Forestry pest intelligent prevention and early warning method and system based on image recognition

Abstract

The invention discloses a forestry pest intelligent prevention and early warning method and system based on image recognition, and the method comprises the steps: firstly obtaining a multi-dimensional image comprising different spectral frequency bands and spatial resolutions, carrying out the processing of a texture optimization model of multi-dimensional image fusion, and deeply mining the texture features of pests to form a fusion image; then, a forestry pest population recognition algorithm is combined with various parameters such as body types and antennas to judge population types; establishing a spatial distribution model based on categories and related parameters, and predicting a development trend in combination with a growth cycle and environmental adaptability parameters; and finally, generating prevention and early warning information by integrating the category, distribution and prediction data. The system comprises six units including a multi-dimensional image acquisition unit and an image fusion processing unit, all the units cooperate with one another, the whole process processing from image acquisition to early warning information generation is achieved, and an efficient and accurate technical scheme is provided for forestry pest prevention and control.

Description

Technical Field

[0001] The present invention relates to the field of forestry pest control, and in particular to an image recognition-based intelligent forestry pest control early warning method and system. Background Art

[0002] With the increasing need to protect forestry resources, forest pest control has become a key link in maintaining ecological balance and ensuring the sustainable development of the forestry economy. Traditional forest pest monitoring relies heavily on manual inspections, where forestry workers use on-site observations, sampling, and identification to determine the type and severity of the pests. This method is not only inefficient and labor-intensive, but also limited by the workers' experience and professional level, making it prone to missed detections and misjudgments. In addition, traditional monitoring methods make it difficult to achieve real-time, dynamic monitoring of large forest areas, and are unable to timely grasp the spatial distribution and development trends of pests, making it difficult to meet the needs of refined modern forestry management.

[0003] In recent years, although image recognition technology has been gradually applied to the field of forest pest monitoring, existing methods still have many shortcomings. On the one hand, in image acquisition and processing, single-band or low-resolution images are difficult to fully present the morphological characteristics and texture details of pests, resulting in a significant reduction in the accuracy of subsequent identification; and existing image fusion technology lacks targeted optimization of the texture characteristics of forest pests, and cannot effectively mine key information about their external morphology. On the other hand, in the pest identification and early warning links, traditional recognition algorithms often ignore parameters such as the growth cycle and environmental adaptability of forest pests, and cannot accurately determine the population category. It is also difficult to make scientific predictions on the spatial distribution and future development trends of pests, resulting in a lack of timeliness and effectiveness in prevention and control warnings, making it difficult to truly play a role in preventing and controlling pest damage. Summary of the Invention

[0004] In order to overcome the shortcomings and deficiencies of the existing technology, the present invention provides an intelligent forest pest prevention and control early warning method and system based on image recognition.

[0005] The technical solution adopted by the present invention is an intelligent forest pest prevention and early warning method based on image recognition, which includes the following steps:

[0006] Step S1: Acquire a multidimensional image containing potential harmful organisms in a forestry area, wherein the multidimensional image includes image data of different spectral bands and spatial resolutions;

[0007] Step S2: Processing the acquired multidimensional images through a texture optimization model of multidimensional image fusion to construct a fused image with enhanced texture representation characteristics, and deeply mining the external morphological texture features of forest pests;

[0008] Step S3: Inputting the fused image into a forestry pest population identification algorithm, determining the population category of the pests based on different parameters of the forestry pests, and obtaining pest population category information;

[0009] Step S4: establishing a pest population spatial distribution model based on the pest population category information and different parameters of forest pests, and determining the spatial distribution range of the pests in the forestry area;

[0010] Step S5: combining the growth cycle parameters and environmental adaptability parameters of forest pests to predict the future development trend of pest populations and obtain pest population development prediction data;

[0011] Step S6: Generate forestry pest intelligent prevention and control early warning information based on pest population category information, spatial distribution range and development forecast data.

[0012] Furthermore, the texture optimization model of the multi-dimensional image fusion is specifically:

[0013] F fused (x,y)=α·T enhanced1 (x,y)+β·T enhanced2 (x,y)

[0014] Among them, F fused (x,y) represents the pixel value of the fused image at the coordinate (x,y); T enhanced1 (x, y) is the pixel value of the first frequency band image at the coordinate (x, y) after enhancement based on the surface texture roughness parameters of forestry pests; T enhanced2 (x, y) is the pixel value of the second frequency band image at the coordinate (x, y) after enhancement based on the pest texture color contrast parameter; α and β are weight coefficients, and satisfy α+β=1. Their values ​​are dynamically adjusted according to the typical texture feature parameters of different types of forestry pests.

[0015] Furthermore, the forest pest population identification algorithm includes the following models:

[0016]

[0017] Among them, P class It represents the probability that the pest belongs to a certain population category; n is the number of characteristic parameters of forest pests used for identification; S featurei is the i-th characteristic parameter, including the pest body size parameters, antennae morphology parameters, and the similarity score with the corresponding population standard characteristic parameters; w i is the weight of the i-th feature parameter, which is determined by machine learning algorithm training based on historical forest pest identification data according to the importance of the feature in distinguishing different populations.

