Intelligent analysis system integrating fish body parameter measurement and disease diagnosis
The real-time data acquisition and intelligent analysis system has solved the problems of efficiency and accuracy in fish growth monitoring and disease diagnosis, realizing integrated intelligent management of fish parameter measurement and disease diagnosis, and improving the scientific nature and efficiency of aquaculture.
Patent Information
- Application Number
- CN202511896720.5
- Authority / Receiving Office
- CN · China
- Patent Type
- Applications(China)
- Current Assignee / Owner
- Filing Date
- 2025-12-16
- Publication Date
- 2026-03-17
AI Technical Summary
Current technologies rely on human experience for monitoring fish growth, resulting in low measurement efficiency and inaccuracy. Diagnosis of fish diseases also relies heavily on subjective experience, making it difficult to detect diseases in a timely manner and impacting aquaculture efficiency.
The system employs a data acquisition module to collect real-time data on the fish's environment and images/videos, a fish measurement module to calculate biological characteristic parameters, a disease diagnosis module for automatic diagnosis, and a fusion analysis module for comprehensive evaluation and early warning decision-making, thereby achieving integrated intelligent analysis of fish parameter measurement and disease diagnosis.
It improves the accuracy of fish body parameter measurement and the efficiency of disease diagnosis, realizes the scientific management of fish health status, reduces human intervention, and improves aquaculture efficiency.
Smart Images
Figure CN121685495A_ABST
Abstract
Description
TECHNICAL FIELD
[0001] The present application relates to the technical field of aquaculture, and particularly relates to an intelligent analysis system integrating fish body parameter measurement and disease diagnosis. BACKGROUND
[0002] Under the background of the development of aquaculture intensification, fish body growth monitoring mainly relies on manual experience for operation, and simple measuring tools are used to measure the fish body. This measurement method is not only inefficient and inaccurate in measurement accuracy, but also increases the work burden of the breeding personnel, and is easy to cause stress reaction to the fish body.
[0003] In terms of fish disease diagnosis, the traditional method mainly relies on the breeding experience of the breeding personnel, and judges whether the fish is diseased by observing the fish body. However, this diagnosis method is subjective, and the diagnosis result is easily affected by subjective factors. Moreover, when the initial symptoms of the fish disease are not obvious, it is difficult to find out in time, which delays the best treatment opportunity and affects the breeding benefit. SUMMARY
[0004] In view of this, the purpose of the present application is to provide an intelligent analysis system integrating fish body parameter measurement and disease diagnosis to solve or at least partially solve the above problems existing in the prior art.
[0005] To achieve the above-mentioned purpose, the present application provides an intelligent analysis system integrating fish body parameter measurement and disease diagnosis, which comprises: a data acquisition module for acquiring real-time aquaculture water environment data and fish image video data; a fish body measurement module for separating fish body moving targets and static backgrounds according to the fish image video data, extracting fish body contours according to the fish body moving targets and static backgrounds, and calculating fish biological feature parameters, wherein the fish biological feature parameters include body length, body width and fatness; a disease diagnosis module for automatically diagnosing and identifying fish diseases according to the fish image video data, positioning the disease area, and calculating the disease area ratio and disease distribution density of the fish body; a fusion analysis module for comprehensively analyzing the fish body measurement module, the disease diagnosis module and the water environment data to realize the evaluation and early warning decision of the fish health status.
[0006] Further, the data acquisition module acquires data in real time through a high-definition industrial camera, a water temperature sensor, a pH sensor and a dissolved oxygen sensor.
[0007] Further, the fish body moving targets and the static backgrounds are separated according to the fish image video data, which specifically comprises the following steps: S11, pre-processing the fish image video data; S12, all pixel points of the fish image video data subjected to the preprocessing operation are acquired, and sample values are randomly selected from the pixel point neighborhood to construct a background model; S13, the pixels of the fish image video data subjected to the preprocessing operation are acquired, the distance between the pixel value and the sample in the background model is calculated, if the distance of the preset sample number is less than the preset distance threshold, it is determined as a static background, otherwise as a fish body moving target.
[0008] Further, the fish body contour is extracted according to the fish body moving target and the static background, specifically including the following steps: S21, based on the fish body moving target and the static background, an edge detection algorithm is used to extract the candidate edge of the fish body by using a double threshold, and a candidate edge point set is constructed; S22, the axis and the axis coordinates of the candidate edge point set are acquired, and the point with the minimum axis coordinate is taken as the pole point; S23, the polar angle of the remaining points on the axis and the axis relative to the pole point is calculated, and the polar angle is sorted in ascending order; S24, based on the polar angle sorted in ascending order, the fish body convex hull is calculated, and the contour concave region is filled to form a complete closed contour; S25, the contour area and the aspect ratio are acquired, and the final fish body contour is selected according to the area and the aspect ratio.
