Visual monitoring method of shrimp-rice symbiosis based on computer vision

Abstract

The present invention discloses a computer vision-based shrimp-rice symbiosis visualization monitoring method, which relates to the field of smart agriculture technology. In order to solve the problem that the existing technology mainly relies on satellite remote sensing data, and lacks effective correlation between multi-dimensional data such as terrain, biomass, and water quality, and has significant deficiencies in real-time early warning response and resource optimization allocation, the present invention integrates multimodal data such as drone aerial photography, underwater laser scanning, and water quality sensors to establish rice biomass estimation, crayfish behavior recognition, and water quality prediction models. Combined with the terrain complexity level classification, a differentiated early warning system is constructed to achieve early identification of risks such as crayfish stress behavior and water quality mutations, simulate system responses under different scenarios, and generate three-dimensional visualization results to provide a scientific basis for dynamically adjusting management measures. The drone and fixed camera work together to achieve dynamic patrol and high-definition acquisition in abnormal areas, effectively improving the intelligent management level of the shrimp-rice symbiosis system.

Description

Technical Field

[0001] The present invention relates to the field of smart agricultural technology, and in particular to a shrimp-rice symbiosis visualization monitoring method based on computer vision. Background Art

[0002] As a highly efficient ecological agricultural model, shrimp-rice symbiosis achieves resource recycling and improved economic benefits through the coordinated development of rice cultivation and crayfish farming. Chinese patent application publication number CN115170977A discloses a method and system for visually monitoring shrimp-rice symbiosis based on multispectral imagery. The method includes: image acquisition and preprocessing, using multispectral satellites to acquire images with cloud cover below 30%, and masking the acquired images; feature calculation and time series analysis, selecting the normalized vegetation index as the index for distinguishing shrimp-rice symbiosis areas from other landforms, and fitting them using time series harmonic analysis to remove contamination points; selecting typical time phases for RGB image synthesis, and highlighting the color of the shrimp-rice symbiosis area in the multispectral imagery through the time series fitting results and typical time phase features; and monitoring the prominent colors of the shrimp-rice symbiosis area in the multispectral imagery. This method solves the problem that existing shrimp-rice symbiosis areas are difficult to accurately distinguish from other surrounding areas.

[0003] However, the current shrimp-rice symbiotic system still has the following problems:

[0004] Existing technologies mainly rely on satellite remote sensing data, which makes it difficult to comprehensively obtain information on the spatiotemporal changes in rice growth status, crayfish behavior, and water quality environment; and multi-dimensional data such as terrain, biomass, and water quality lack effective correlation, and cannot reveal the coupling mechanism within the system. There are significant deficiencies in real-time early warning response and resource optimization allocation. Summary of the Invention

[0005] The purpose of the present invention is to provide a computer vision-based shrimp-rice symbiosis visualization monitoring method, which realizes comprehensive monitoring and precise management of the shrimp-rice symbiosis system through multi-source data fusion and intelligent analysis technology, so as to solve the problems raised in the above background technology.

[0006] To achieve the above object, the present invention provides the following technical solutions:

[0007] The computer vision-based visualization monitoring method for shrimp-rice symbiosis includes:

[0008] The image acquisition device collects the growth image data of shrimp and rice in the monitoring area in real time, performs target feature recognition based on the growth image data, and judges the growth status of shrimp and rice based on the recognition results;

[0009] Establish a water quality parameter prediction model, evaluate water quality indicators in the monitoring area, and combine statistical analysis of the identification results to predict the symbiotic trend of shrimp and rice in the future, including:

[0010] Based on the aerial image data obtained by UAV, the basic topography of the monitoring area is obtained, the number of geometric units required to cover the terrain at different scales is measured, the roughness of the terrain surface is quantified, and the hydrological characteristics are extracted;

[0011] The weight of the terrain factors is determined based on the basic terrain and hydrological characteristics of the monitoring area to generate a comprehensive terrain complexity index;

[0012] Based on the distribution characteristics of the comprehensive terrain complexity index, the monitoring area is divided into four levels, including very low complexity, low complexity, medium complexity and high complexity;