[0018] Furthermore, the process of establishing the pest population spatial distribution model is further based on the following model:

[0019] D distribution (x,y)=f(P detection (x,y),S density (x,y))

[0020] Among them, D distribution (x,y) represents the spatial distribution of pests at the coordinate (x,y); P detection (x, y) is the probability of detecting pests at the coordinate (x, y), which is determined by the pest characteristic parameters of the corresponding area in the fused image and the population recognition algorithm; S density (x,y) is the population density at coordinate (x,y) calculated based on the population quantity parameters and activity range parameters of forest pests; f is the spatial distribution mapping function, and its parameters are adaptively adjusted according to the topographic parameters and vegetation cover parameters of the forestry area.

[0021] Furthermore, the process of predicting the future development trend of the pest population utilizes the following model:

[0022]

[0023] Among them, G future G represents the predicted value of forest pest population size at a certain moment in the future; current is the current pest population size, which is determined by the population identification algorithm and spatial distribution model; r is the growth coefficient calculated based on the pest growth rate parameters and reproduction cycle parameters; E suitability E is the suitability parameter of the current forestry environment for pests, which is a comprehensive combination of temperature, humidity, and light environment parameters; standard It is the standard environmental suitability parameter for the growth and reproduction of pests.

[0024] Furthermore, the process of generating forestry pest intelligent prevention and control early warning information adopts the following model:

[0025] W information =I category ⊕I distribution ⊕I prediction

[0026] Among them, W information Indicates the generated intelligent prevention and control warning information; I category It is the information code of pest population category, which is converted from the result of population identification algorithm; I distribution Encode the spatial distribution information of pests and generate output data based on the spatial distribution model; Iprediction It is the encoding of pest development prediction information, which is determined according to population development prediction data; ⊕ is the information fusion operator, which integrates different information according to the preset coding rules.

[0027] Furthermore, the step S3 specifically includes:

[0028] Step S31: extracting morphological characteristic parameters of the pests in the fused image, including but not limited to body outline parameters, color texture parameters, and structure ratio parameters, to provide basic data for population identification;

[0029] Step S32: matching the extracted morphological characteristic parameters with a preset database of characteristic parameters of various forestry pest populations, and calculating a similarity index of each characteristic parameter;

[0030] Step S33: Based on the similarity index, a forestry pest population identification algorithm is used to perform preliminary population classification screening of the pests;

[0031] Step S34: Through the secondary feature verification mechanism, combined with the pest behavior habit parameters, the preliminary screening results are corrected to determine the final pest population category information.

[0032] Furthermore, the step S4 specifically includes:

[0033] Step S41: Divide the forest area into grids, and count the number parameters of the pests in each grid based on the pest population category information;

[0034] Step S42: determining the influence range of each individual forest pest within the grid based on the activity radius parameter of the forest pest;

[0035] Step S43: Calculate the probability parameters of pest spread between adjacent grids using a spatial distribution model based on the number of pests and their impact range in each grid;

[0036] Step S44: Based on the diffusion probability parameters and the distribution status of the pests in each grid, a complete spatial distribution model of the pest population is constructed to determine the spatial distribution range.

[0037] Furthermore, the step S5 specifically includes:

[0038] Step S51: obtaining the growth cycle stage parameters of the current forestry pests, and analyzing their growth characteristics and reproduction patterns in this stage;

[0039] Step S52: collecting real-time environmental parameters within the forestry area, and combining them with the environmental adaptability parameters of the pests to assess the impact of the current environment on the growth and reproduction of the pests;

[0040] Step S53: Based on the pest population parameters and growth and reproduction patterns, using the population development prediction model, calculate the predicted population growth value in the short term in the future;

[0041] Step S54: The short-term prediction value is corrected in combination with the environmental change trend parameter to obtain the future longer-term development prediction data of the pest population.

[0042] An intelligent forest pest prevention and early warning system based on image recognition, which includes:

[0043] A multi-dimensional image acquisition unit for acquiring multi-dimensional images containing potential harmful organisms in forestry areas;

[0044] An image fusion processing unit is connected to the multi-dimensional image acquisition unit and is used to process the acquired multi-dimensional image through a texture optimization model of multi-dimensional image fusion to construct a fused image;

[0045] The population identification processing unit is connected to the image fusion processing unit and is used to input the fused image into the forestry pest population identification algorithm to determine the population category of the pests;

[0046] A spatial distribution modeling unit, connected to the population identification processing unit, is used to establish a pest population spatial distribution model based on pest population category information and different parameters of forest pests;

[0047] Develop a prediction unit, connected to the spatial distribution modeling unit, to combine the growth cycle parameters and environmental adaptability parameters of forest pests to predict the future development trend of pest populations;

[0048] The early warning information generation unit is connected to the development forecast unit and is used to generate early warning information for intelligent prevention and control of forestry pests based on the pest population category information, spatial distribution range and development forecast data.

