtain eggshell surface feature data.

[0110] Among them, the three-dimensional visible light feature recognition of the eggshell can be carried out by using image recognition and neural network technologies to identify the optical features of the egg surface to obtain information such as the eggshell's color, smoothness, texture structure, crack conditions and light reflection characteristics.

[0111] Among them, the eggshell surface feature data can be the external physical and optical feature information of the eggshell obtained by recognizing the three-dimensional visible light features of the eggshell, including color distribution, brightness uniformity, reflective properties, surface roughness, crack morphology, concave-convex changes and texture structure.

[0112] Specifically, the eggs to be analyzed are imaged using three-dimensional visible light imaging technologies such as industrial camera shooting, multi-spectral imaging, structured light scanning or laser triangulation, to obtain multi-source eggshell collection data. The multi-source eggshell collection data is input into an image feature recognition model, and the multi-source eggshell collection data of the eggs to be analyzed is subjected to surface optical feature recognition through the image feature recognition model, wherein the recognition content includes the color distribution, texture structure, glossiness, crack conditions and surface roughness of the eggshell, and combined with three-dimensional reconstruction technology, the pattern obtained by surface optical feature recognition is embedded in the three-dimensional surface model of the eggshell, and finally the eggshell surface feature data is generated.

[0113] Step 804, based on the eggshell surface feature data, perform eggshell optical coherence tomography analysis on the multi-source eggshell collected data of the egg to be analyzed to obtain eggshell shallow layer feature data.

[0114] Among them, eggshell optical coherence tomography can be a high-precision imaging technology based on low-coherence light interference, which is used to detect the shallow structure of the eggshell, that is, the microscopic characteristics below the surface of the eggshell. This technology can penetrate the outer surface of the eggshell and provide non-destructive cross-sectional imaging to analyze the internal microcracks, interlayer thickness, calcification degree, pore structure and mineral deposition of the eggshell.

[0115] Among them, the shallow characteristic data of the eggshell can be the internal surface structure information of the eggshell obtained through optical coherence tomography (OCT), including microcrack distribution, local thickness changes, mineral deposition state, interlayer bonding strength, etc.

[0116] Specifically, the area and dimension of the optical coherence tomography are determined based on the surface feature data of the eggshell, and optical coherence tomography (OCT) technology is further used to scan the shallow structure inside the eggshell (such as microcracks in the eggshell, internal interlayer thickness distribution, degree of calcification and potential hidden cracks) through the principle of low-coherence light interference, so as to realize non-invasive detection of the shallow structure of the eggshell. Microscopic defects that are invisible to the naked eye can be detected, such as microcracks caused by stress concentration in the eggshell, or local structural abnormalities caused by uneven mineral deposition. Finally, the shallow characteristic data of the eggshell are obtained to supplement the limitations of the surface characteristics.

[0117] Step 806, fusing the eggshell surface feature data and the eggshell shallow layer feature data to determine the eggshell image feature data.

[0118] Specifically, according to the eggshell surface feature data obtained by the three-dimensional visible light feature recognition of the eggshell, key areas (such as abnormal surface color and smoothness change areas) are determined, and these areas are spatially mapped with the shallow structure data obtained by optical coherence tomography (OCT) so as to analyze the correspondence between the surface and the shallow layer in the same coordinate space. The feature extraction and fusion algorithm is used to reduce the data dimension and pattern match the texture, crack distribution, optical transmittance, microstructure and other information of the surface and shallow layer. For example, if the surface feature data detects an optical reflection anomaly, and the shallow feature data shows that there are microcracks in the area, it is comprehensively judged that the eggshell may have structural damage or mineral deposition anomalies. In the fusion process, in order to improve the accuracy of data fusion, multi-scale analysis technology is used to perform feature fusion at the macroscopic (overall surface morphology), mesoscopic (local area details), and microscopic (internal microcracks) levels to generate fused eggshell image feature data.

[0119] In this embodiment, the eggshell three-dimensional visible light feature recognition is performed through multi-source eggshell collection data to obtain information such as eggshell color, smoothness, cracks, reflectivity and surface morphology, and evaluate the eggshell quality from the appearance level; based on the eggshell surface feature data, OCT technology is used to conduct in-depth analysis of the eggshell shallow structure, detect hidden defects such as microcracks, local thickness changes, mineral deposition, pore structure, etc., to supplement the limitations of visible light detection; the surface feature data and shallow feature data are integrated to construct complete eggshell image feature data to ensure that the detection results not only include surface abnormalities, but also reveal potential quality risks inside the eggshell. It can achieve a comprehensive analysis from macroscopic appearance to microstructure, greatly improve the accuracy of eggshell quality assessment, help screen abnormal eggshells, improve the reliability of egg grading detection, and provide key data support for chicken health traceability.