[0013] Based on the classification results of the monitoring areas, the terrain complexity data is spatially matched with biomass and water quality parameters and time series analysis is carried out to determine the association rules between terrain complexity and biomass fluctuations and water quality mutations, and to build a differentiated monitoring strategy. Specifically, in high-biomass areas, high-density visual sensor nodes are deployed to implement high-frequency monitoring every 10 minutes. In complex terrain areas, multispectral imagers and micro-meteorological stations are configured to strengthen the coordinated monitoring of environmental parameters. High-risk periods are predicted based on historical data, monitoring resources are intelligently allocated, and a terrain complexity-warning threshold mapping table is built to automatically adjust the warning thresholds of water quality and biomass indicators according to the terrain complexity level.

[0014] The prediction of the symbiotic trend of shrimp and rice over a period of time in the future also includes: calculating a health index of the monitoring area based on terrain complexity data, rice biomass value, and crayfish density data, correlating it with water quality parameters output by a water quality parameter prediction model, and quantitatively evaluating the intensity of the symbiotic effect in different monitoring areas;

[0015] The symbiotic effect intensity index is obtained by the following formula:

[0016]

[0017] Among them, P is the symbiotic effect intensity index, B i is the rice biomass value, W i is the water quality index, A i Crayfish density data, T i is the comprehensive index of terrain complexity, ω i is the weight coefficient, and ∈ is the smoothing factor.

[0018] Furthermore, the monitoring area specifically includes:

[0019] Monitoring area division: Based on the actual topography and area of ​​the shrimp-rice symbiotic field, divide it into no less than four monitoring sub-areas, number each monitoring sub-area, and determine the boundary coordinates of each monitoring sub-area;

[0020] Monitoring equipment installation: Determine the type and quantity of equipment for each monitoring sub-area based on the location characteristics of each monitoring sub-area, including the location and number of cameras, the location and number of lighting equipment, and the number and location of underwater laser scanning arrays;

[0021] Data network construction: Based on the wireless network, a data transmission channel is established for the monitoring equipment in each monitoring sub-area, a monitoring sensor network is built, and data sharing and real-time data transmission are carried out based on the monitoring sensor network.

[0022] Furthermore, the growth image data further includes:

[0023] Based on the low-altitude flight data collected by drones, the monitoring equipment carried by the drones collects images according to the predetermined shooting frequency to obtain remote sensing image data of the rice fields;

[0024] When the fixed camera detects abnormal growth in a local area, the drone generates a cruising path based on the boundary coordinates of the corresponding monitoring area and adjusts the shooting frequency and camera resolution.

[0025] The aerial image data obtained by the drone is integrated with the monitoring image data obtained by the monitoring equipment to generate growth image data of shrimp and rice in the monitoring area.

[0026] Furthermore, the target feature recognition specifically includes:

[0027] Extract image data of rice at different growth stages based on growth image data, and extract color features, texture features, and shape features of rice;

[0028] Establish the relationship between rice color, texture and shape characteristics and biomass, estimate rice biomass, and obtain the corresponding biomass value for each monitoring area;

[0029] The model expression for estimating rice biomass is:

[0030] y=β0+β1x1+β2x2+...+β m x m +α

[0031] Among them, y is the biomass, x1, x2..., x m are rice characteristics, β0, β1..., β m is the model coefficient, α is the error term;

[0032] Acquire underwater three-dimensional point cloud data of the monitoring area through the deployed underwater laser scanning array;

[0033] A three-dimensional model of the crayfish habitat is generated based on the underwater three-dimensional point cloud data, and the number of crayfish per unit area is determined by combining the monitoring image data obtained by the monitoring equipment in each monitoring sub-area.