[0049] Beneficial effects: The present invention proposes an intelligent forest pest prevention and early warning method and system based on image recognition. The present invention acquires multidimensional images including different spectral frequency bands and spatial resolutions, and uses a texture optimization model to enhance and fuse the images based on parameters such as the roughness of the pest surface texture and color contrast, and deeply explores the external morphological texture characteristics of the pests. Compared with the traditional single image acquisition and processing method, it can provide richer and more accurate information for subsequent identification. In the pest identification and early warning link, the system combines various parameters such as the size, growth cycle, and environmental adaptability of forest pests, and uses population identification algorithms, spatial distribution models, and development prediction models to accurately determine the population category, determine the spatial distribution range, and predict future development trends. Compared with the problems of traditional algorithms that ignore key parameters and lack timeliness and effectiveness in early warning, this system builds a scientific model system through comprehensive multi-parameter analysis to achieve accurate control of the entire process of pest identification from category identification to development trend prediction. It can timely generate intelligent prevention and control early warning information containing population category, spatial distribution and development forecast data, provide efficient and accurate decision-making basis for forest pest control, significantly improve the efficiency and accuracy of forest pest control work, and effectively meet the needs of refined management of modern forestry. BRIEF DESCRIPTION OF THE DRAWINGS

[0050] Figure 1 is a flow chart of the method steps of the present invention;

[0051] Figure 2 It is a diagram of the system unit composition of the present invention. DETAILED DESCRIPTION

[0052] It should be noted that, unless there is a conflict, the embodiments and features in the embodiments of this application can be combined with each other. The present application will be further described in detail below with reference to the accompanying drawings and specific embodiments.

[0053] like Figure 1 As shown, the forestry pest intelligent prevention and early warning method based on image recognition includes the following steps:

[0054] Step S1: Acquire a multidimensional image containing potential harmful organisms in a forestry area, wherein the multidimensional image includes image data of different spectral bands and spatial resolutions;

[0055] Specifically, this step is the starting point of the entire intelligent prevention and early warning method, and its significance lies in providing a comprehensive and accurate raw data basis for subsequent analysis and processing. Image data in different spectral bands can show the characteristic differences between pests and forestry vegetation under different spectra. For example, some pests will show different reflectance characteristics from healthy vegetation in the near-infrared spectrum band. By acquiring images in this band, signs of pests that are difficult to detect with the naked eye can be effectively identified. Image data with different spatial resolutions can meet the needs of observation from macro to micro. High spatial resolution images can clearly capture the individual morphological details of pests, while low spatial resolution images can help to grasp the overall distribution of pests in large forestry areas.

[0056] During implementation, drones equipped with multiple spectral sensors and cameras of varying resolutions can be used to cruise and photograph forestry areas. Drones can capture all-around images of forestry areas at varying altitudes and angles along pre-set routes, ensuring that the multidimensional images captured cover the entire target area without missing any data. Furthermore, fixed image acquisition devices equipped with adjustable spectral bands and resolutions can be strategically deployed within forestry areas. These devices regularly capture images of the surrounding area and transmit the collected multidimensional image data in real time to a data processing center, providing raw image data for subsequent steps.

[0057] Step S2: Processing the acquired multidimensional images through a texture optimization model of multidimensional image fusion to construct a fused image with enhanced texture representation characteristics, and deeply mining the external morphological texture features of forest pests;

[0058] Specifically, this step improves image quality and feature expression capabilities, solving the problem that a single image cannot fully display the texture features of pests. The external morphological texture features of forest pests are an important basis for identifying their types and judging the degree of damage. However, images of different spectral bands and resolutions each contain only partial feature information. Through the texture optimization model of multi-dimensional image fusion, these scattered feature information are integrated and enhanced, so that the fused image can more clearly and completely present the texture details of pests, such as the lines on insect wings, the concave and convex structure on the surface of insect eggs, etc., providing richer and more recognizable image data for the subsequent accurate identification of pest population categories.

[0059] During its implementation, the model first pre-processes the acquired multi-dimensional images, converting image data in different formats and coordinate systems into a standard format and coordinate system for subsequent fusion operations. Next, based on preset texture optimization rules, texture enhancement is performed on each image in the multi-dimensional image. For different types of pests, targeted adjustments are made to the images of corresponding frequency bands and resolutions based on their typical texture feature parameters, such as texture roughness and color contrast. Finally, an image fusion algorithm is used to fuse the texture-enhanced multi-dimensional images. Through weighted calculations and other methods, a fused image with enhanced texture representation characteristics is generated, enabling in-depth exploration and integration of the external morphological texture features of pests.

[0060] Step S3: Inputting the fused image into a forestry pest population identification algorithm, determining the population category of the pests based on different parameters of the forestry pests, and obtaining pest population category information;

[0061] Specifically, this step is a key step in achieving accurate pest identification. Its significance lies in converting the characteristic information contained in the fused image into specific pest population categories, providing a basis for the subsequent development of targeted prevention and control strategies. Forest pests are numerous and diverse, and different species differ in morphology, physiology, and ecology. The forest pest population identification algorithm, combined with multiple parameters such as pest size, antennae morphology, wing structure, and color markings, can accurately distinguish different population categories, avoiding inappropriate prevention and control measures due to misjudgment and improving the accuracy and effectiveness of prevention and control efforts.

[0062] The implementation process first requires the establishment of a large and accurate database of forest pest population characteristic parameters, storing detailed characteristic parameter information for various known pests. When the fused image is fed into the population identification algorithm, the algorithm automatically extracts the morphological characteristic parameters of the pests in the image and then compares and matches these extracted parameters with the data in the characteristic parameter database. By calculating the similarity index between the characteristic parameters, the degree of similarity between the pests in the fused image and each known population is assessed. Based on pre-set judgment rules, the pest population category is ultimately determined, and accurate pest population category information is output.