[0120] In an exemplary embodiment, Fig. 9 As shown, the eggshell optical coherence tomography analysis is performed on the multi-source eggshell data of the egg to be analyzed to obtain the eggshell shallow layer characteristic data, including steps 902 to 908. Among them: Step 902, partitioning the eggshell of the egg to be analyzed according to the eggshell surface characteristic data to obtain surface characteristic data of each partition.

[0121] Among them, eggshell partitioning can be based on the physical and optical properties of the eggshell surface feature data, such as color, smoothness, crack distribution, thickness change, optical reflectivity, etc., to divide the eggshell of the egg to be analyzed into multiple areas, so that subsequent optical coherence tomography (OCT) can accurately detect different partitions.

[0122] Among them, the partition surface feature data can be the eggshell surface feature data corresponding to different areas extracted based on the eggshell partition, that is, the eggshell surface physical feature and optical property data corresponding to different eggshell partitions.

[0123] Specifically, based on surface features such as color uniformity, smoothness, reflectivity, crack distribution, local thickness variation, etc., image segmentation algorithms (such as K-Means clustering, region growing, or edge detection) are used to automatically identify different regions of the eggshell and divide them into multiple partitions with similar features. For example, the top, bottom, and sides of the eggshell may exhibit different thicknesses or optical properties due to different stress distributions, so they can be divided separately; at the same time, if the surface detection finds abnormalities (such as cracks or optical reflectivity abnormalities), the area can be subdivided into high-priority scanning areas for subsequent higher-precision OCT analysis. After the partitioning is completed, a unique regional identifier is assigned to each eggshell partition, and the corresponding partition surface feature data is stored, which includes the color distribution, smoothness, crack morphology, local thickness, and light reflection characteristics of the partition.

[0124] Step 904, for any eggshell partition, determine the coherent tomography dimension of the eggshell partition based on the surface feature data of the partition.

[0125] Among them, the coherent tomography dimension can be to determine the OCT imaging parameters suitable for the area according to the partition surface feature data of the eggshell before the OCT scan, including scanning depth, lateral resolution, scanning step length, light source power, signal acquisition rate, etc.

[0126] Specifically, for any selected eggshell partition, key parameters such as scanning depth, lateral resolution and scanning density are determined based on the surface feature data of the eggshell partition, such as color reflectivity, surface smoothness, crack conditions and local thickness. For example, for areas with thicker, smoother eggshells and no obvious cracks, a larger scanning step and a deeper OCT penetration depth can be used to improve scanning efficiency; for areas with cracks or abnormal surface reflections, a higher lateral resolution and a smaller step interval should be used to ensure that the morphology, depth and expansion trend of microcracks can be clearly captured. Combined with the optical properties of eggshell materials, such as the absorption and scattering characteristics of eggshells for light of specific wavelengths, the incident angle and light source power of OCT are optimized to avoid excessive scattering or signal attenuation. Finally, the coherent tomography scanning dimension corresponding to each eggshell partition is obtained to ensure that each eggshell partition can obtain the clearest interlayer structure information during OCT scanning.

[0127] Step 906, for any eggshell partition, scan the eggshell partition according to the coherence tomography dimension to obtain coherence tomography data.

[0128] Among them, the coherent tomography data can be the imaging data of the shallow structure inside the eggshell obtained based on the principle of low-coherence light interference during the OCT scanning process. This data contains information such as the distribution of eggshell microcracks, local thickness changes, mineral deposition, pore structure, and interlayer bonding strength, and can be used to detect internal defects of the eggshell that cannot be identified by visible light.