[0034] Furthermore, the number of crayfish per unit area is determined, including:

[0035] The obtained underwater three-dimensional point cloud data is preprocessed, and the surface of the preprocessed three-dimensional point cloud data is reconstructed to convert the discrete point cloud data into a continuous surface model to obtain a three-dimensional model of the crayfish habitat;

[0036] Preprocess the acquired monitoring image data, and perform crayfish target detection on the preprocessed video images;

[0037] The three-dimensional model of the crayfish habitat is spatially registered with the monitoring image data to determine the three-dimensional spatial area corresponding to each video image, and the number of crayfish detected in the area is counted. The number of crayfish per unit area is determined based on the area of ​​the area.

[0038] Furthermore, the number of crayfish detected in the statistical area also includes:

[0039] Extract the biological characteristics of each crayfish from the monitoring image data and establish a unique identifier for each crayfish;

[0040] Establishing a time axis according to the data length of the collected monitoring image data, performing frame processing on the monitoring image data, determining the initial position corresponding to each crayfish, determining the actual position of each initial position in the three-dimensional space area, and analyzing the position characteristics of the crayfish;

[0041] Track the movement position of each crayfish in the continuous frames, and establish the movement feature of the crayfish based on the initial position. At the same time, extract the data proportion of each crayfish in the continuous frames to generate the pause feature of each crayfish.

[0042] Based on the movement characteristics of each crayfish and the corresponding pause characteristics, a behavioral characteristic corresponding to each crayfish is obtained, and based on the behavioral characteristic, whether the crayfish is in a stress behavior state is determined;

[0043] Among them, whether the crayfish is in a stress behavior state is determined by the comprehensive stress behavior judgment index, and the formula is as follows:

[0044] S=ω1×v avg +ω2×a max +ω3×Δθ avg

[0045] Among them, ω1, ω2, ω3 are weight coefficients, and ω1+ω2+ω3=1, v avg is the average speed of crayfish, a max is the maximum acceleration, Δθ avg is the average change of attitude angle within a certain period of time;

[0046] When the comprehensive stress behavior judgment index exceeds a preset threshold, the crayfish is determined to be in a stress behavior state.

[0047] Furthermore, the target feature recognition further includes:

[0048] Based on the aerial image data obtained by UAV, the basic topography of the monitoring area is obtained, the number of geometric units required to cover the terrain at different scales is measured, the roughness of the terrain surface is quantified, and the hydrological characteristics are extracted;

[0049] The weight of the terrain factors is determined based on the basic terrain and hydrological characteristics of the monitoring area to generate a comprehensive terrain complexity index;

[0050] The monitoring area is divided into four levels based on the distribution characteristics of the comprehensive terrain complexity index, including extremely low complexity, low complexity, medium complexity and high complexity.

[0051] Compared with the prior art, the present invention has the following beneficial effects:

[0052] By integrating multimodal data such as drone aerial photography, underwater laser scanning, and water quality sensors, we have broken through the limitations of traditional single-indicator monitoring, established rice biomass estimation, crayfish behavior identification, and water quality prediction models, and combined with the terrain complexity level division to build a differentiated early warning system to achieve early identification of risks such as crayfish stress behavior and water quality mutations. We simulate the system response under different scenarios and generate three-dimensional visualization results to provide a scientific basis for the dynamic adjustment of management measures. Drones and fixed cameras work together to achieve dynamic cruising and high-definition acquisition in abnormal areas, ensuring that there are no blind spots in monitoring, effectively improving the intelligent management level of the shrimp-rice symbiosis system. BRIEF DESCRIPTION OF THE DRAWINGS

[0053] Figure 1 This is a flow chart of the computer vision-based shrimp-rice symbiosis visualization monitoring method of the present invention. DETAILED DESCRIPTION

[0054] The following will clearly and completely describe the technical solutions in the embodiments of the present invention in conjunction with the accompanying drawings. Obviously, the described embodiments are only part of the embodiments of the present invention, not all of the embodiments. Based on the embodiments of the present invention, all other embodiments obtained by ordinary technicians in this field without making creative efforts are within the scope of protection of the present invention.