[0063] Step S4: establishing a pest population spatial distribution model based on the pest population category information and different parameters of forest pests, and determining the spatial distribution range of the pests in the forestry area;

[0064] Specifically, these steps aim to clarify the specific distribution and range of pests within forestry areas, providing data support for scientific planning of control areas and the rational allocation of control resources. Understanding the spatial distribution of pests helps forestry managers quickly locate infested areas, avoid blind control efforts, improve control efficiency, and reduce control costs. Furthermore, by analyzing the spatial distribution range, it is possible to identify pest spread trends, enabling proactive preventive measures to prevent further expansion of the infestation.

[0065] During the implementation process, the entire forestry area is first divided into grids according to certain rules to form multiple grid units with uniform specifications. Then, based on the pest population category information, the number, density and other parameters of different types of pests in each grid unit are counted. Combined with parameters such as the activity radius and flight ability of the pests, the influence range of each individual pest within the grid is determined. Using the spatial distribution model, the number of pests in the grid, the influence range, and the geographical environmental factors between adjacent grids (such as topography, vegetation type, etc.) are comprehensively considered to calculate the probability parameters of the spread of pests between adjacent grids. Finally, based on these parameters and the distribution status of pests in each grid, a complete spatial distribution model of pest populations is constructed, which clearly and intuitively presents the spatial distribution range of pests within the forestry area.

[0066] Step S5: combining the growth cycle parameters and environmental adaptability parameters of forest pests to predict the future development trend of pest populations and obtain pest population development prediction data;

[0067] Specifically, this step is of great significance for formulating long-term prevention and control plans in advance and preventing potential hazards. The growth and reproduction of forest pests are affected by both their own growth cycle and external environmental factors. By combining their growth cycle parameters (such as the duration of the egg stage, larval stage, pupal stage, and adult stage, and the growth characteristics of each stage) and environmental adaptability parameters (such as the tolerance range and suitability to environmental factors such as temperature, humidity, light, and soil conditions), it is possible to scientifically predict the changes in the number of pest populations, the direction of spread, and the degree of harm in the future, so that forestry managers can make preparations in advance and avoid passive situations caused by sudden disasters.

[0068] In the specific implementation, first of all, by analyzing the information on the categories of pest populations, the corresponding growth cycle parameters and environmental adaptability parameters are determined. Next, real-time environmental data for the current and future period in the forestry area is collected, including meteorological data such as temperature, humidity, rainfall, and duration of sunlight, as well as soil data such as soil pH and fertility. Then, based on the current population size, growth stage, and growth and reproduction patterns of the pests, combined with environmental data and environmental adaptability parameters, the population development prediction model is used to calculate the population growth forecast value and diffusion trend in the future short period (such as one week, one month). Finally, considering the environmental change trend parameters (such as environmental changes brought about by future seasonal changes), the short-term prediction value is revised to obtain the development forecast data of the pest population in the future for a longer period (such as one quarter, one year), providing an accurate basis for formulating scientific and reasonable prevention and control strategies.

[0069] Step S6: Generate forestry pest intelligent prevention and control early warning information based on pest population category information, spatial distribution range and development forecast data.

[0070] Specifically, this step is the final output of the entire intelligent pest prevention and early warning method. Its purpose is to integrate the various key information obtained in the previous steps and present it to forestry managers in an intuitive and easy-to-understand manner, providing comprehensive and accurate reference for their decision-making. The generated intelligent pest prevention and early warning information can clearly inform managers of the pest type, current distribution area, and future development trends, allowing managers to quickly understand the overall status of pests in the forest area, thereby formulating and implementing appropriate prevention and control measures in a timely and accurate manner, effectively reducing the damage caused by pests to forestry resources.

[0071] During implementation, the pest population category information, spatial distribution range, and development forecast data are first encoded and processed, and these data are converted into a format that is easy for computers to process and transmit. Then, according to the preset information fusion rules, the encoded information is integrated to form intelligent prevention and control early warning information containing basic pest information, distribution maps, development trend charts, etc. Visualization technology can be used to present the spatial distribution range in the form of map annotations and to display the development forecast data in the form of charts such as line charts and bar charts, making the early warning information more intuitive and clear. Finally, the generated intelligent prevention and control early warning information is sent to relevant forestry management personnel in a timely manner through various means such as text messages, emails, and forestry management information system push, ensuring that they can obtain information and take action in the first time.

[0072] Preferably, the texture optimization model of the multi-dimensional image fusion is specifically:

[0073] F fused (x,y)=α·T enhanced1 (x,y)+β·Tenhanced2 (x,y)

[0074] Among them, F fused (x,y) represents the pixel value of the fused image at the coordinate (x,y); T enhanced1 (x, y) is the pixel value of the first frequency band image at the coordinate (x, y) after enhancement based on the surface texture roughness parameters of forestry pests; T enhanced2 (x, y) is the pixel value of the second frequency band image at the coordinate (x, y) after enhancement based on the pest texture color contrast parameter; α and β are weight coefficients, and satisfy α+β=1. Their values ​​are dynamically adjusted according to the typical texture feature parameters of different types of forestry pests.

[0075] Specifically, through the texture optimization model of multi-dimensional image fusion, targeted enhancement is performed on images of different frequency bands. Its technical parameters include the first frequency band image enhanced based on the surface texture roughness parameters of forestry pests, the second frequency band image enhanced based on the texture color contrast parameters, and the weight coefficient dynamically adjusted according to different types of pests. The significance of this model lies in breaking through the limitations of traditional image fusion that only pursues visual effects, focusing on enhancing the texture characteristics of pests. By performing differentiated processing on images of different frequency bands, the fused image can more prominently present the key texture details of the pests, providing better quality image data for subsequent accurate identification. During implementation, the images of different frequency bands are first preprocessed, then enhanced according to the preset texture optimization rules, and finally fused by dynamically adjusting the weight coefficient to ensure that the fused image retains the texture feature information of the pests to the greatest extent.

