Acoustic-optical precision targeting and expelling method and system for detecting wild animals based on infrared

Abstract

The application discloses an acousto-optic precision targeting and expelling method and system based on infrared detection of wild animals, and relates to the technical field of image processing. The method comprises the following steps: acquiring a continuous image sequence captured by an infrared night vision camera arranged around crops; performing trajectory analysis on the image sequence to extract time sequence trajectory information thereof; calling a plurality of preset trajectory analysis modules and acquiring a response strategy sub-library associated with each trajectory analysis module; constructing an expelling scheduling mechanism according to the target response strategy; inputting the acquired time sequence parameters related to the approach of animals to crops into the expelling scheduling mechanism, performing an acousto-optic expelling operation, and outputting an expelling effect report. The application is based on infrared detection technology and realizes acousto-optic precision targeting and expelling, intelligently intervenes in the behavior of wild animals approaching crops, and significantly improves the accuracy, response speed and efficiency of the expelling process.

Description

TECHNICAL FIELD

[0001] The present application relates to the field of animal repelling technology, and in particular to an acousto-optic precision targeting repelling method and system based on infrared detection of wild animals. BACKGROUND

[0002] With the increasing demand for crop protection in agricultural production, traditional methods of wild animal repelling such as manual driving, hunting, and chemical repelling have been unable to meet the needs of modern agriculture. These traditional methods not only have low efficiency and are difficult to monitor the effect, but also may have negative impacts on the ecological environment. Therefore, animal repelling technology based on intelligent and precise means has become a problem to be solved in the field of agriculture.

[0003] In recent years, the progress of infrared detection technology and image processing technology has provided new opportunities for the precise identification and tracking of animal behavior, especially in complex environments. With the help of modern sensors and artificial intelligence technology, real-time monitoring, behavior analysis, and precise repelling of animals can be achieved. However, the existing technology still has some defects in the precise repelling of wild animals, such as how to use infrared detection technology to accurately identify animal behavior in real time and automatically generate a precise response strategy based on the timing information of the animal approaching the crops, and how to optimize the repelling effect in real time based on animal behavior. SUMMARY

[0004] This section aims to summarize some aspects of the embodiments of the present application and briefly introduce some preferred embodiments. Some simplifications or omissions may be made in this section and the abstract and title of the specification to avoid obscuring the purpose of this section, abstract and title, and such simplifications or omissions cannot be used to limit the scope of the present application.

[0005] In view of the problems existing in the prior art, the present application is proposed.

[0006] To solve the above technical problems, the present application provides the following technical solutions:

[0007] The acousto-optic precision targeting repelling method based on infrared detection of wild animals comprises:

[0008] Obtaining a sequence of continuous images captured by an infrared night vision camera arranged around the crops;

[0009] Performing trajectory analysis on the image sequence to extract timing trajectory information; the timing trajectory information includes a dynamic trajectory feature module for predicting animal behavior trends;

[0010] Calling a plurality of preset trajectory analysis modules and obtaining a response strategy sub-library associated with each trajectory analysis module;

[0011] For each trajectory parsing module, the target response strategy corresponding to the time-series trajectory information is searched in the corresponding response strategy sub-library in sequence, and a drive-off scheduling mechanism is constructed based on the target response strategy.

[0012] The acquired time-series parameters related to animal proximity to crops are input into the deportation scheduling mechanism to execute audio-visual deportation operations and output a deportation effect report.

[0013] As a preferred embodiment of the acoustic-optical precision-targeted repulsion method for wild animals based on infrared detection described in this invention, the step of sequentially searching for the target response strategy corresponding to the temporal trajectory information from the response strategy sub-library for each trajectory parsing module, and constructing a repulsion scheduling mechanism accordingly, specifically includes:

[0014] Obtain a preset set of deportation scheduling mechanisms, and search for a predetermined deportation mechanism in the set of deportation scheduling mechanisms based on the time-series parameters related to animal approach to crops;

[0015] If a pre-defined removal mechanism is successfully found, it will be designated as the current removal scheduling mechanism.

[0016] If the search for the established expulsion mechanism fails, for each trajectory parsing module, the target response strategy that matches the time-series trajectory information is searched from each response strategy sub-library in turn, and the target expulsion scheduling mechanism is constructed accordingly.

[0017] As a preferred embodiment of the acoustic-optical precision targeting and removal method for wildlife based on infrared detection described in this invention, wherein: when constructing a target removal scheduling mechanism according to a target response strategy, the target removal scheduling mechanism is dynamically added to the removal scheduling mechanism set, thereby generating an updated removal scheduling mechanism set.

[0018] As a preferred embodiment of the acoustic-optical precision-targeted repulsion method for wildlife based on infrared detection described in this invention, wherein: for each trajectory parsing module, sequentially searching for the target response strategy corresponding to the temporal trajectory information from each response strategy sub-library includes:

[0019] Acquire historical trajectory data and construct a strategy mapping model based on this historical trajectory data;

[0020] The time-series trajectory information is input into the strategy mapping model to determine the correspondence, and the target response strategy for each trajectory parsing module is output.

[0021] As a preferred embodiment of the acoustic-optical precision targeting and repelling method for wild animals based on infrared detection described in this invention, the step of sequentially searching for a target response strategy matching the temporal trajectory information from each response strategy sub-library for each trajectory parsing module includes:

[0022] According to the time sequence track information, an effective unit is identified from each track analysis module, and an effective response strategy sub-library corresponding to the effective unit in each response strategy sub-library is determined;

[0023] For each effective unit, a target response strategy matching the time sequence track information is retrieved from the effective response strategy sub-library in sequence.

[0024] As a preferred scheme of the sound-light precision targeting and driving away method based on infrared detection of wild animals, the method further comprises: monitoring the execution effect of the target driving away scheduling mechanism by using a preset process coordinator;

[0025] If it is monitored that the driving away scheduling has been executed, it is determined that the sound-light driving away process is completed, and the running resources occupied by each track analysis module are released; and / or, if a process termination instruction is received during the execution of the sound-light driving away process, a termination operation is performed on any track analysis module in the target driving away scheduling mechanism by using the process coordinator.

