A forestry forest pest image intelligent identification prevention and control decision system

Abstract

The application discloses a forestry forest disease and pest image intelligent identification prevention and control decision system, and both sides of the fixing frame are provided with fixing bands, the fixing bands are provided with fixing holes, the fixing holes are provided with fasteners, the identification assembly is located on the front side of the fixing frame and comprises a hinged plate, the hinged plate is fixedly connected with the fixing frame, a sensing camera is rotationally connected to the front side of the hinged plate, the sensing camera collects macro images of leaves, branches, whole plants and forest areas, captures disease spots, insect holes, discoloration and wilting disease and pest characteristics, and both sides of the fixing frame are provided with sensing assemblies. Compared with the prior art, the application has the advantages that the application adopts image visual identification and multi-dimensional environment and pest sensing fusion sensing technology, and breaks through the technical limitation of single image identification.

Description

Technical Field

[0001] This invention relates to the field of decision-making systems, specifically a forestry tree disease and pest image intelligent recognition and control decision-making system. Background Technology

[0002] Forestry resources are a core component of ecological environmental protection and ecological civilization construction. Routine monitoring, accurate identification, and scientific control of forest pests and diseases are crucial for ensuring stable forest growth and reducing forestry disaster losses. Currently, pest and disease control in domestic forest areas still largely relies on manual inspections and traditional fixed-point monitoring equipment. The overall level of intelligence and precision is low, and there are many technical shortcomings and industry pain points in practical applications, seriously hindering the high-quality development of forestry management.

[0003] First, existing pest and disease identification and monitoring technologies suffer from low accuracy and insufficient early warning capabilities. Current mainstream monitoring methods mostly rely on single cameras for image capture and recognition, which can only identify obvious lesions, insect holes, and wilting features that appear in the middle and later stages of pest and disease development. They are ineffective in identifying latent pests and diseases that lack obvious outward signs of infection in their early or dormant stages. Furthermore, the complex forest environment, with its uneven lighting, foliage obstruction, and overlapping trees, makes single-visual identification highly susceptible to misjudgments and missed detections. This hinders early detection and warning of pests and diseases, often resulting in reactive control measures only after large-scale outbreaks, leading to extremely poor timeliness.

[0004] Secondly, traditional monitoring equipment suffers from poor adaptability and insufficient stability in field operation. Most forestry monitoring equipment on the market has a fixed structure, which can only be installed on tree trunks of fixed specifications. It cannot adapt to the monitoring needs of trees of different ages and diameters, making it difficult to cover monitoring operations in various forest areas such as young forests, mature forests, and economic forests. Moreover, the fixed structure of traditional equipment has poor stability. In the wild, strong winds, rain, snow, frost, and other severe weather conditions can easily cause the equipment to shake, shift, or even fall off, resulting in a high failure rate. This requires frequent on-site inspections and maintenance by staff, leading to high manual maintenance costs. At the same time, the rigid fixed structure can easily compress and damage tree branches and trunks, affecting the normal growth of trees.

[0005] Secondly, data transmission in remote forest areas is limited, resulting in poor continuity of monitoring data and delayed decision-making. Most of my country's forest areas are located in mountainous and remote wilderness areas with incomplete public network signal coverage. Traditional monitoring equipment often uses a single communication transmission mode, which cannot adapt to the complex communication environment of forest areas. This easily leads to problems such as data delays, disconnections, and data loss, resulting in fragmented and discontinuous monitoring data. The cloud cannot obtain complete and effective forest area monitoring data, causing delays in the prediction of pest and disease risks and failing to meet the needs of normalized and uninterrupted monitoring of the entire forest area. Summary of the Invention

[0006] The technical problem to be solved by this invention is to overcome the above-mentioned technical defects.