[0076] Preferably, the forestry pest population identification algorithm includes the following model:

[0077]

[0078] Among them, P class It represents the probability that the pest belongs to a certain population category; n is the number of characteristic parameters of forest pests used for identification; S featurei is the i-th characteristic parameter, including the pest body size parameters, antennae morphology parameters, and the similarity score with the corresponding population standard characteristic parameters; w i is the weight of the i-th feature parameter, which is determined by machine learning algorithm training based on historical forest pest identification data according to the importance of the feature in distinguishing different populations.

[0079] Specifically, the forest pest population identification algorithm achieves accurate determination of pest population categories by comprehensively considering multiple feature parameters and their weights. Its technical parameters include the number of feature parameters used for identification, the similarity score between each feature parameter and the standard parameter, and the feature weight determined based on historical data training. The significance of this algorithm lies in overcoming the limitations of single feature judgment in traditional identification methods, and improving the accuracy and robustness of identification through multi-feature fusion and dynamic adjustment of weights. During implementation, the morphological feature parameters of the pests in the fused image are first extracted, and then matched with the preset feature parameter library to calculate the similarity. Finally, the similarity scores are weighted averaged according to the feature weights to obtain the probability that the pest belongs to a certain population category, thereby achieving accurate identification.

[0080] Preferably, the process of establishing the pest population spatial distribution model is further based on the following model:

[0081] S distribution (x,y)=f(P detection (x,y),S density (x,y))

[0082] Among them, D distribution (x,y) represents the spatial distribution of pests at the coordinate (x,y); P detection (x, y) is the probability of detecting pests at the coordinate (x, y), which is determined by the pest characteristic parameters of the corresponding area in the fused image and the population recognition algorithm; S density (x,y) is the population density at coordinate (x,y) calculated based on the population quantity parameters and activity range parameters of forest pests; f is the spatial distribution mapping function, and its parameters are adaptively adjusted according to the topographic parameters and vegetation cover parameters of the forestry area.

[0083] Specifically, the spatial distribution model of pest populations comprehensively considers detection probability and population density, and accurately determines the spatial distribution range of pests within forestry areas through adaptively adjusted spatial distribution mapping functions. Its technical parameters include detection probability, population density calculated based on population size and activity range, and mapping function parameters that are adaptively adjusted according to parameters such as topography, landform, and vegetation cover. The significance of this model lies in fully considering the complexity of the forestry environment and the activity characteristics of pests, and improving the accuracy of spatial distribution prediction by dynamically adjusting model parameters. During implementation, the detection probability of each area is first determined using fused images and population recognition algorithms, and then the population density is calculated based on the population size and activity range. Finally, the detection probability and population density are converted into spatial distribution states through an adaptively adjusted mapping function to construct a complete spatial distribution model.

[0084] Preferably, the process of predicting the future development trend of the pest population uses the following model:

[0085]

[0086] Among them, G future G represents the predicted value of forest pest population size at a certain moment in the future; current is the current pest population size, which is determined by the population identification algorithm and spatial distribution model; r is the growth coefficient calculated based on the pest growth rate parameters and reproduction cycle parameters; E suitability E is the suitability parameter of the current forestry environment for pests, which is a comprehensive combination of temperature, humidity, and light environment parameters; standard It is the standard environmental suitability parameter for the growth and reproduction of pests.

[0087] Specifically, the population development prediction model achieves scientific predictions on the future development trends of pests by comprehensively considering the current population size, growth coefficient and environmental suitability. Its technical parameters include the current population size, the growth coefficient calculated based on the growth rate and reproduction cycle, the suitability parameters of the comprehensive environmental parameters, and the standard environmental suitability parameters. The significance of this model lies in breaking through the limitations of traditional prediction methods that only consider a single factor. Through the fusion of multiple parameters, it comprehensively evaluates the impact of environmental factors on the growth and reproduction of pests, thereby improving the accuracy and foresight of the prediction. During implementation, the current population size is first determined through the population identification algorithm and the spatial distribution model, and then the growth coefficient is calculated in combination with the growth rate and reproduction cycle. At the same time, the current environmental suitability is evaluated, and finally these parameters are substituted into the model to calculate the future population size forecast value, providing a scientific basis for prevention and control decisions.

[0088] Preferably, the process of generating forestry pest intelligent prevention and control early warning information adopts the following model:

[0089] W information =I category ⊕I distribution ⊕I prediction

[0090] Among them, W information Indicates the generated intelligent prevention and control warning information; I category It is the information code of pest population category, which is converted from the result of population identification algorithm; I distribution Encode the spatial distribution information of pests and generate output data based on the spatial distribution model; I prediction It is the encoding of pest development prediction information, which is determined according to population development prediction data; ⊕ is the information fusion operator, which integrates different information according to the preset coding rules.