[0055] In order to solve the technical problems that existing technologies mainly rely on satellite remote sensing data, lack effective correlation between multi-dimensional data such as terrain, biomass, and water quality, and have significant deficiencies in real-time early warning response and resource optimization allocation, please refer to Figure 1 , this embodiment provides the following technical solutions:

[0056] The computer vision-based visualization monitoring method for shrimp-rice symbiosis includes:

[0057] The image acquisition device collects the growth image data of shrimp and rice in the monitoring area in real time, performs target feature recognition based on the growth image data, and judges the growth status of shrimp and rice based on the recognition results;

[0058] A water quality parameter prediction model was established to assess water quality indicators within the monitoring area. Based on statistical analysis of the identification results, the responses of shrimp-rice symbiosis to different scenarios were simulated based on four typical scenarios: normal growth, extreme weather, pest and disease outbreaks, and management intervention. Using 3D dynamic visualization technology, the simulation results were displayed in real time, including biomass change curves, crayfish distribution heat maps, and water quality parameter fluctuation cloud maps.

[0059] Predict the future trend of shrimp and rice symbiosis. Calculate the health index of the monitored area based on terrain complexity data, rice biomass values, and crayfish density data. Correlate this with the water quality parameters output by the water quality parameter prediction model to quantitatively assess the intensity of the symbiotic effect in different monitored areas.

[0060] The symbiotic effect intensity index is obtained by the following formula:

[0061]

[0062] Among them, P is the symbiotic effect intensity index, B i is the rice biomass value, which is obtained by calculating the biomass y through the linear regression formula; W i A is a water quality indicator, which is collected by water quality sensors and optimized by water quality parameter prediction models, including dissolved oxygen, pH value, ammonia nitrogen content, etc., and is normalized to a value between 0 and 1 using weighted normalization processing; i Crayfish density data is obtained through underwater 3D point cloud model and image space registration method; T i is the comprehensive index of terrain complexity, with very low / low / medium / high complexity corresponding to values ​​1 / 2 / 3 / 4 respectively; ω i is the weight coefficient, ∈ is the smoothing factor, which is 0.01-0.1 and is used to avoid i Calculation errors caused by too small a value.

[0063] In this embodiment, for each monitoring sub-region, the single-region symbiotic effect component is first calculated, and then the components of all sub-regions are summed to obtain the symbiotic effect intensity index of the entire monitoring region, for example:

[0064] Substitute the formula to calculate P for each monitoring sub-area i , if a sub-region A i =15 tails / m 2 , B i =500kg / mu, W i =0.8, T i =2,ω i =0.25, then For all sub-regions P i The sum is obtained by P value. The higher the P value, the better the symbiotic state, indicating more coordinated shrimp-rice growth, better water quality, higher terrain adaptability, and stronger system stability. The lower the P value, the worse the symbiotic state, indicating that there are significant contradictions in the system and urgent adjustments are needed. The time series changes of the P value can intuitively reflect the symbiotic trend, and combined with association rules, provide a basis for differentiated management. If the P value continues to decline and the main contribution comes from areas with high terrain complexity, then priority should be given to optimizing irrigation or improving water quality in this area.

[0065] In this embodiment, by integrating multi-dimensional data such as terrain complexity, biomass, water quality, and crayfish density, comprehensive dynamic monitoring of the shrimp-rice symbiosis system is achieved, breaking through the limitations of traditional single indicator monitoring, and establishing a rice biomass estimation model, a crayfish behavior recognition model, and a water quality parameter prediction model. Combined with the terrain complexity level division, a differentiated early warning threshold system is constructed. Through correlation analysis of the coupling relationship between terrain, biology, and environment, early identification and accurate early warning of risks such as crayfish stress behavior and water quality mutation are achieved, and three-dimensional visualization of biomass change curves, crayfish distribution heat maps, and water quality fluctuation cloud maps are generated to provide data support for the scientific formulation of feeding strategies, water change frequency, and other management measures.