[0026] As a preferred scheme of the sound-light precision targeting and driving away method based on infrared detection of wild animals, the method further comprises: monitoring the execution effect of the target driving away scheduling mechanism by using a preset process coordinator;

[0027] Using the target driving away scheduling mechanism, a plurality of target response strategies are executed in parallel based on the time sequence parameters;

[0028] The driving away process records generated during the execution of each target response strategy are stored in a driving away log library;

[0029] The driving away process records in the driving away log library are called to generate a sound-light driving away effect report.

[0030] As a preferred scheme of the sound-light precision targeting and driving away method based on infrared detection of wild animals, the method further comprises: monitoring the execution effect of the target driving away scheduling mechanism by using a preset process coordinator;

[0031] According to the execution sequence information of each target response strategy, a primary response strategy in a first execution sequence level and a senior response strategy in a second execution sequence level are identified from the target response strategies;

[0032] The second execution sequence level is higher than the first execution sequence level;

[0033] The primary driving away process records generated during the execution of each primary response strategy are stored in a first driving away log sub-library;

[0034] When the high-level response strategy is running in parallel, a high-level driving process record is generated based on the primary driving process record in the first driving log sub-library called, and is stored in the second driving log sub-library.

[0035] The system applied to the above-mentioned sound-light precision targeted driving method based on infrared detection of wild animals comprises:

[0036] An infrared detection module, an infrared night-vision camera arranged around crops, is responsible for collecting a continuous image sequence of animals approaching crops.

[0037] A trajectory analysis module, which performs trajectory analysis of the image sequence, extracts time sequence trajectory information of the animals, and performs dynamic feature extraction thereon.

[0038] A response strategy sub-library module for storing response strategies related to different animal behaviors and environmental scenarios.

[0039] A driving scheduling mechanism module, which performs sound-light driving operation after obtaining time sequence parameters of animals approaching crops.

[0040] A log recording module, which stores driving process records generated in the execution of each target response strategy into a driving log library, and calls the driving process records in the library to generate a driving effect report.

[0041] A flow coordination and resource management module, which monitors the execution effect of the target driving scheduling mechanism in real time through a preset flow coordinator, judges whether the driving is completed, and releases the running resources occupied by each trajectory analysis module.

[0042] A strategy mapping and historical data analysis module, which is used for constructing a strategy mapping model according to historical trajectory data, inputting time sequence trajectory information into the model for corresponding relationship judgment, and outputting corresponding target response strategies for each trajectory analysis module.

[0043] The application further discloses a computer readable storage medium, which stores a computer program, and the computer program is executed by a processor to realize the steps of the sound-light precision targeted driving method and system based on infrared detection of wild animals.

[0044] The application has the following beneficial effects:

[0045] The application realizes real-time response and prediction of animals approaching crops through the combination of multiple trajectory analysis modules and dynamic time sequence parameters. Through multiple strategy iterations, the driving strategy is further optimized, the driving effect is improved, the driving scheduling mechanism is dynamically updated, the system can automatically adjust the response strategy according to different animal approaching conditions, and the strategy is monitored and adjusted in real time during the execution process, so that invalid resource consumption is avoided, and the resource utilization rate is improved.

[0046] In summary, the present application is based on infrared detection technology, through acousto-optic precise targeting and driving, intelligently intervening the behavior of wild animals approaching crops, significantly improving the accuracy, response speed and efficiency of the driving process. BRIEF DESCRIPTION OF DRAWINGS

[0047] In order to more clearly illustrate the technical solutions of the embodiments of the present application, the drawings needed to be used in the embodiment description will be briefly introduced. Obviously, the drawings in the following description are only some embodiments of the present application, and other drawings can be obtained by those skilled in the art without creative labor. Among them:

[0048] Fig. 1 The overall flowchart of the acousto-optic precise targeting and driving method based on infrared detection of wild animals proposed by the present application;

[0049] Fig. 2 The overall framework diagram of the acousto-optic precise targeting and driving system based on infrared detection of wild animals proposed by the present application. DETAILED DESCRIPTION

[0050] In order to make the above-mentioned purposes, features and advantages of the present application more obvious and easy to understand, the specific embodiments of the present application will be described in detail below with reference to the drawings of the specification.

[0051] In the following description, many specific details are set forth in order to provide a thorough understanding of the present application, but the present application can also be implemented in other ways different from those described herein, and those skilled in the art can make similar generalizations without departing from the connotation of the present application, therefore the present application is not limited to the specific embodiments disclosed below.

[0052] Secondly, the "one embodiment" or "embodiment" referred to herein means that the specific features, structures or characteristics can be included in at least one implementation of the present application. In this specification, "in one embodiment" does not mean the same embodiment, nor is it an independent or alternative embodiment that excludes other embodiments.

[0053] REFERENCE Figs. 1-2 For one embodiment of the present application, the acousto-optic precise targeting and driving method and system based on infrared detection of wild animals are provided, which includes the following steps:

[0054] Step 1: Obtain the continuous image sequence captured by the infrared night vision camera arranged around the crops;

[0055] Step two: Perform trajectory analysis on the image sequence to extract its time-series trajectory information; the time-series trajectory information includes dynamic trajectory feature modules used to predict animal behavior trends.

[0056] Specifically, to achieve accurate detection and repulsion of wild animals, the first step is to perform detailed analysis on the acquired image sequence. That is, through automated image processing algorithms, the animal's activity trajectory is extracted from the infrared image. The main steps of this process include:

[0057] Automated image processing algorithms are used to analyze the image sequence, and domain experts can also manually identify and confirm certain image features as needed. This flexibility reduces technical barriers, allowing the system to adapt to different application scenarios.