[0129] Specifically, for any selected eggshell partition, the scanning parameters, including light source wavelength, scanning depth, lateral resolution, scanning step length and signal acquisition rate, are adjusted based on the coherent tomography dimension to ensure the best imaging effect in different eggshell areas. Then the OCT system uses the principle of low-coherence light interference to emit a light beam to the eggshell surface, and generates a tomographic image of the eggshell in real time by detecting the returned interference signal. For different partitions, the OCT scanning method is different. For example, in areas with smooth surfaces and no cracks, a faster scanning rate is used to improve the detection efficiency; while in areas with cracks or abnormal thickness, high-density scanning is used to ensure that more subtle structural changes are captured. During the scanning process, the optical focus and incident angle are automatically adjusted to optimize the signal penetration depth and reduce image blur caused by light scattering; in addition, problems such as uneven eggshell thickness, surface reflection interference, and signal attenuation may be encountered during the scanning process. The light source power and data acquisition gain are dynamically adjusted to ensure that a clear tomographic signal is obtained. Finally, each eggshell partition will generate high-resolution OCT scanning data, which contains key features such as interlayer structure information, microcrack morphology, calcification degree, and pore distribution in the area.

[0130] Step 908, reconstruct each coherent tomography data to obtain eggshell shallow layer characteristic data.

[0131] Specifically, the coherent tomography data of each partition is preprocessed, including noise removal (such as adaptive filtering denoising), signal enhancement (such as histogram equalization), and contrast optimization; the coherent tomography data of each partition is spatially aligned using a spatial registration algorithm (such as feature point-based registration or phase correlation method) to ensure that the depth information of different scanning areas can be seamlessly spliced ​​to form a complete eggshell shallow structure model. Further, the scan slices of each partition are hierarchically spliced ​​using tomographic reconstruction technology (such as Fourier transform reconstruction or stacked imaging technology) to restore the three-dimensional shallow structure of the eggshell and ensure the continuity of eggshell microcracks, pores, interlayer thickness and other features. In order to improve the accuracy of the data, multi-scale fusion will also be performed to integrate the preprocessed coherent tomography data at the microscopic (nanoscale), local (millimeter) and overall (centimeter) levels, so that the shallow structure information of the eggshell can be effectively analyzed at different scales, and the final output of the eggshell shallow characteristic data includes key parameters such as the distribution of eggshell microcracks, local thickness differences, mineral deposition, and structural integrity.

[0132] In this embodiment, the eggshell is divided into regions through the eggshell surface feature data to ensure that different regions (such as the top, bottom, side or high-risk areas with cracks) can be targeted for depth detection, thereby improving detection efficiency and accuracy. Based on the partition surface feature data corresponding to each eggshell partition, the coherence tomography (OCT) dimension is set respectively, and key parameters such as scanning depth, lateral resolution, and step length are dynamically adjusted to adapt to the eggshell characteristics of different regions; OCT scanning can penetrate the eggshell surface to obtain shallow features such as microcracks, local thickness changes, mineral deposition status, and pore structure inside the eggshell, ensuring that not only surface defects visible to the naked eye can be detected, but also hidden quality risks can be identified. Through data reconstruction and multi-region fusion, complete eggshell shallow feature data is generated to ensure that the OCT data of each partition can be seamlessly integrated to form a high-precision three-dimensional shallow structure analysis result. It can improve the accuracy of eggshell quality assessment, realize automatic identification of abnormal eggshells, improve the intelligent level of egg product screening, and provide high-value data support for chicken health traceability.

[0133] Based on the same inventive concept, the embodiment of the present application also provides a chicken abnormality prediction device based on egg images for implementing the chicken abnormality prediction method based on egg images mentioned above. Fig.10 As shown, it includes: an eggshell data recognition module 1002, an abnormal eggshell selection module 1004, an abnormal eggshell analysis module 1006 and a chicken disease prediction module 1008. The implementation scheme for solving the problem provided by the device is similar to the implementation scheme recorded in the above method. Therefore, the specific limitations in one or more embodiments of the chicken abnormality prediction device based on egg images provided below can refer to the limitations of the chicken abnormality prediction method based on egg images above, which will not be repeated here. Each module in the above-mentioned chicken abnormality prediction device based on egg images can be implemented in whole or in part by software, hardware and their combination.

[0134] In an exemplary embodiment, a computer is provided, which may be a server, and its internal structure diagram may be as follows: Fig.11 The computer includes a processor, a memory, an input / output interface (I / O for short) and a communication interface. Those skilled in the art will understand that Fig.11 The structure shown in the figure is only a block diagram of a part of the structure related to the scheme of the present application, and does not constitute a limitation on the computer to which the scheme of the present application is applied. The specific computer may include more or fewer components than shown in the figure, or combine certain components, or have a different arrangement of components.