[0066] In this embodiment, the monitoring area specifically includes:

[0067] Monitoring area division: Based on the actual topography and area of ​​the shrimp-rice symbiotic field, divide it into no fewer than four monitoring sub-areas and number each sub-area. At the same time, determine the boundary coordinates of each sub-area to ensure that the subsequently collected images can accurately correspond to the specific location;

[0068] Monitoring equipment installation: The type and quantity of equipment for each monitoring sub-area will be determined based on its locational characteristics. This includes determining the location and number of cameras and lighting equipment, as well as the number and location of underwater laser scanning arrays. For example, panoramic cameras will be installed high up at the edge of the field for overall monitoring, while close-up cameras will be installed inside the rice field near crayfish activity areas and densely populated rice fields. Lighting equipment will automatically turn on and off based on ambient light conditions to ensure clear images at night or in low light conditions.

[0069] Data network construction: Based on the wireless network, a data transmission channel is established for the monitoring equipment in each monitoring sub-area, a monitoring sensor network is built, and data sharing and real-time data transmission are carried out based on the monitoring sensor network.

[0070] In this embodiment, the growth image data further includes:

[0071] Based on the low-altitude flight data collected by drones, the monitoring equipment carried by the drones collects images according to the predetermined shooting frequency to obtain remote sensing image data of the rice fields;

[0072] When the fixed camera detects abnormal growth in a local area, the drone generates a cruising path based on the boundary coordinates of the corresponding monitoring area. At the same time, it adjusts the shooting frequency and camera resolution. For example, according to the abnormality level (level 1-5), the drone dynamically adjusts the resolution from 5cm / pixel (normal) to 0.8cm / pixel (highest), the frame rate from 2fps (normal) to 15fps (highest), and the flight altitude from 50m (normal) to 10m (emergency).

[0073] The aerial image data obtained by the drone is integrated with the monitoring image data obtained by the monitoring equipment to generate growth image data of shrimp and rice in the monitoring area.

[0074] In this embodiment, an efficient and comprehensive monitoring system is constructed by finely dividing the monitoring areas and installing equipment in the shrimp-rice symbiotic fields. Real-time data transmission and sharing are realized by combining wireless networks, ensuring accurate positioning and real-time response of monitoring. Remote sensing images are obtained by drones, providing macro images of the rice fields, supplementing regional information that cannot be monitored by fixed cameras. Drones can dynamically adjust shooting parameters and provide detailed images with high resolution and high frame rate, thereby enhancing emergency response capabilities. This not only improves the level of intelligent agricultural management, but also provides a scientific basis for agricultural production, achieves the goal of precision agricultural management, and significantly improves the efficiency and quality of agricultural production.

[0075] In this embodiment, the target feature recognition specifically includes:

[0076] Based on growth image data, image data of rice at different growth stages are extracted, and color features (such as mean and standard deviation in RGB color space), texture features (such as contrast, correlation, energy and homogeneity) and shape features of rice are extracted to reflect the morphology and growth status of rice plants.

[0077] Establish the relationship between rice color, texture and shape characteristics and biomass, estimate rice biomass, and obtain the corresponding biomass value for each monitoring area;

[0078] The model expression for estimating rice biomass is:

[0079] y=β0+β1x1+β2x2+...+β m x m +α

[0080] Among them, y is the biomass, x1, x2..., x m are rice characteristics, β0, β1..., β m is the model coefficient, and α is the error term. The model coefficient is solved by the least square method and other methods to obtain the biomass value of each monitoring area. By calculating the statistical parameters such as the mean value and standard deviation of the biomass, the distribution pattern and change trend of the biomass are analyzed.

[0081] Acquire underwater three-dimensional point cloud data of the monitoring area through the deployed underwater laser scanning array;

[0082] A three-dimensional model of the crayfish habitat is generated based on underwater three-dimensional point cloud data. The number of crayfish per unit area is determined by combining the monitoring image data obtained by the monitoring equipment in each monitoring sub-area. The normal activity and stress behavior states are distinguished through the behavioral pattern recognition algorithm.

[0083] In this embodiment, the color, texture, and shape characteristics of rice are extracted to reflect the morphology and growth status of rice plants. The relationship between these characteristics and biomass is then established, and rice biomass is estimated. This improves the precision and accuracy of crop growth monitoring and promotes scientific and refined farmland management.