[0007] To solve the above problems, the technical solution of the present invention is: an intelligent image recognition and control decision-making system for forestry tree diseases and pests, including a fixing frame, a recognition component and a sensing component; A fixing strap is inserted between the two sides of the fixing frame. The fixing strap has fixing holes, and fasteners are inserted into the fixing holes. The identification component is located on the front side of the fixing frame and includes a hinge plate. The hinge plate is fixedly connected to the fixing frame. A sensing camera is rotatably connected to the front side of the hinge plate. The sensing camera collects macroscopic images of tree leaves, branches, whole trees, and forest areas, capturing disease and pest characteristics such as lesions, insect holes, discoloration, and wilting. The two sides of the fixing frame are equipped with sensing components, including temperature and humidity, light, and soil moisture sensors, forest acoustic sensors, and pheromone / volatile organic compound sensors, to supplement indirect characteristics between the environment and pests. A transmission layer is connected to the top of the fixing frame. A control button is located on the side of the fixing frame. The control button is electrically connected to the transmission layer and the sensing camera.

[0008] Furthermore, the fixing strap is made of elastic material and can be adjusted and fixed according to different tree diameters.

[0009] Furthermore, the transmission layer uses G / G, LoRa, and satellite communication methods to upload the perception layer data to the cloud platform in real time, ensuring stable data transmission in remote forest areas.

[0010] Furthermore, the bottom of the fixed frame is equipped with a decision layer, which, together with the transmission layer, includes a risk warning module, a zoned prevention and control decision module, and an operation optimization planning module. The risk warning module outputs low, medium, and high risk warning signals based on the pest and disease prediction results and preset risk thresholds. The zoned prevention and control decision module automatically divides the core prevention and control area, prevention buffer zone, and monitoring and observation area by combining the forest terrain, tree species, meteorological conditions, and pest and disease damage characteristics, and matches corresponding physical control, biological control, and chemical control strategies. The operation optimization planning module generates the optimal prevention and control operation path, pesticide dosage, and natural enemy release ratio scheme.

[0011] Further: By using macroscopic remote sensing images from satellites and drones to make preliminary judgments on abnormal areas in forest areas, preliminary diagnostic hypotheses and initial confidence levels for pests and diseases are generated. Microscopic images from the sensing layer and environmental sensor data are retrieved as verification evidence. The diagnostic confidence level is updated using the Bayesian formula, and the fusion weights of macroscopic and microscopic data are dynamically adjusted to output highly reliable pest and disease diagnosis results and damage level ratings, effectively reducing the misjudgment rate and missed judgment rate of single image recognition.

[0012] The advantages of this invention compared to existing technologies are: This invention employs image visual recognition combined with multi-dimensional environmental and pest sensing technologies, overcoming the limitations of single-image recognition. Traditional pest monitoring relies solely on single-camera imaging, which is susceptible to factors such as forest light obstruction, overlapping branches and leaves, and the lack of obvious visual characteristics in early-stage latent pests, leading to misjudgments and missed detections. This system uses sensing cameras to collect multi-scale images of tree leaves, branches, entire trees, and the forest area, accurately capturing the visible characteristics of pests and diseases. Simultaneously, it incorporates multiple sensors for temperature and humidity, soil moisture, forest acoustics, and volatile organic compounds to collect environmental conditions for pest and disease growth and latent pest activity characteristics, enabling precise monitoring of pests and diseases during their early incubation and initial infection stages.

[0013] The equipment is highly adaptable to installation and has high stability in the field, making it suitable for monitoring needs in all forest areas. It is fixed by using elastic material fixing straps with multiple sets of fixing holes and fasteners. The overall lightweight structure design ensures that it will not squeeze or damage tree branches after installation, and will not affect the normal growth of trees. At the same time, the locking and fixing structure can withstand strong winds, rain and snow in the forest. Attached Figure Description

[0014] Figure 1 This invention relates to a three-dimensional intelligent image recognition and control decision-making system for forestry tree diseases and pests. Figure 1 .

[0015] Figure 2 This invention relates to a three-dimensional intelligent image recognition and control decision-making system for forestry tree diseases and pests. Figure 2 .