[0058] The analyzed image sequence is converted into time-series trajectory information, which represents the animal's movement trajectory along a specific path on the time axis. Each trajectory information contains the animal's motion characteristics, and different activity states are marked according to the timestamp.

[0059] Each animal trajectory is refined into a dynamic trajectory feature module, which includes the following attributes:

[0060] Function: This trajectory block describes the animal's motion state, such as moving along path X;

[0061] Input: Includes image data and related information such as timestamps;

[0062] Interface relationship: Share real-time data with other modules (such as threat assessment module, repulsion control module) to achieve real-time decision-making and control; through this process, vague requirements are converted into standardized, machine-understandable time-series trajectory information, providing a clear basis for subsequent strategy matching and repulsion execution.

[0063] For example, suppose the system detects a wild boar entering the farmland area, then the following operations are performed:

[0064] Animal detection: First, through infrared camera image processing, the wild boar image entering the farmland area is extracted and converted into analyzable trajectory data.

[0065] Trajectory analysis: According to the animal's motion trajectory, the generated time-series trajectory information includes the animal's path information from the boundary to the center of the farmland, continuous time markers (timestamps), motion speed, etc.

[0066] Threat assessment: Based on the data in the dynamic trajectory feature module, assess whether the animal's activity will cause damage, such as the distance and length of time the animal is close to the crop area.

[0067] Drive-off control: By analyzing the above information, the system will determine the appropriate drive-off strategy (such as playing predator sounds and starting light flickering).

[0068] Finally, the generated temporal trajectory information converts user requirements into specific and operable trajectory data, serving as input for subsequent threat assessment and drive-off control steps. Through this step, animal activities are converted into structured data, making subsequent strategy matching more efficient. The structured definition of temporal trajectory information allows the system to be flexibly configured for different animal species and scenarios, achieving precise drive-off.

[0069] Step three: Call multiple preset trajectory analysis modules and obtain the response strategy sub-library associated with each trajectory analysis module.

[0070] Specifically, the trajectory analysis module is a basic module in the system, responsible for analyzing and processing the temporal trajectory information extracted from the infrared image sequence. Each trajectory analysis module has independent functions and can run in parallel, improving the efficiency of the entire system. The essential properties of the trajectory analysis module are:

[0071] Functional independence: Each trajectory analysis module is responsible for a specific task, such as detecting animal movement trajectories or identifying activities, ensuring modularization and mutual non-interference.

[0072] Parallel operation: Each trajectory analysis module can execute simultaneously, providing parallel processing capabilities for large-scale, multi-target environments and significantly improving processing speed.

[0073] Collaboration: Although each trajectory analysis module executes tasks independently, they collaborate through strategic configuration (dynamic selection of behavior patterns) and asynchronous communication mechanisms (such as shared databases or message queues) to complete complex multi-task processing.

[0074] Covering the entire life cycle: The trajectory analysis module covers every stage of the monitoring and drive-off process, from animal detection to activity assessment and drive-off suggestion output.

[0075] For example, the system has multiple preset trajectory analysis modules. For example, the animal detection unit: responsible for identifying the location and movement trajectory of animals from images; the threat assessment unit: determines the potential threat to crops based on the animal's activity path and outputs the threat level; the behavior analysis unit: analyzes the animal's movement patterns to identify harmful behavior.

[0076] Each trajectory analysis module has a corresponding response strategy sub-library to address different scenarios. The response strategy sub-library contains all possible response methods for that unit, allowing it to select appropriate strategies based on different input data and contexts.

[0077] Definition of response strategy sub-library: the response strategy sub-library of each trajectory analysis module is a set of specific strategies that the unit can execute, which varies according to different simulation scenarios and requirements. Response strategies provide the system with the ability to respond flexibly to various situations, including flexibility: by setting different strategies, the system can adjust the reaction behavior according to different animal species, activity situations and crop protection needs, adaptability: the design of the strategy sub-set can be dynamically adjusted to adapt to various environments and scenarios.

[0078] For example, for the animal detection unit, possible response strategies include:

[0079] Threat response strategy: when the detected animal species is a high-risk species, activate the sound and light drive;

[0080] Guidance strategy: for some non-threatening animals, provide strategies to guide or change animal behavior, such as playing specific sounds to make them leave a certain area.

[0081] For example, the strategy sub-set of the threat assessment unit may include:

[0082] High-risk level strategy: when the animal approaches the crop area, activate the high-intensity drive mode;

[0083] Low-risk level strategy: if the animal is far away from the crop area, only use low-intensity interference sound or flashing light.

[0084] In summary, by designing a rich response strategy sub-library, the system can flexibly respond to changing environments and animal behavior, ensuring that each trajectory analysis module can make appropriate responses based on actual conditions, thereby optimizing the effectiveness of the entire drive process. Improve flexibility: the system can select the most appropriate strategy for processing according to different scenarios. Improve configurability: different environments and animal behavior can be configured to drive in different ways by selecting different strategies, making the system highly customizable.

[0085] Step four: for each trajectory analysis module, find the target response strategy corresponding to the time sequence trajectory information from the corresponding response strategy sub-library in order, and construct the drive scheduling mechanism according to the target response strategy.

[0086] Specifically, obtain a set of preset drive scheduling mechanisms, search for a given drive mechanism in the set of drive scheduling mechanisms based on the time sequence parameters related to the animal approaching the crop;

[0087] If the given drive mechanism is successfully searched, it is determined as the current drive scheduling mechanism;

[0088] Specifically, by calculating the similarity between the current timing parameters and the preset parameters of each mechanism (such as cosine similarity based on feature vectors, it should be noted that the essence of cosine similarity is to calculate the difference in direction between two vectors, and the influence of its absolute length is weakened, so it is particularly suitable for measuring the similarity of behavior patterns, and this calculation method is well known to those skilled in the art), if the similarity is higher than the preset threshold, it is determined that the search is successful.