[0135] In one embodiment, a computer is further provided, including a memory and a processor, wherein a computer program is stored in the memory, and the processor implements the steps in the above method embodiments when executing the computer program.

[0136] In one embodiment, a computer-readable storage medium is provided, storing a computer program, which implements the steps in the above method embodiments when executed by a processor.

[0137] In one embodiment, a computer program product or computer program is provided, the computer program product or computer program includes computer instructions, the computer instructions are stored in a computer-readable storage medium. A computer processor reads the computer instructions from the computer-readable storage medium, and the processor executes the computer instructions, so that the computer performs the steps in the above-mentioned method embodiments.

[0138] The technical features of the above embodiments may be combined arbitrarily. To make the description concise, not all possible combinations of the technical features in the above embodiments are described. However, as long as there is no contradiction in the combination of these technical features, they should be considered to be within the scope of this specification.

Claims

1. A chicken abnormality prediction method based on egg images, characterized in that: The method comprises: Performing eggshell optical feature recognition on the multi-source eggshell collected data of the egg to be analyzed to obtain eggshell image feature data; Determine eggs with abnormal eggshells from the eggs to be analyzed according to the eggshell image feature data, and select abnormal eggshell feature data corresponding to the eggs with abnormal eggshells from the eggshell image feature data; According to the abnormal eggshell characteristic data, a disease dynamics simulation is performed on the suspicious chickens corresponding to the eggs with abnormal eggshells to obtain the chicken disease dynamics data corresponding to the suspicious chickens; According to the chicken disease dynamics data, disease risk prediction is performed on the suspicious chicken to obtain disease risk prediction data of the suspicious chicken.

2. The method according to claim 1, characterized in that According to the abnormal eggshell characteristic data, the disease dynamics simulation is performed on the suspicious chickens corresponding to the eggs with abnormal eggshells to obtain the chicken disease dynamics data corresponding to the suspicious chickens, including: According to the abnormal eggshell characteristic data, the disease transmission evolution analysis is performed on the suspicious chicken to obtain the macroscopic disease analysis data of the suspicious chicken; According to the abnormal eggshell characteristic data, a spatiotemporal analysis of disease transmission of the suspicious chickens is performed to obtain microscopic disease analysis data of the suspicious chickens; The chicken macroscopic disease analysis data and the chicken microscopic disease analysis data are coupled to obtain the chicken disease dynamics data.

3. The method according to claim 2, characterized in that The method of performing disease transmission evolution analysis on the suspicious chickens according to the abnormal eggshell characteristic data to obtain macroscopic disease analysis data of the suspicious chickens includes: According to the actual flock size and actual health status of the suspected chickens, the initial inventory of the disease transmission evolution model is set; Inputting the abnormal eggshell characteristic data into the disease transmission evolution model to obtain a calculation flow rate of the disease transmission evolution model; updating the initial inventory according to the calculated flow rate to obtain an updated inventory of the disease transmission evolution model; Using the updated inventory as the initial inventory, and using the abnormal eggshell feature data at the next moment as the abnormal eggshell feature data, returning to the step of inputting the abnormal eggshell feature data into the disease transmission evolution model to obtain the calculation flow rate of the disease transmission evolution model, until the actual number of cycles of the disease transmission evolution model triggers the preset number of cycles, and obtaining the iterative disease analysis data of the chickens; The chicken house environment information corresponding to the abnormal chicken is used to perform scene intervention correction on the chicken iterative disease analysis data to obtain the chicken macroscopic disease analysis data.

4. The method according to claim 2, characterized in that: The method of performing a spatiotemporal analysis of disease transmission on the suspicious chickens based on the abnormal eggshell characteristic data to obtain microscopic disease analysis data of the suspicious chickens includes: Constructing a spatiotemporal model of disease transmission according to the actual flock size and actual health status corresponding to the suspicious chickens; the spatiotemporal model of disease transmission includes an initial infection probability and an initial latent state; According to the abnormal eggshell characteristic data, the initial infection probability and the initial latent state are optimized to obtain a real-time infection probability and a real-time latent state; According to the disease transmission rules and proximity rules of the disease transmission spatiotemporal model, the interactive evolution of diseases of each of the suspicious chickens is analyzed to obtain chicken transmission disease analysis data; According to the chicken house environment information corresponding to the abnormal chickens, the chicken infectious disease analysis data is subjected to scene intervention correction to obtain the chicken microscopic disease analysis data.