[0091] Specifically, the intelligent pest prevention and control early warning information generation model generates comprehensive and accurate early warning information by encoding and fusing pest categories, distribution, and forecast information. Its technical parameters include category information encoding converted from population identification results, distribution information encoding based on the output of the spatial distribution model, forecast information encoding determined based on development forecast data, and fusion operators for information integration according to preset rules. The significance of this model lies in the standardization and organic integration of complex multi-source information, making the early warning information more intuitive, easy to understand, and easy to apply. During implementation, various types of information are first encoded and converted, and then integrated according to preset rules through fusion operators, ultimately generating intelligent pest prevention and control early warning information containing pest categories, distribution ranges, and development trends, providing decision support for forestry managers.

[0092] Preferably, the step S3 specifically includes:

[0093] Step S31: extracting morphological characteristic parameters of the pests in the fused image, including but not limited to body outline parameters, color texture parameters, and structure ratio parameters, to provide basic data for population identification;

[0094] Step S32: matching the extracted morphological characteristic parameters with a preset database of characteristic parameters of various forestry pest populations, and calculating a similarity index of each characteristic parameter;

[0095] Step S33: Based on the similarity index, a forestry pest population identification algorithm is used to perform preliminary population classification screening of the pests;

[0096] Step S34: Through the secondary feature verification mechanism, combined with the pest behavior habit parameters, the preliminary screening results are corrected to determine the final pest population category information.

[0097] Specifically, step S3 realizes accurate identification of pest population categories through four sub-steps: extracting feature parameters, matching and calculating similarity, preliminary screening and secondary verification. Extracting morphological feature parameters provides basic data for subsequent identification, matching and calculating similarity evaluates the similarity between pests and standard populations, preliminary screening quickly narrows the identification range, and secondary verification corrects the preliminary results in combination with behavioral habit parameters to ensure the accuracy of identification. The significance of this step is to overcome the limitations of single feature identification and improve the reliability of identification through rigorous process design and multi-dimensional verification. During implementation, the four sub-steps are carried out in sequence, each step is closely connected, and the identification results are gradually refined to finally determine the accurate pest population category information.

[0098] Preferably, the step S4 specifically includes:

[0099] Step S41: Divide the forest area into grids, and count the number parameters of the pests in each grid based on the pest population category information;

[0100] Step S42: determining the influence range of each individual forest pest within the grid based on the activity radius parameter of the forest pest;

[0101] Step S43: Calculate the probability parameters of pest spread between adjacent grids using a spatial distribution model based on the number of pests and their impact range in each grid;

[0102] Step S44: Based on the diffusion probability parameters and the distribution status of the pests in each grid, a complete spatial distribution model of the pest population is constructed to determine the spatial distribution range.

[0103] Specifically, step S4 achieves accurate determination of the spatial distribution range of pests through four sub-steps: gridding, determining the impact range, calculating the diffusion probability, and constructing a spatial distribution model. Gridding simplifies complex forestry areas into regular units for easy statistics and analysis; determining the impact range clarifies the area of ​​influence of each individual pest; calculating the diffusion probability takes into account the activity characteristics of the pests and environmental factors; and constructing a spatial distribution model to integrate various parameters and intuitively present the spatial distribution status of the pests. This step provides a scientific basis for prevention and control area planning and resource allocation through refined spatial analysis. During implementation, the four sub-steps must be strictly followed in order to ensure the accuracy and reliability of the spatial distribution model.

[0104] Preferably, the step S5 specifically includes:

[0105] Step S51: obtaining the growth cycle stage parameters of the current forestry pests, and analyzing their growth characteristics and reproduction patterns in this stage;

[0106] Step S52: collecting real-time environmental parameters within the forestry area, and combining them with the environmental adaptability parameters of the pests to assess the impact of the current environment on the growth and reproduction of the pests;

[0107] Step S53: Based on the pest population parameters and growth and reproduction patterns, using the population development prediction model, calculate the predicted population growth value in the short term in the future;

[0108] Step S54: The short-term prediction value is corrected in combination with the environmental change trend parameter to obtain the future longer-term development prediction data of the pest population.

[0109] Specifically, step S5 realizes the scientific prediction of the future development trend of the pest population through four sub-steps: obtaining growth cycle stage parameters, evaluating environmental impacts, calculating short-cycle prediction values, and revising long-cycle prediction data. Acquiring growth cycle stage parameters clarifies the current growth status and reproduction patterns of pests; evaluating environmental impacts to analyze the promoting or inhibiting effects of the current environment on the growth and reproduction of pests; calculating short-cycle prediction values ​​to provide a reference for short-term prevention and control decision-making; revising long-cycle prediction data to consider environmental change trends and improve the accuracy of long-term predictions. This step provides a reliable basis for the formulation of long-term prevention and control plans by comprehensively analyzing the characteristics of the pests themselves and environmental factors. During implementation, the four sub-steps are completed in sequence to ensure that the prediction results can reflect the current actual situation and adapt to future environmental changes.