[0089] If the search for the specified repelling mechanism fails, for each trajectory analysis module, the target response strategy matching the timing trajectory information is searched from each response strategy sub-library in turn, and the target repelling scheduling mechanism is constructed accordingly.

[0090] In the present application, the preset repelling scheduling mechanism set is constructed in advance according to historical data and known wild animal behavior patterns, as well as the contact situation with crops. Each mechanism in the set corresponds to different repelling scenarios and response strategy combinations. These mechanisms are created based on different animal species, crop species and environmental conditions by analyzing and summarizing effective repelling strategies in different scenarios. The source and composition of the repelling scheduling mechanism set are as follows:

[0091] Historical data and scenario analysis: The repelling scheduling mechanism set is not randomly set, but is constructed in advance by analyzing a large amount of historical data, including animal species, motion trajectory, threat assessment and crop damage situation, etc.

[0092] Mechanism composition: These mechanisms integrate trajectory analysis modules and response strategy sub-libraries to form a series of repelling scheduling strategies that can cope with different scenarios.

[0093] Priority use scheme: For known and common animal and crop contact scenarios, the preset repelling scheduling mechanism can provide efficient solutions, thereby improving the response efficiency of the system.

[0094] When the system receives a new repelling demand, it will first extract the key timing trajectory information by analyzing the demand. This information will be used to match with each mechanism in the preset repelling scheduling mechanism set: First, the demand needs to be analyzed: the system analyzes the new repelling demand and extracts the relevant timing trajectory information, including the type of animal, the time and location of approaching the crops, etc. Then match with the mechanism: match the extracted timing trajectory information with the mechanisms in the repelling scheduling mechanism set. The matching is mainly based on the similarity of the scene information and the applicable range of the mechanism, and the most suitable repelling strategy can be selected through similarity calculation. When matching, not only the timing data input is considered, but also the applicable scene and historical success rate of the mechanism are considered to ensure that the selected mechanism can be effectively executed under the current conditions. The specific method of "considering the applicable scene" is to perform the above feature vector similarity calculation and threshold comparison, and the consideration of "historical success rate" indicates that the preset repelling scheduling mechanism set construction method (historical data and scene analysis: the repelling scheduling mechanism set is not randomly set, but is pre-constructed by analyzing a large amount of historical data, including animal species, movement trajectory, threat assessment and crop damage, etc.), which ensures that the matching search operation is in a high-quality solution space itself. Therefore, the specific implementation of "considering the historical success rate when matching" is to ensure that the matching search operation is performed in a preset set consisting of high historical success rate mechanisms.

[0095] Example: First step: construct timing parameters into feature vectors, extract a set of key features from timing trajectory information to form a multi-dimensional vector. For example: the current scene vector V = [animal type code, normalized speed, normalized acceleration, distance from crop center, motion direction angle, ambient light intensity,...].

[0096] To prevent a dimension value from dominating the entire similarity calculation, normalize all features (scale to 0-1 range). For example, the animal type code can be pre-set (wild boar = 0.8, deer = 0.5, bird = 0.2).

[0097] Second step: define a reference vector for each preset mechanism, that is, each preset repelling scheduling mechanism is bound to a reference feature vector V_i defined by historical success data when creating the mechanism. For example, the reference vector of mechanism A is V_A.

[0098] Third step: calculate the cosine similarity and make a decision, calculate the cosine similarity between the current scene vector V and the reference vector V_i of each preset mechanism. The calculation formula is:

[0099] Similarity = (V · V_i) / (||V|| × ||V_i||).

[0100] where, · represents the dot product of vectors, ||V|| represents the modulus (length) of the vector. The closer the result of this calculation is to 1, the more consistent the directions of the two vectors are, that is, the more similar the scene is, that is, the more suitable it is.

[0101] That is, if the system can find a highly matched mechanism, it directly takes it as the established driving scheduling mechanism and executes the strategy in the mechanism. If no suitable mechanism is found, the system will enable the dynamic generation process, parse the trajectory module for each trajectory, retrieve the target response strategy matching the time sequence trajectory information from the corresponding response strategy sub-library, and generate a new driving scheduling mechanism by combining these strategies.

[0102] In some embodiments, the method further comprises:

[0103] In the case of constructing a target driving scheduling mechanism according to a target response strategy, the target driving scheduling mechanism is dynamically added to the driving scheduling mechanism set, thereby generating an updated driving scheduling mechanism set;

[0104] Wherein, the target driving scheduling mechanism is generated based on the time sequence parameters of the specific animal approaching the crops. In order to improve the response speed and flexibility of the system, when a new driving scheduling mechanism is generated, it will be dynamically updated to the driving scheduling mechanism set through the programming interface or data management module, thereby generating a new driving scheduling mechanism set. This process not only ensures that the system quickly adapts to new scenarios, but also continuously optimizes and enriches the mechanism library, improving the intelligent level of the system.

[0105] The trigger condition for dynamic updating can trigger updating in the following situations: no matching established mechanism, the system cannot find a mechanism that highly matches the current time sequence trajectory information in the existing driving scheduling mechanism set, triggering updating. For example: the approach angle or speed of the animal does not match the historical situation; the existing mechanism execution effect is insufficient, the process coordinator monitors that the driving effect is lower than the preset threshold (such as the driving success rate <80%, or the animal still stays in the dangerous area), triggering the generation of a new mechanism; new behavior patterns appear, when the trajectory analysis module finds that the trajectory features are significantly different from the historical trajectory library (the behavior pattern similarity is lower than the threshold), it is determined that it is a new behavior pattern, and a new mechanism needs to be constructed.

[0106] Specifically, the process coordinator monitors the driving effect: this is achieved through the driving log library and the preset evaluation method. The monitoring object: the process coordinator regularly queries the driving log library (see step five), and then quantifies the effect: if the animal leaves the dangerous area within T time after starting the driving, it is recorded as successful, otherwise it is recorded as failed. The driving success rate is the proportion of the number of successes in a period of time (such as the last 10 times), and the driving effect is judged according to the success rate.