[0084] In this embodiment, determining the number of crayfish per unit area specifically includes:

[0085] The obtained underwater three-dimensional point cloud data is preprocessed, and the surface of the preprocessed three-dimensional point cloud data is reconstructed to convert the discrete point cloud data into a continuous surface model to obtain a three-dimensional model of the crayfish habitat;

[0086] Preprocess the acquired monitoring image data, such as grayscale conversion and histogram equalization, to enhance image contrast and clarity for easier target detection, and perform crayfish target detection on the preprocessed video images;

[0087] The three-dimensional model of the crayfish habitat is spatially registered with the monitoring image data to determine the three-dimensional spatial area corresponding to each video image, and the number of crayfish detected in the area is counted. The number of crayfish per unit area is determined based on the area of ​​the area.

[0088] In this embodiment, counting the number of crayfish detected in the area also includes:

[0089] Extract the biological characteristics of each crayfish from the monitoring image data and establish a unique identifier for each crayfish;

[0090] Establishing a time axis according to the data length of the collected monitoring image data, performing frame processing on the monitoring image data, determining the initial position corresponding to each crayfish, determining the actual position of each initial position in the three-dimensional space area, and analyzing the position characteristics of the crayfish;

[0091] Track the movement position of each crayfish in the continuous frames, and establish the movement feature of the crayfish based on the initial position. At the same time, extract the data proportion of each crayfish in the continuous frames to generate the pause feature of each crayfish.

[0092] Based on the movement characteristics of each crayfish and the corresponding pause characteristics, the behavioral characteristics corresponding to each crayfish are obtained, and based on the behavioral characteristics, whether the crayfish is in a stress behavior state is determined to achieve accurate tracking of the individual crayfish trajectory;

[0093] Among them, whether the crayfish is in a stress behavior state is determined by the comprehensive stress behavior judgment index, and the formula is as follows:

[0094] S=ω1×v avg +ω2×a max +ω3×Δθ avg

[0095] Among them, ω1, ω2, ω3 are weight coefficients, and ω1+ω2+ω3=1, v avg is the average speed of crayfish, a max is the maximum acceleration, Δθ avg It is the average change of posture angle within a certain period of time. When the comprehensive judgment index of stress behavior exceeds the preset threshold, the crayfish is judged to be in a stress behavior state.

[0096] In this embodiment, by preprocessing underwater three-dimensional point cloud data and performing surface reconstruction, combined with the preprocessing of monitoring images and crayfish target detection, an accurate estimation of the number of crayfish per unit area is achieved; at the same time, by extracting the biological characteristics of crayfish, establishing a unique identifier, analyzing position and movement characteristics, and generating pause characteristics, the behavior patterns of crayfish can be finely analyzed and the stress behavior state can be intelligently determined, thereby providing a scientific basis for crayfish farming, timely discovering abnormal behavior, reducing losses, improving farming benefits, and overall improving the intelligence level of farming management and production efficiency.

[0097] In this embodiment, the target feature recognition further includes:

[0098] Based on the aerial image data obtained by UAVs, the basic topography of the monitoring area is obtained, including slope, aspect, and undulation. The number of geometric units required to cover the terrain at different scales is measured, the roughness of the terrain surface is quantified, and hydrological characteristics are extracted.

[0099] Based on the basic topography and hydrological characteristics of the monitoring area, the weights of terrain factors such as slope, aspect, and terrain relief extracted from DEM data were determined (slope: 0.35, relief: 0.25, roughness: 0.20, other factors: 0.20) to generate a comprehensive terrain complexity index;

[0100] The monitoring area is divided into four levels based on the distribution characteristics of the comprehensive terrain complexity index, including extremely low complexity, low complexity, medium complexity and high complexity.