[0016] As shown in the figure: 1. Fixing frame; 2. Fixing strap; 3. Fixing hole; 4. Fastener; 5. Hinge plate; 6. Sensing camera; 7. Transmission layer; 8. Control button. Detailed Implementation

[0017] The specific embodiments of the present invention will be further described below with reference to the accompanying drawings. Identical components are indicated by the same reference numerals.

[0018] It should be noted that the terms “front,” “back,” “left,” “right,” “up,” and “down” used in the following description refer to the directions shown in the attached diagram, while the terms “inside” and “outside” refer to the directions toward or away from the geometric center of a specific component, respectively.

[0019] To make the content of this invention easier to understand, the technical solutions in the embodiments of this invention will be clearly and completely described below with reference to the accompanying drawings.

[0020] like Figures 1 to 2As shown, a forestry tree disease and pest image intelligent recognition and control decision-making system consists of a recognition component on the front side of the fixed frame 1 and sensing components on both sides forming a system perception layer. It collects direct feature data of diseases and pests and indirect feature data of the environment from all directions, providing data support for intelligent diagnosis.

[0021] The identification component is rigidly fixed to the mounting bracket 1 via the hinge plate 5. The sensing camera 6 can adaptively rotate and adjust the shooting angle from multiple angles. In daily operation, it continuously collects four types of core image data: First, microscopic images of tree leaves, accurately capturing subtle disease and pest characteristics such as leaf spots, insect holes, yellowing, curling and wilting, and mold spots; second, local images of branches and trunks, identifying disease and pest symptoms such as branch and trunk rot, insect holes, phloem damage, and gum exudation; third, overall images of individual trees, monitoring the overall abnormal state of the entire tree, such as growth, wilting, and sparse foliage; and fourth, macroscopic images of small areas of forest, recording the overall growth of trees in the area, achieving the effect of single-point monitoring and area coverage.

[0022] Meanwhile, the multi-dimensional sensing components integrated on both sides of the fixed frame 1 simultaneously collect environmental and pest data. Among them, temperature, humidity, light, and soil moisture sensors monitor local microclimate and soil environmental data in the forest area in real time, capturing environmental conditions that easily induce pests and diseases, such as high temperature and humidity, and soil drought / waterlogging. Forest acoustic sensors collect forest insect sounds and vibration sound wave data of pest activity to determine the activity level and population density of pests. Pheromones / volatile organic compounds sensors capture specific volatile substances released by trees after being attacked by pests and diseases. Combined with environmental data and image data, it makes up for the limitations of single image recognition, accurately supplements the indirect characteristics of pest incubation and early infection, and realizes "early detection and early warning" of pests and diseases.

[0023] The control button 8 on the side of the mounting bracket 1 enables manual operation such as starting and stopping the equipment, manually calibrating parameters, fine-tuning the shooting angle, and switching the data acquisition frequency. It is suitable for on-site debugging and emergency monitoring by forest patrol personnel. At the same time, the control button 8 is electrically linked with the sensing camera 6 and the transmission layer 7 throughout the process to ensure the real-time nature of equipment control commands and data transmission.

[0024] The transmission layer 7 mounted on top of the mounting frame 1 adopts a multi-mode converged communication transmission scheme to adapt to the communication environment of different forest areas. For forest areas with good signal, such as suburbs and plains, 4G / 5G network transmission is given priority; for forest areas in deep mountains and remote areas without public network coverage, LoRa narrowband IoT communication mode is automatically switched to achieve low power consumption and long-distance data transmission; for extremely remote forest areas with no signal, satellite communication is used as a backup transmission channel.

[0025] The transmission layer 7 encrypts and packages the image data, environmental sensor data, and pest characteristic data collected by the perception layer in real time, and uploads them to the cloud platform in real time. It also has the functions of data interruption resume and local caching. When the signal is briefly interrupted, it automatically stores local data and retransmits it in batches after the signal is restored. It completely solves the problems of data transmission lag, loss, and delay in remote forest areas, and ensures the stability and integrity of data transmission at all times and in all areas, providing continuous and effective data support for backend intelligent decision-making.