[0107] Implementation means of triggering the generation of new mechanisms: Here the new mechanism is not created out of nothing, but the strategy mapping model already recorded is called. This model is a trained machine learning model (such as a neural network) that can output a recommended "target response strategy" combination according to the input "current time trajectory information" (i.e. new behavior pattern). Construction process: The system packages this strategy combination output by the model, which has been verified to be effective, with the current scene feature vector, in the pre-set format, through the dynamic programming interface of the data management module, and adds it to the evicting scheduling mechanism set, thus completing the construction and storage of the new mechanism.

[0108] Implementation means of determining new behavior patterns: achieved by calculating the "behavior pattern similarity" and comparing it with a new one, example: convert the behavior (such as vector [average speed, speed change rate, moving direction, approach angle, activity radius]) into a set of numbers (behavior feature vector), then calculate the similarity of these numbers through a standard mathematical formula (cosine similarity) (the calculation method is the same as the scene similarity calculation principle), and finally make a decision through a pre-set numerical threshold (threshold value).

[0109] To ensure the continuous expansion of the mechanism library while maintaining stability, the following steps need to be taken: Similarity determination and redundancy filtering, before generating a new evicting mechanism, calculate the similarity with the existing mechanism (such as based on trajectory feature vector + strategy combination vector), if the similarity > 90%, it is determined as a variant of the existing mechanism, do not store repeatedly, only update the execution parameters. Priority marking and conflict arbitration, set the priority for each mechanism, according to the animal threat level (high risk > low risk), execution resource occupation (low occupation > high occupation), effect historical data (high success rate > low success rate) and other weighted calculations, when multiple mechanisms meet the conditions at the same time, select the one with the highest priority to execute. Version management and expiration cleaning, when a new mechanism is added, give it a version number (such as "Mechanism A_v2"), if the old version is not called continuously for many times (such as N = 5 times), it is automatically archived or cleaned up, to ensure that the mechanism set does not expand and avoid reducing the execution efficiency. Conflict detection and hierarchical merging, if two mechanisms output conflicts under the same time trajectory conditions (such as one requires "high intensity sound evicting" and the other requires "low intensity light evicting"), the system performs hierarchical merging through the strategy merger. Preferentially execute the primary mechanism (low cost, low intensity); if the primary execution fails, add the senior mechanism (high intensity, multi-modal). Dynamic update record and traceability, every time a mechanism is added or modified, it is recorded in the evicting log library, and the trigger reason (unmatched / low effect / new pattern / environment change) is marked. Subsequent mechanism optimization can be traced back according to the log to avoid repeated development of the same mechanism.

[0110] The implementation means of dynamic updating is that dynamic updating is realized through a specific programming interface or data management module. This enables the system to automatically add new target drive-off scheduling mechanisms to the mechanism set whenever a new target drive-off scheduling mechanism is generated, without manual intervention. In this process, the data management module is responsible for ensuring the effectiveness and compatibility of the newly generated drive-off scheduling mechanism, ensuring that the update operation proceeds smoothly and does not disrupt the stability of the existing system.

[0111] Through dynamic updating, the system can quickly respond to new drive-off needs without the need for large-scale reconstruction or upgrading of the entire system. This not only improves flexibility but also shortens the time to respond to new scenarios. Whenever similar animal behavior or crop proximity situations are encountered, the system can directly reuse previously successful drive-off scheduling mechanisms, thereby avoiding the hassle of redesigning drive-off strategies each time and greatly improving work efficiency.

[0112] In some embodiments, for each trajectory analysis module, the target response strategy corresponding to the time sequence trajectory information is sequentially searched from each response strategy sub-library, including:

[0113] Obtain historical trajectory data and construct a strategy mapping model based on the historical trajectory data;

[0114] Input the time sequence trajectory information into the strategy mapping model for corresponding relationship determination, and output the target response strategy for each trajectory analysis module.

[0115] The strategy mapping model is constructed based on historical trajectory data and time sequence trajectory information, and is mainly used to analyze and predict the response strategy of animal behavior. The construction process of the model is as follows:

[0116] Extract historical trajectory data from the system database, including past animal activity data, environmental conditions, and drive-off strategy execution results, etc. Perform preprocessing operations such as cleaning, formatting, and normalization on these historical data to ensure data quality and provide accurate input for subsequent analysis.

[0117] Then a machine learning algorithm (such as neural networks, decision trees, etc.) is used to build a strategy mapping model. The goal of this model is to learn the correlation between animal behavior and the driving strategy in historical data. Specifically, the input of the strategy mapping model must be multi-dimensional data that reflects the trend of animal behavior and the characteristics of the scene, which includes: animal basic characteristic parameters (such as animal species; animal size or thermal imaging area; and animal quantity), trajectory time sequence parameters (such as animal movement trajectory coordinate sequence (including timestamp); moving speed, acceleration; movement direction and turning point), environment and scene parameters (such as current time; weather conditions (whether there is fog, rain); crop growth stage (seedling / mature period)), and the input parameters are normalized (such as 0~1 interval mapping) to obtain the input vector, represented as X={x1,x2,…,xn}, where each xi represents an input feature (such as animal type, speed, trajectory length, environmental parameters, etc.). The output vector Y={y1,y2,…,ym} where each yj represents a candidate driving strategy (such as sound driving, light driving, joint driving, etc.), and the strategy mapping model can be defined as: ;

[0118] where represents a nonlinear mapping function (such as neural networks, decision trees, etc.) with parameters Here, taking the neural network form as an example, for the neural network model:

[0119] Hidden layer output: ;

[0120] Output layer prediction strategy probability: ;

[0121] where: , input features; , are the weights and biases trained respectively, and in the above represents any intermediate layer, represents the last layer (output layer); is the activation function (such as ReLU); represents the activation function, which is used to convert real vectors to probability distributions, represents the probability value given to each strategy, and the final selection probability is the highest as the target response strategy.