[0110] like Figure 2 As shown in the figure, the forestry pest intelligent prevention and early warning system based on image recognition includes:

[0111] A multi-dimensional image acquisition unit for acquiring multi-dimensional images containing potential harmful organisms in forestry areas;

[0112] An image fusion processing unit is connected to the multi-dimensional image acquisition unit and is used to process the acquired multi-dimensional image through a texture optimization model of multi-dimensional image fusion to construct a fused image;

[0113] The population identification processing unit is connected to the image fusion processing unit and is used to input the fused image into the forestry pest population identification algorithm to determine the population category of the pests;

[0114] A spatial distribution modeling unit, connected to the population identification processing unit, is used to establish a pest population spatial distribution model based on pest population category information and different parameters of forest pests;

[0115] Develop a prediction unit, connected to the spatial distribution modeling unit, to combine the growth cycle parameters and environmental adaptability parameters of forest pests to predict the future development trend of pest populations;

[0116] The early warning information generation unit is connected to the development forecast unit and is used to generate early warning information for intelligent prevention and control of forestry pests based on the pest population category information, spatial distribution range and development forecast data.

[0117] This image recognition-based intelligent forest pest prevention and early warning method and system effectively overcomes the shortcomings of existing technologies, such as limited image information and insufficient feature extraction, through multidimensional image acquisition and deep texture optimization. Traditional methods rely on single-band or low-resolution images, making it difficult to fully capture the morphological details of pests. This system, however, acquires multidimensional images encompassing different spectral bands and spatial resolutions. Using a texture optimization model, it enhances and fuses these multidimensional images based on parameters such as the pest's surface texture roughness and color contrast. This system deeply explores the pest's external morphological and texture features, constructing a fused image with enhanced characterization, laying a solid foundation for subsequent accurate identification.

[0118] In the process of identifying and analyzing pest populations, the system comprehensively integrates various parameters of forest pests, breaking through the limitations of insufficient utilization of parameters in traditional algorithms. Traditional identification algorithms often ignore key parameters such as the growth cycle and behavioral habits of pests, resulting in inaccurate identification and prediction. This system uses a population identification algorithm to integrate multiple characteristic parameters such as body size and antennae morphology, calculate feature similarity and weights, and accurately determine the population category; when establishing a spatial distribution model, it combines parameters such as population size, activity range, topography, etc. to determine the spatial distribution range of pests within the forestry area; when predicting future development trends, it considers parameters such as growth cycle and environmental adaptability, scientifically evaluates changes in population size, and achieves precise control of the entire process of pest identification from category identification to development trend prediction.

[0119] Furthermore, the system achieves intelligent and precise management of the entire process, from image acquisition to early warning information generation, significantly improving the efficiency of forest pest control. Six functional units, including the multidimensional image acquisition unit and the image fusion processing unit, work closely together. Through the organic integration of models and algorithms in each link, they can promptly generate intelligent early warning information containing population classification, spatial distribution, and development forecast data. Compared with traditional manual inspections or simple image recognition methods, this greatly shortens the monitoring cycle and reduces labor costs, providing efficient and accurate decision-making basis for forest pest control, and effectively promoting the development of forest pest control work in the direction of intelligence and refinement.

[0120] In the description of the present invention, it should be noted that, unless otherwise expressly specified or limited, the terms "disposed," "installed," "connected," "connected," and "fixed" should be understood in a broad sense. For example, they may refer to fixed connections, detachable connections, or integral connections; they may refer to mechanical connections or electrical connections; they may refer to direct connections or indirect connections through an intermediate medium; and they may refer to internal communication between two components. Those skilled in the art will understand the specific meanings of the above terms in the present invention based on specific circumstances.

[0121] Although embodiments of the present invention have been shown and described, it will be understood by those skilled in the art that various equivalent changes, modifications, substitutions and variations may be made to these embodiments without departing from the principles and spirit of the present invention, and that the scope of the present invention is defined by the appended claims and their equivalents.

Claims

1. An intelligent forest pest prevention and early warning method based on image recognition, characterized in that: The following steps are involved: Step S1: Acquire a multidimensional image containing potential harmful organisms in a forestry area, wherein the multidimensional image includes image data of different spectral bands and spatial resolutions; Step S2: Processing the acquired multidimensional images through a texture optimization model of multidimensional image fusion to construct a fused image with enhanced texture representation characteristics, and deeply mining the external morphological texture features of forest pests; Step S3: Inputting the fused image into a forestry pest population identification algorithm, determining the population category of the pests based on different parameters of the forestry pests, and obtaining pest population category information; Step S4: establishing a pest population spatial distribution model based on the pest population category information and different parameters of forest pests, and determining the spatial distribution range of the pests in the forestry area; Step S5: combining the growth cycle parameters and environmental adaptability parameters of forest pests to predict the future development trend of pest populations and obtain pest population development prediction data; Step S6: Generate forestry pest intelligent prevention and control early warning information based on pest population category information, spatial distribution range and development forecast data.

2. The method for intelligent prevention and early warning of forest pests based on image recognition according to claim 1, characterized in that: The texture optimization model of the multi-dimensional image fusion is specifically: F fused (x,y)=α·T enhanced1 (x,y)+β·T enhanced2 (x,y) Among them, F fused (x,y) represents the pixel value of the fused image at the coordinate (x,y); T enhanced1 (x, y) is the pixel value of the first frequency band image at the coordinate (x, y) after enhancement based on the surface texture roughness parameters of forestry pests; T enhanced2 (x, y) is the pixel value of the second frequency band image at the coordinate (x, y) after enhancement based on the pest texture color contrast parameter; α and β are weight coefficients, and satisfy α+β=1. Their values ​​are dynamically adjusted according to the typical texture feature parameters of different types of forestry pests.