[0026] The decision-making layer and transmission layer 7 integrated at the bottom of the fixed frame 1 communicate with each other in real time. Relying on cloud big data algorithms, it completes intelligent pest and disease control decisions through three major modules: risk warning, zoned prevention and control, and operation optimization. At the same time, it combines multi-source data fusion algorithms from air, space, and ground to improve diagnostic accuracy.

[0027] First, macroscopic images of the entire forest area are acquired through satellite remote sensing and drone aerial photography to quickly screen high-risk areas such as withered trees, areas of chlorosis, and abnormal growth, completing the initial judgment of abnormal areas of pests and diseases and generating preliminary diagnostic hypotheses and initial confidence levels for pests and diseases. Subsequently, microscopic images of leaves and branches collected by the device's sensing layer, as well as sensor data such as temperature, humidity, insect acoustics, and volatile organic compounds, are retrieved as verification evidence. The confidence level of pest and disease diagnosis is updated in real time using Bayesian formulas. The fusion weight of macroscopic remote sensing data and locally collected microscopic data is dynamically adjusted according to the forest environment and pest and disease types. Finally, highly reliable diagnostic results of pest and disease types are output, and mild, moderate, and severe damage levels are rated. This effectively avoids problems such as light interference, occlusion misjudgment, and missed detection of hidden pests in single image recognition, improving the accuracy of pest and disease identification to over 95%.

[0028] The risk warning module has a built-in database of pest and disease damage thresholds, which includes critical indicators for different tree species, seasons, and pests and diseases. Combining real-time diagnostic results, pest and disease spread trends, and environmental triggering conditions, it automatically compares the data with preset risk thresholds and outputs three levels of warning signals: low risk is sporadic pest and disease occurrence with no spread trend, and only monitoring records are kept; medium risk is small-scale pest and disease infection with a risk of spread, and a warning is pushed out; high risk is a large-scale pest and disease outbreak with rapid spread, and an emergency warning is pushed out and linked to the forest area prevention and control terminal to remind staff to take emergency measures.

[0029] The zoned prevention and control decision-making module integrates four core parameters: forest topography, tree species, meteorological conditions, and pest and disease damage characteristics, to intelligently divide the monitored forest area into zones. Areas with concentrated outbreaks and high levels of pest and disease damage are designated as core control zones, employing high-intensity control strategies including precise chemical control, manual extermination, and concentrated pest eradication. Areas surrounding the core zones and at risk of further infestation are designated as prevention buffer zones, using a strategy primarily based on biological control supplemented by physical control methods, such as releasing natural enemies of pests and diseases, hanging insect-attracting lamps, and applying insecticides to block the spread of pests and diseases. Areas free of pests and diseases and with excellent environmental conditions are designated as monitoring and observation zones, requiring only routine monitoring without active control, thus minimizing control costs and ecological damage.

[0030] The operation optimization planning module, based on the results of zoning prevention and control, combined with the forest road network, operation scope, and distribution density of pests and diseases, automatically generates the optimal prevention and control operation path, avoiding terrain obstacles and shortening the time for inspection and pest control operations. At the same time, it accurately calculates the amount of pesticides to be sprayed based on the number of trees and the level of pest and disease damage, avoiding environmental pollution caused by excessive pesticides or prevention and control failure due to insufficient dosage. For biological control scenarios, it intelligently calculates the release ratio and release points of natural enemy insects such as ladybugs and parasitic wasps, realizing scientific, precise, and low-cost optimization of prevention and control operations.