5. The method according to claim 2, characterized in that: The coupling of the chicken macroscopic disease analysis data and the chicken microscopic disease analysis data to obtain the chicken disease dynamics data includes: Using the chicken macroscopic disease analysis data, optimizing the interactive evolution trend of the chicken microscopic disease analysis data to obtain optimized microscopic disease analysis data; Integrating the microscopic disease analysis data of the chickens to construct global microscopic disease analysis data corresponding to the suspicious chickens; Using the global microscopic disease analysis data, the chicken macroscopic disease analysis data is corrected to obtain corrected macroscopic disease analysis data; The optimized microscopic disease analysis data is nested into the modified macroscopic disease analysis data to obtain the chicken disease dynamics data.

6. The method according to claim 1, characterized in that The method of performing disease risk prediction on the suspicious chickens according to the chicken disease dynamics data to obtain disease risk prediction data of the suspicious chickens includes: According to the chicken disease dynamics data, a disease prediction period corresponding to the disease risk prediction is set; the disease prediction period includes a short-term disease prediction, a mid-term disease prediction and a long-term disease prediction; Using a chicken disease discrete dynamic algorithm, based on the chicken disease dynamics data, short-term prediction of the disease is performed on the suspected chicken to obtain short-term disease risk data; Using the chicken disease intervention evolutionary algorithm, based on the chicken disease dynamics data, mid-term prediction of the disease is performed on the suspected chicken to obtain mid-term disease risk data; Using a nonlinear algorithm for chicken diseases, based on the chicken disease dynamics data, long-term prediction of the disease is performed on the suspected chickens to obtain long-term disease risk data; The short-term disease risk data, the medium-term disease risk data and the long-term disease risk data are spliced ​​to obtain the disease risk prediction data.

7. The method according to any one of claims 1 to 6, characterized in that: The eggshell optical feature recognition is performed on the multi-source eggshell collected data of the egg to be analyzed to obtain eggshell image feature data, including: Performing eggshell three-dimensional visible light feature recognition on the multi-source eggshell collected data of the egg to be analyzed to obtain eggshell surface feature data; Based on the eggshell surface feature data, performing eggshell optical coherence tomography analysis on the multi-source eggshell collection data of the egg to be analyzed to obtain eggshell shallow layer feature data; The eggshell surface feature data and the eggshell shallow layer feature data are fused to determine the eggshell image feature data.

8. The method according to claim 7, characterized in that The eggshell optical coherence tomography analysis is performed on the multi-source eggshell data collected from the egg to be analyzed to obtain the eggshell shallow layer characteristic data, including: According to the eggshell surface characteristic data, the eggshell of the egg to be analyzed is partitioned to obtain surface characteristic data of each partition; any of the partition surface characteristic data has a corresponding eggshell partition; For any of the eggshell partitions, determining the coherence tomography dimension of the eggshell partition according to the surface feature data of the partition; For any of the eggshell partitions, scanning the eggshell partition according to the coherence tomography dimension to obtain coherence tomography data; Each of the coherent tomography data is reconstructed to obtain the shallow layer characteristic data of the eggshell.

9. A chicken abnormality prediction device based on egg images, characterized in that: The device comprises: An eggshell data recognition module is used to perform eggshell optical feature recognition on the multi-source eggshell collected data of the eggs to be analyzed to obtain eggshell image feature data; An abnormal eggshell selection module is used to determine eggs with abnormal eggshells from the eggs to be analyzed according to the eggshell image feature data, and to select abnormal eggshell feature data corresponding to the eggs with abnormal eggshells from the eggshell image feature data; An abnormal eggshell analysis module is used to perform disease dynamics simulation on suspicious chickens corresponding to the eggs with abnormal eggshells according to the abnormal eggshell characteristic data, so as to obtain disease dynamics data of the chickens corresponding to the suspicious chickens; The chicken disease prediction module is used to predict the disease risk of the suspicious chickens according to the chicken disease dynamics data to obtain the disease risk prediction data of the suspicious chickens.

10. A computer comprising a memory and a processor, wherein the memory stores a computer program, characterized in that: When the processor executes the computer program, the steps of the method according to any one of claims 1 to 8 are implemented.

Images