[0122] Use historical data to train the model, divide the training set and test set, and optimize through the backpropagation algorithm to ensure that the model can accurately predict the response strategy under different situations. Through cross-validation, accuracy, recall rate and other indicators to evaluate the model, and further optimize the model structure according to the evaluation results.

[0123] In summary, the constructed strategy mapping model will be used for actual trajectory analysis and animal repelling decisions. The application process is as follows: new time sequence trajectory information (such as animal type, position, motion trajectory, etc.) is input into the trained strategy mapping model. The model processes and analyzes the time sequence trajectory information and judges its similarity to the existing patterns in the historical data. The model outputs the target response strategy according to the input time sequence trajectory information, that is, provides appropriate repelling strategies for each trajectory analysis module. These strategies may include playing predator sounds, starting light interference, etc., to drive animals away from crops. After using the strategy mapping model, the system can quickly and accurately select the best response strategy for new animal activity scenarios, avoiding the tedious process of manual selection and configuration, and greatly improving simulation efficiency. The model trained by historical data can learn the optimal strategy combination, improve the repelling effect, and make the system's response more accurate, avoiding unnecessary interference or over-repelling.

[0124] In some embodiments, for each trajectory analysis module, the target response strategy matching the time sequence trajectory information is sequentially searched from each response strategy sub-library, including:

[0125] According to the time sequence trajectory information, the effective units are identified from each trajectory analysis module, and the effective response strategy sub-libraries corresponding to the effective units in each response strategy sub-library are determined;

[0126] For each effective unit, the target response strategy matching the time sequence trajectory information is sequentially searched from the effective response strategy sub-library.

[0127] The effective unit refers to the part of the trajectory information related to the current scene and capable of providing useful data. Through analysis of the time sequence trajectory information, the system can accurately identify which trajectory data best matches the current simulation scenario, ensuring the efficiency and accuracy of subsequent strategies. For each identified effective unit, the system will filter out the effective strategy subset corresponding to the effective unit from the corresponding response strategy sub-library. The effective response strategy sub-library is a dynamically adjusted strategy set based on historical data and the current scene, ensuring that only relevant strategies are considered, thereby avoiding interference from irrelevant strategies. The system sequentially searches for the target response strategy matching the current time sequence trajectory information from each effective response strategy sub-library. This process performs accurate matching based on the specific characteristics of each effective unit to ensure that the selected strategy optimally addresses the current repelling scheduling needs and provides efficient repelling or intervention responses. Once the target response strategy is determined, the system applies these strategies to the corresponding trajectory analysis module to guide subsequent actions, thereby improving simulation efficiency and accuracy. This step ensures that the system can respond quickly and efficiently without wasting resources.

[0128] In some embodiments, the method further comprises: monitoring the execution effect of the target repelling scheduling mechanism by using the preset process coordinator;

[0129] If it is monitored that the repelling scheduling has been executed completely, it is determined that the acousto-optic repelling process is completed, and the running resources occupied by each trajectory analysis module are released; and / or, during the execution of the acousto-optic repelling process, if a process termination instruction is received, a termination operation is performed on any trajectory analysis module in the target repelling scheduling mechanism by the process coordinator.

[0130] Specifically, the process coordinator serves as the core management module of the system, and is responsible for real-time monitoring of the execution state of each trajectory analysis module. When all relevant modules complete the scheduled tasks and output results, the process coordinator determines whether the tasks have been executed completely by checking the state flags or output data of each module. Once it is confirmed that the tasks are completed, the process coordinator triggers the resource release mechanism. This mechanism releases the occupied resources by calling the API of the operating system, including: closing the processes that are no longer needed; releasing the memory; disconnecting the network connections that are no longer used; clearing the cache data, etc. This process ensures the efficient use of system resources, avoids the waste of resources after task execution, and thus improves the performance and response speed of the overall system.

[0131] During the execution of the acousto-optic repelling process, if the system receives a process termination instruction from a user or other system, the process coordinator will immediately perform a termination operation on any trajectory analysis module in the target repelling scheduling mechanism. The termination operation includes the following steps: stopping the calculation of the current module; saving the current state (if possible) for subsequent recovery; notifying the relevant nodes or systems to ensure that other modules can respond in time and make synchronous adjustments.

[0132] Step five: input the acquired time sequence parameters related to the animal approaching the crops into the repelling scheduling mechanism, execute the acousto-optic repelling operation, and output the repelling effect report.

[0133] Specifically, using the target repelling scheduling mechanism, multiple target response strategies are executed in parallel based on the time sequence parameters; the repelling process records generated during the running of each target response strategy are stored in the repelling log library; each repelling process record in the repelling log library is called to generate an acousto-optic repelling effect report.

[0134] Specifically, the target repulsion scheduling mechanism uses parallel computing technology to improve efficiency in the process of executing multiple target response strategies. The specific steps are as follows: through multi-threading, multi-processing or distributed computing and other technical means, the parallel execution of multiple target response strategies is realized. This parallel computing method allows different response strategies to be processed simultaneously, shortening the overall execution time and improving the efficiency of the system. Each target response strategy is executed independently, and data exchange and synchronization are performed with other strategies through a simulation database. In order to ensure the efficiency of the system, message queues, shared memory and other inter-process communication (IPC) technologies can also be used for data sharing and coordination. During the execution of the strategy, the generated repulsion process data will be stored in the repulsion log library in real time or periodically. This storage method ensures the persistence and traceability of the data, so as to facilitate subsequent analysis and report generation.

[0135] During the execution of the target response strategy, the generated repulsion process data will be stored in the repulsion log library and processed and analyzed through pre-set algorithms and models. These data are stored and processed centrally, ensuring the consistency and integrity of the information and avoiding data loss and inconsistency caused by decentralized storage. The storage of all repulsion process data is realized through a database, which not only provides convenient access but also effectively manages the data. The storage format, structure and access permissions of the data can be standardized in the database.