[0031] In actual operation, the system first secures the equipment in the forest using the fixing frame 1 and fixing strap 2. Staff then activate the equipment via control button 8. The perception layer's identification and sensing components continuously collect images of forest pests and diseases, as well as environmental and pest characteristic data, 24 / 7. The transmission layer 7 adaptively switches transmission modes based on forest area communication conditions, uploading data to the cloud in real time. The decision-making layer relies on multi-source data fusion algorithms to complete intelligent diagnosis and hazard rating of pests and diseases, simultaneously conducting risk warnings, zoned prevention and control, and operational plan optimization. Finally, the warning information, prevention and control zone results, and optimized operational plans are pushed to forest area management terminals and mobile mini-programs to guide staff in accurately carrying out pest and disease control operations, achieving an intelligent upgrade of forest pest and disease management from passive treatment to proactive prediction and precise control.

[0032] The present invention and its embodiments have been described above. This description is not restrictive, and the accompanying drawings are only one embodiment of the present invention; the actual structure is not limited thereto. In conclusion, if those skilled in the art are inspired by this description and design similar structures and embodiments without departing from the spirit of the invention, such designs should fall within the protection scope of the present invention.

Claims

1. A forestry tree disease and pest image intelligent recognition and control decision-making system, characterized in that: Includes a mounting bracket (1), an identification component, and a sensing component; A fixing strap (2) is inserted between the two sides of the fixing frame (1). The fixing strap (2) has a fixing hole (3). A fastener (4) is inserted into the fixing hole (3). The identification component is located on the front side of the fixing frame (1). It includes a hinge plate (5). The hinge plate (5) is fixedly connected to the fixing frame (1). A sensing camera (6) is rotatably connected to its front side. The sensing camera (6) collects macroscopic images of tree leaves, branches, whole trees and forest areas, and captures disease and pest characteristics such as lesions, insect holes, discoloration, and wilting. The two sides of the fixing frame (1) are provided with sensing components. The sensing components include temperature and humidity, light, and soil moisture sensors, forest acoustic sensors, and pheromone / volatile organic compound sensors to supplement the indirect characteristics of the environment and pests. A transmission layer (7) is connected above the fixing frame (1). A control button (8) is provided on the side of the fixing frame (1). The control button (8) is electrically connected to the transmission layer (7) and the sensing camera (6).

2. The intelligent image recognition and control decision-making system for forestry tree diseases and pests according to claim 1, characterized in that: The fixing strap (2) is made of elastic material and can be adjusted and fixed according to different tree diameters.

3. The intelligent image recognition and control decision-making system for forestry tree diseases and pests according to claim 1, characterized in that: The transmission layer (7) uses 5G / 4G, LoRa, and satellite communication to upload the perception layer data to the cloud platform in real time, ensuring stable data transmission in remote forest areas.

4. The intelligent image recognition and control decision-making system for forestry tree diseases and pests according to claim 1, characterized in that: The bottom of the fixed frame (1) is provided with a decision layer. The decision layer and the transmission layer (7) include a risk warning module, a zoned prevention and control decision module and an operation optimization planning module. The risk warning module outputs low, medium and high risk warning signals based on the pest and disease prediction results and preset risk thresholds. The zoned prevention and control decision module automatically divides the core prevention and control area, prevention buffer zone and monitoring and observation area by combining the forest area terrain, tree species, meteorological conditions and pest and disease damage characteristics, and matches the corresponding physical control, biological control and chemical control strategies. The operation optimization planning module generates the optimal prevention and control operation path, pesticide dosage and natural enemy release ratio scheme.

5. The intelligent image recognition and control decision-making system for forestry tree diseases and pests according to claim 1, characterized in that: Preliminary identification of abnormal areas in forest areas is completed by using macroscopic remote sensing images from satellites and drones, generating preliminary diagnostic hypotheses and initial confidence levels for pests and diseases, retrieving microscopic images from the sensing layer and environmental sensor data as verification evidence, updating the diagnostic confidence level through Bayesian formula, dynamically adjusting the fusion weights of macroscopic and microscopic data, and outputting highly reliable pest and disease diagnosis results and damage level ratings, effectively reducing the misjudgment rate and missed judgment rate of single image recognition.

Images