[0101] Based on the categorization of monitoring areas, spatial matching and time series analysis were performed between terrain complexity data and biomass and water quality parameters. Correlations between terrain complexity and biomass fluctuations and water quality mutations were identified. For example, a strong correlation rule was found: "High terrain complexity + low dissolved oxygen → a 37% increase in the probability of crayfish stress behavior." This led to the development of differentiated monitoring strategies. High-biomass areas deployed a high-density visual sensor node and implemented high-frequency monitoring every 10 minutes. Complex terrain areas deployed multispectral imagers and micro-meteorological stations to strengthen coordinated monitoring of environmental parameters. High-risk periods were predicted based on historical data, and monitoring resources were intelligently allocated. A mapping table was constructed between terrain complexity and early warning thresholds, automatically adjusting early warning thresholds for indicators such as water quality and biomass based on the terrain complexity level. Stricter threshold standards were applied to complex terrain areas, while those in simpler areas were relaxed.

[0102] In this example, basic terrain information for the monitored area is obtained through drone aerial imagery data. The surface roughness and hydrological characteristics of the terrain are quantified, generating a comprehensive terrain complexity index that categorizes the monitored area into different levels. By combining spatial matching of terrain complexity with biomass and water quality parameters and time series analysis, the correlation rules between terrain complexity and biomass fluctuations and water quality mutations are identified. Based on this, a differentiated monitoring strategy is constructed, enabling intelligent allocation of monitoring resources and precise control of early warnings. This effectively improves the relevance and efficiency of monitoring, reduces aquaculture risks, and enhances the intelligent level of agricultural management.

[0103] The above description is only a preferred specific embodiment of the present invention, but the scope of protection of the present invention is not limited thereto. Any technician familiar with the technical field, within the technical scope disclosed by the present invention, who makes equivalent replacements or changes based on the technical solution and inventive concept of the present invention, should be covered by the scope of protection of the present invention.

Claims

1. A computer vision-based visual monitoring method for shrimp-rice symbiosis, characterized in that: include: The image acquisition device collects the growth image data of shrimp and rice in the monitoring area in real time, performs target feature recognition based on the growth image data, and judges the growth status of shrimp and rice based on the recognition results; Establish a water quality parameter prediction model, evaluate water quality indicators in the monitoring area, and combine statistical analysis of the identification results to predict the symbiotic trend of shrimp and rice in the future, including: Based on the aerial image data obtained by UAV, the basic topography of the monitoring area is obtained, the number of geometric units required to cover the terrain at different scales is measured, the roughness of the terrain surface is quantified, and the hydrological characteristics are extracted; The weight of the terrain factors is determined based on the basic terrain and hydrological characteristics of the monitoring area to generate a comprehensive terrain complexity index; Based on the distribution characteristics of the comprehensive terrain complexity index, the monitoring area is divided into four levels, including very low complexity, low complexity, medium complexity and high complexity; Based on the classification results of the monitoring areas, the terrain complexity data is spatially matched with biomass and water quality parameters and time series analysis is carried out to determine the association rules between terrain complexity and biomass fluctuations and water quality mutations, and to build a differentiated monitoring strategy. Specifically, in high-biomass areas, high-density visual sensor nodes are deployed to implement high-frequency monitoring every 10 minutes. In complex terrain areas, multispectral imagers and micro-meteorological stations are configured to strengthen the coordinated monitoring of environmental parameters. High-risk periods are predicted based on historical data, monitoring resources are intelligently allocated, and a terrain complexity-warning threshold mapping table is built to automatically adjust the warning thresholds of water quality and biomass indicators according to the terrain complexity level. The prediction of the symbiotic trend of shrimp and rice over a period of time in the future also includes: calculating a health index of the monitoring area based on terrain complexity data, rice biomass value, and crayfish density data, correlating it with water quality parameters output by a water quality parameter prediction model, and quantitatively evaluating the intensity of the symbiotic effect in different monitoring areas; The symbiotic effect intensity index is obtained by the following formula: Among them, P is the symbiotic effect intensity index, B i is the rice biomass value, W i is the water quality index, A i Crayfish density data, T i is the comprehensive index of terrain complexity, ω i is the weight coefficient, and ∈ is the smoothing factor.