[0136] It should be noted that, according to the execution sequence information of each target response strategy, the primary response strategy in the first execution order level and the advanced response strategy in the second execution order level are identified; the second execution order level is higher than the first execution order level;

[0137] The primary repulsion process records generated by running each primary response strategy are stored in the first repulsion log sub-library;

[0138] When running the advanced response strategy in parallel, the advanced repulsion process records are generated based on the primary repulsion process records in the first repulsion log sub-library and stored in the second repulsion log sub-library.

[0139] Specifically: according to the execution sequence information of the target response strategy, the system first identifies the primary response strategy in the first execution order level. These strategies are usually the first to be executed and are responsible for handling basic or pre-emptive repulsion tasks. On this basis, the system further identifies the advanced response strategy in the second execution order level, which will be executed in parallel after the primary response strategy is executed to optimize the repulsion effect. The second execution order level is higher than the first execution order level, ensuring that the execution time and effect of the advanced strategy are maximized.

[0140] The embodiment also discloses a judgment of a primary / advanced response strategy and a switching method thereof. The purpose of the method is mainly to ensure efficiency and accuracy in the process of driving away, so that the system can first use basic and low-cost means to try, and if the effect is not enough, rely on higher intensity and more complex strategies for supplementation. The judgment method can be realized through the following sub-steps:

[0141] S1: After trajectory analysis, the system evaluates the threat level of the animal's behavior of approaching the crops. Examples: low threat: the animal is far away from the crops (such as >10 meters), the speed is slow or the stay time is short; medium threat: the animal has entered the buffer area (such as 5-10 meters), and shows a trend of continuous stay or approach; high threat: the animal has entered the core protection area (such as <5 meters), and the trajectory shows obvious foraging or destruction behavior. The primary strategy is preferentially used for low to medium threat, and the advanced strategy is triggered for high threat.

[0142] S2: Based on the division of execution complexity, the primary response strategy is defined as: low system resource consumption, small action amplitude, and fast execution of the driving means. Examples: single sound source (such as playing the driving sound), low-frequency flashing light, single-point light irradiation. The advanced response strategy is defined as: requiring higher intensity or combined driving means, involving multi-modal parallel execution. Examples: multi-sound source stereo sound playing + high-frequency flashing light + mobile light source linkage. When the primary strategy is executed and the animal is still detected, the advanced strategy is automatically upgraded.

[0143] Step S3: Dynamic switching based on time / feedback, primary strategy time threshold: the system sets an observation time (such as 10 seconds) when executing the primary strategy. If the animal trajectory deviates from the crop area within this time, the primary strategy is considered effective, and the process ends; if there is no effect after this time, the primary strategy is considered ineffective. Upgrade mechanism: primary strategy failure→automatic switching to advanced strategy. The advanced strategy calls the process record of the primary strategy as an input parameter (such as sound source type, light source intensity), and adds stronger strategies on this basis to avoid repeated ineffective operations.

[0144] The driving process record generated by the primary response strategy in the execution process will be stored in the first driving log sub-database. This sub-database is specially used to store driving data related to the primary response strategy, ensuring complete recording of the preliminary driving process for subsequent analysis and backtracking. When the system executes the advanced response strategy in parallel, the advanced driving process record is generated based on the primary driving process record stored in the first driving log sub-database. These records will be stored in the second driving log sub-database.

[0145] In this way, the system can rely on the output data of the primary strategy to generate the driving process record of the advanced strategy when executing the strategies in parallel, realizing data correlation and effective utilization between strategies.

[0146] The embodiment also provides an acousto-optic precision targeting and driving away system based on infrared detection of wild animals, which is applied to the above-mentioned acousto-optic precision targeting and driving away method based on infrared detection of wild animals. The system comprises: an infrared detection module, an infrared night-vision camera arranged around crops, responsible for collecting a continuous image sequence of animals approaching crops; a trajectory analysis module, which performs trajectory analysis of the image sequence, extracts time sequence trajectory information of the animals, and performs dynamic feature extraction thereon; a response strategy sub-library module, which is used for storing response strategies related to different animal behaviors and environmental scenarios; a driving away scheduling mechanism module, which performs acousto-optic driving away operation after obtaining time sequence parameters of animals approaching crops; a log recording module, which stores driving away process records generated in the execution process of each target response strategy into a driving away log library, and calls the driving away process records in the library to generate a driving away effect report; a process coordination and resource management module, which monitors the execution effect of the target driving away scheduling mechanism in real time through a preset process coordinator, judges whether the driving away is completed, and releases the running resources occupied by each trajectory analysis module; a strategy mapping and historical data analysis module, which is used for constructing a strategy mapping model according to historical trajectory data, inputting the time sequence trajectory information into the model for corresponding relationship judgment, and thus outputting corresponding target response strategies for each trajectory analysis module.

[0147] The embodiment also provides a storage medium having a computer program stored thereon, the program being executed by a processor to implement the acousto-optic precision targeting and driving away method based on infrared detection of wild animals proposed in the above-mentioned embodiment. The storage medium can be implemented by any type of volatile or non-volatile storage device or a combination thereof, such as a static random access memory (SRAM), an electrically erasable programmable read-only memory (EEPROM), an erasable programmable read-only memory (EPROM), a programmable read-only memory (PROM), a read-only memory (ROM), a magnetic memory, a flash memory, a magnetic disk or an optical disk.

[0148] It should be noted that the above examples are only used to illustrate the technical solutions of the present application but not limit the present application. Although the present application is described in detail with reference to the preferred embodiments, those skilled in the art should understand that the technical solutions of the present application can be modified or equivalently replaced, without departing from the spirit and scope of the technical solutions of the present application, which should be covered in the scope of the claims of the present application.