3. The method for intelligent prevention and early warning of forest pests based on image recognition according to claim 1, characterized in that: The forest pest population identification algorithm includes the following models: Among them, P class It represents the probability that the pest belongs to a certain population category; n is the number of characteristic parameters of forest pests used for identification; S featurei is the i-th characteristic parameter, including the pest body size parameters, antennae morphology parameters, and the similarity score with the corresponding population standard characteristic parameters; w i is the weight of the i-th feature parameter, which is determined by machine learning algorithm training based on historical forest pest identification data according to the importance of the feature in distinguishing different populations.

4. The method for intelligent prevention and early warning of forest pests based on image recognition according to claim 1, characterized in that: The process of establishing a pest population spatial distribution model is further based on the following model: D distribution (x,y)=f(P detection (x,y),S density (x,y)) Among them, D distribution (x,y) represents the spatial distribution of pests at the coordinate (x,y); P detection (x, y) is the probability of detecting pests at the coordinate (x, y), which is determined by the pest characteristic parameters of the corresponding area in the fused image and the population recognition algorithm; S density (x,y) is the population density at coordinate (x,y) calculated based on the population quantity parameters and activity range parameters of forest pests; f is the spatial distribution mapping function, and its parameters are adaptively adjusted according to the topographic parameters and vegetation cover parameters of the forestry area.

5. The method for intelligent prevention and early warning of forest pests based on image recognition according to claim 1, characterized in that: The process of predicting the future development trend of pest populations uses the following model: Among them, G future G represents the predicted value of forest pest population size at a certain moment in the future; current is the current pest population size, which is determined by the population identification algorithm and spatial distribution model; r is the growth coefficient calculated based on the pest growth rate parameters and reproduction cycle parameters; E suitability E is the suitability parameter of the current forestry environment for pests, which is a comprehensive combination of temperature, humidity, and light environment parameters; standard It is the standard environmental suitability parameter for the growth and reproduction of pests.

6. The method for intelligent prevention and early warning of forest pests based on image recognition according to claim 1, characterized in that: The process of generating forestry pest intelligent prevention and control early warning information adopts the following model: Among them, W information Indicates the generated intelligent prevention and control warning information; I category It is the information code of pest population category, which is converted from the result of population identification algorithm; I distribution Encode the spatial distribution information of pests and generate output data based on the spatial distribution model; I prediction Encoding of pest development forecast information, determined based on population development forecast data; It is an information fusion operator that integrates different information according to preset encoding rules.

7. The method for intelligent prevention and early warning of forest pests based on image recognition according to claim 1, characterized in that: The step S3 specifically includes: Step S31: extracting morphological characteristic parameters of the pests in the fused image, including but not limited to body contour parameters, color texture parameters, and structure ratio parameters; Step S32: matching the extracted morphological characteristic parameters with a preset database of characteristic parameters of various forestry pest populations, and calculating a similarity index of each characteristic parameter; Step S33: Based on the similarity index, a forestry pest population identification algorithm is used to perform preliminary population classification screening of the pests; Step S34: Through the secondary feature verification mechanism, combined with the pest behavior habit parameters, the preliminary screening results are corrected to determine the final pest population category information.

8. The method for intelligent prevention and early warning of forest pests based on image recognition according to claim 1, characterized in that: The step S4 specifically includes: Step S41: Divide the forest area into grids, and count the number parameters of the pests in each grid based on the pest population category information; Step S42: determining the influence range of each individual forest pest within the grid based on the activity radius parameter of the forest pest; Step S43: Calculate the probability parameters of pest spread between adjacent grids using a spatial distribution model based on the number of pests and their impact range in each grid; Step S44: Based on the diffusion probability parameters and the distribution status of the pests in each grid, a complete spatial distribution model of the pest population is constructed to determine the spatial distribution range.

9. The method for intelligent prevention and early warning of forest pests based on image recognition according to claim 1, characterized in that: The step S5 specifically includes: Step S51: obtaining the growth cycle stage parameters of the current forestry pests, and analyzing their growth characteristics and reproduction patterns in this stage; Step S52: collecting real-time environmental parameters within the forestry area, and combining them with the environmental adaptability parameters of the pests to assess the impact of the current environment on the growth and reproduction of the pests; Step S53: Based on the pest population parameters and growth and reproduction patterns, using the population development prediction model, calculate the predicted population growth value in the short term in the future; Step S54: The short-term prediction value is corrected in combination with the environmental change trend parameter to obtain the future longer-term development prediction data of the pest population.

10. The forestry pest intelligent prevention and early warning system based on image recognition is characterized by: include: A multi-dimensional image acquisition unit for acquiring multi-dimensional images containing potential harmful organisms in forestry areas; An image fusion processing unit is connected to the multi-dimensional image acquisition unit and is used to process the acquired multi-dimensional image through a texture optimization model of multi-dimensional image fusion to construct a fused image; The population identification processing unit is connected to the image fusion processing unit and is used to input the fused image into the forestry pest population identification algorithm to determine the population category of the pests; A spatial distribution modeling unit, connected to the population identification processing unit, is used to establish a pest population spatial distribution model based on pest population category information and different parameters of forest pests; Develop a prediction unit, connected to the spatial distribution modeling unit, to combine the growth cycle parameters and environmental adaptability parameters of forest pests to predict the future development trend of pest populations; The early warning information generation unit is connected to the development forecast unit and is used to generate early warning information for intelligent prevention and control of forestry pests based on the pest population category information, spatial distribution range and development forecast data.

Images