2. The computer vision-based shrimp-rice symbiosis visualization monitoring method according to claim 1, characterized in that: The monitoring area specifically includes: Monitoring area division: Based on the actual topography and area of ​​the shrimp-rice symbiotic field, divide it into no less than four monitoring sub-areas, number each monitoring sub-area, and determine the boundary coordinates of each monitoring sub-area; Monitoring equipment installation: Determine the type and quantity of equipment for each monitoring sub-area based on the location characteristics of each monitoring sub-area, including the location and number of cameras, the location and number of lighting equipment, and the number and location of underwater laser scanning arrays; Data network construction: Based on the wireless network, a data transmission channel is established for the monitoring equipment in each monitoring sub-area, a monitoring sensor network is built, and data sharing and real-time data transmission are carried out based on the monitoring sensor network.

3. The computer vision-based shrimp-rice symbiosis visualization monitoring method according to claim 1, characterized in that: The growth image data further includes: Based on the low-altitude flight data collected by drones, the monitoring equipment carried by the drones collects images according to the predetermined shooting frequency to obtain remote sensing image data of the rice fields; When the fixed camera detects abnormal growth in a local area, the drone generates a cruising path based on the boundary coordinates of the corresponding monitoring area and adjusts the shooting frequency and camera resolution. The aerial image data obtained by the drone is integrated with the monitoring image data obtained by the monitoring equipment to generate growth image data of shrimp and rice in the monitoring area.

4. The computer vision-based shrimp-rice symbiosis visualization monitoring method according to claim 3, characterized in that: The target feature recognition specifically includes: Extract image data of rice at different growth stages based on growth image data, and extract color features, texture features, and shape features of rice; Establish the relationship between rice color, texture and shape characteristics and biomass, estimate rice biomass, and obtain the corresponding biomass value for each monitoring area; Acquire underwater three-dimensional point cloud data of the monitoring area through the deployed underwater laser scanning array; A three-dimensional model of the crayfish habitat is generated based on the underwater three-dimensional point cloud data, and the number of crayfish per unit area is determined by combining the monitoring image data obtained by the monitoring equipment in each monitoring sub-area.

5. The computer vision-based shrimp-rice symbiosis visualization monitoring method according to claim 4, characterized in that: Determine the number of crayfish per unit area, including: The obtained underwater three-dimensional point cloud data is preprocessed, and the surface of the preprocessed three-dimensional point cloud data is reconstructed to convert the discrete point cloud data into a continuous surface model to obtain a three-dimensional model of the crayfish habitat; Preprocess the acquired monitoring image data, and perform crayfish target detection on the preprocessed video images; The three-dimensional model of the crayfish habitat is spatially registered with the monitoring image data to determine the three-dimensional spatial area corresponding to each video image, and the number of crayfish detected in the area is counted. The number of crayfish per unit area is determined based on the area of ​​the area.

6. The computer vision-based shrimp-rice symbiosis visualization monitoring method according to claim 5, characterized in that: The number of crayfish detected in the statistical area also includes: Extract the biological characteristics of each crayfish from the monitoring image data and establish a unique identifier for each crayfish; Establishing a time axis according to the data length of the collected monitoring image data, performing frame processing on the monitoring image data, determining the initial position corresponding to each crayfish, determining the actual position of each initial position in the three-dimensional space area, and analyzing the position characteristics of the crayfish; Track the movement position of each crayfish in the continuous frames, and establish the movement feature of the crayfish based on the initial position. At the same time, extract the data proportion of each crayfish in the continuous frames to generate the pause feature of each crayfish. Based on the movement characteristics of each crayfish and the corresponding pause characteristics, the behavioral characteristics corresponding to each crayfish are obtained, and based on the behavioral characteristics, it is determined whether the crayfish is in a stress behavior state.

7. The computer vision-based shrimp-rice symbiosis visualization monitoring method according to claim 1, characterized in that: Also includes: Based on the association rules between terrain complexity and biomass fluctuations and water quality mutations, the shrimp-rice symbiotic responses under different scenarios are simulated. Based on three-dimensional dynamic visualization technology, the simulation results are displayed in real time, including biomass change curves, crayfish distribution heat maps, and water quality parameter fluctuation cloud maps.

Images