Claims

1. A precise acoustic-optical targeting method for repelling wild animals based on infrared detection, characterized in that, The method includes: Acquire a continuous sequence of images captured by infrared night vision cameras deployed around the crops; The image sequence is subjected to trajectory parsing analysis to extract its temporal trajectory information; the temporal trajectory information includes a dynamic trajectory feature module for predicting animal behavior trends; Invoke multiple preset trajectory parsing modules and obtain the response strategy sub-library associated with each trajectory parsing module; For each trajectory parsing module, the target response strategy corresponding to the time-series trajectory information is searched in the corresponding response strategy sub-library in sequence, and a drive-off scheduling mechanism is constructed based on the target response strategy. The acquired time-series parameters related to animal approach to crops are input into the deportation scheduling mechanism to execute the audio-visual deportation operation and output a deportation effect report. Specifically, for each trajectory parsing module, the process of sequentially searching for a target response strategy corresponding to the time-series trajectory information from the corresponding response strategy sub-library, and constructing a drive-off scheduling mechanism based on the target response strategy, includes: Obtain a preset set of deportation scheduling mechanisms, and search for a predetermined deportation mechanism in the set of deportation scheduling mechanisms based on the time-series parameters related to animal approach to crops; If a pre-defined removal mechanism is successfully found, it will be designated as the current removal scheduling mechanism. If the search for the established expulsion mechanism fails, for each trajectory parsing module, the target response strategy that matches the time-series trajectory information is searched from each response strategy sub-library in turn, and the target expulsion scheduling mechanism is constructed accordingly. When constructing a drive-off scheduling mechanism based on a target response strategy, the target drive-off scheduling mechanism is dynamically added to the drive-off scheduling mechanism set, thereby generating an updated drive-off scheduling mechanism set.

2. The method for precise acoustic and optical targeting and repelling of wild animals based on infrared detection according to claim 1, characterized in that: The step of sequentially searching for the target response strategy corresponding to the time-series trajectory information from the corresponding response strategy sub-library for each trajectory parsing module includes: Acquire historical trajectory data and construct a strategy mapping model based on this historical trajectory data; The time-series trajectory information is input into the strategy mapping model to determine the correspondence, and the target response strategy for each trajectory parsing module is output.

3. The method for precise acoustic and optical targeting and repelling of wild animals based on infrared detection according to claim 2, characterized in that: The step of sequentially searching for the target response strategy corresponding to the time-series trajectory information from the corresponding response strategy sub-library for each trajectory parsing module includes: Based on the time-series trajectory information, valid units are identified from each trajectory parsing module, and valid response strategy sub-libraries corresponding to the valid units are determined in each response strategy sub-library. For each valid unit, the target response strategy that matches the time-series trajectory information is retrieved sequentially from the valid response strategy sub-library.

4. The method for precise acoustic and optical targeting and repelling of wild animals based on infrared detection according to claim 3, characterized in that: The method also includes: using a preset process coordinator to monitor the execution effect of the expulsion scheduling mechanism; If the expulsion scheduling is detected to have been completed, the audible and visual expulsion process is determined to be finished, and the operating resources occupied by each trajectory analysis module are released; and / or, if a process termination instruction is received during the execution of the audible and visual expulsion process, the process coordinator will terminate any trajectory analysis module in the expulsion scheduling mechanism.

5. The method for precise acoustic and optical targeting and repelling of wild animals based on infrared detection according to claim 4, characterized in that: The process involves inputting the acquired time-series parameters related to animal proximity to crops into the deportation scheduling mechanism, executing audio-visual deportation operations, and outputting a deportation effect report, including: A drive-off scheduling mechanism is used to execute multiple target response strategies in parallel based on time-series parameters; The expulsion process records generated during the operation of each target response strategy are stored in the expulsion log library. The expulsion process records in the expulsion log library are called to generate an expulsion effect report.

6. The method for precise acoustic and optical targeting and repelling of wild animals based on infrared detection according to claim 5, characterized in that: The process involves calling the eviction log database to record each eviction process and generating an eviction effectiveness report, including: Based on the execution order information of each target response strategy, the primary response strategy in the first execution order level and the advanced response strategy in the second execution order level are identified. The second execution order level is higher than the first execution order level; The initial expulsion process records generated by running each initial response strategy are stored in the first expulsion log sub-database; When running the advanced response strategy in parallel, an advanced removal process record is generated based on the primary removal process record in the first removal log sub-library, and stored in the second removal log sub-library.

7. A precise acoustic-optical targeting and deterrence system for wildlife based on infrared detection, applied to the precise acoustic-optical targeting and deterrence method for wildlife based on infrared detection described in claim 6, the system comprising: The infrared detection module, consisting of infrared night vision cameras deployed around the crops, is responsible for collecting a continuous sequence of images of animals approaching the crops. The trajectory parsing module performs trajectory parsing and analysis on the image sequence, extracts the temporal trajectory information of the animal, and extracts its dynamic features. The response strategy sub-library module is used to store response strategies related to different animal behaviors and environmental scenarios; After acquiring the timing parameters of the animal approaching the crop, the drive-away scheduling mechanism module executes an audio-visual drive-away operation. The logging module stores the expulsion process records generated during the execution of each target response strategy into the expulsion log library, and calls the expulsion process records in the library to generate an expulsion effect report; The process coordination and resource management module monitors the execution effect of the expulsion scheduling mechanism in real time through a preset process coordinator, determines whether the expulsion is completed, and releases the running resources occupied by each trajectory analysis module. The strategy mapping and historical data analysis module is used to construct a strategy mapping model based on historical trajectory data, and input time-series trajectory information into the model to determine the correspondence, thereby outputting the corresponding target response strategy for each trajectory parsing module.

8. A computer-readable storage medium having a computer program stored thereon, characterized in that: When the computer program is executed by the processor, it implements the steps of the acoustic-optical precision targeting and removal method for wild animals based on infrared detection, as described in any one of claims 1 to 6.

Images