[0009] Further, the fish biological feature parameters are calculated and represented as follows:
[0010]
[0011]
[0012] wherein, is the body length, is the body width, is the fatness, is the pixel length of the main axis of the fish body contour, is the maximum width pixel number perpendicular to the main axis of the fish body contour, is the fish body pixel-physical conversion coefficient.
[0013] Further, the disease diagnosis module specifically includes the following steps: S31, the collected fish image video data is quality checked, and the data passing the quality check is subjected to a preprocessing operation and input into a deep learning model; S32, based on the deep learning model, an adaptive attention mechanism is introduced, and the adaptive attention mechanism is used to enhance the feature extraction capability of the fish disease area; S33, using the deep learning model, multi-scale features of the fish disease area are extracted through different levels of feature pyramid, and the multi-scale features are fully fused through the feature pyramid network and the path aggregation network, the multi-scale features include high-level features and low-level features; S34, based on the fully fused multi-scale features, the prediction head of the deep learning model is used to predict at different scales respectively, and the prediction head output is filtered to retain the best prediction result, the prediction head output includes boundary box coordinates, disease confidence and disease category probability; S35, the existing farmed fish disease data is obtained and integrated into a fish disease dataset, the statistical characteristics of the dataset are analyzed, the size and proportion of the preset anchor box are optimized according to the statistical characteristics, and the clustering algorithm is used to analyze the fish disease area size in the dataset to generate the final fish disease detection anchor box; S36, the offset of the boundary box coordinates output by step S34 relative to the fish disease detection anchor box is calculated, so that the boundary box coordinates are aligned with the fish disease detection anchor box, and the disease area of the fish is accurately positioned.
[0014] Further, the calculation of the disease area ratio of the fish includes: Based on the boundary box coordinates aligned with the fish disease detection anchor box, the disease area is cropped from the fish image video data, and the percentage of the disease area to the fish surface area is calculated, which is expressed as follows:
[0015] Wherein, The percentage of the disease area to the fish surface area is The image area of the disease area is The image area of the fish is
[0016] Further, the calculation of the disease distribution density of the fish includes the following steps: S41, the fish area is divided into uniform grids according to the fish contour, and the total number of pixels of the fish area is obtained; S42, based on the divided uniform grid, the number of disease pixels in each grid unit is counted; S43, based on the number of disease pixels in each grid unit, the disease distribution density of the fish is calculated, the disease distribution density includes local density and global density variation coefficient, which is expressed as follows:
[0017]
[0018] wherein, is a local density, is a global density coefficient of variation, is a number of disease pixels, is a total number of pixels, is a local density average, is a local density standard deviation.
[0019] Further, the fusion analysis module specifically comprises the following steps: S51, aligning the fish biological feature parameters, the fish disease diagnosis results and the water environment data in time and space, so that different data have the same time stamp and spatial reference system; S52, based on the various types of data after time and space alignment, calculating the daily growth rate and the morphological change rate of the fish according to the fish biological feature parameters, calculating the disease expansion speed according to the fish disease diagnosis results, and calculating the environmental pH value gradient and the dissolved oxygen change rate according to the water environment data; S53, based on the change rates of various types of data calculated in step S52, analyzing the correlation between the fish biological feature parameters, the fish disease diagnosis results and the water environment data, and evaluating the health status of the fish according to the correlation.
[0020] Compared with the prior art, the beneficial effects of the present application are: The present application proposes an intelligent analysis system integrating fish parameter measurement and disease diagnosis, which realizes the integration and intelligentization of fish parameter measurement and disease diagnosis, improves the measurement accuracy and diagnosis efficiency, and provides strong support for the scientific management of aquaculture. BRIEF DESCRIPTION OF DRAWINGS
[0021] In order to more clearly illustrate the technical solutions in the embodiments of the present application, the following will briefly introduce the drawings needed in the embodiment description. Obviously, the drawings in the following description are only preferred embodiments of the present application, and other drawings can also be obtained by those skilled in the art without creative labor on the basis of these drawings.
[0022] Fig. 1 A structure principle diagram of an intelligent analysis system integrating fish parameter measurement and disease diagnosis provided by the present application embodiment; Fig. 2A main interface schematic diagram of an intelligent analysis system fusing fish body parameter measurement and disease diagnosis is provided for an embodiment of the present application. Fig. 3 An analysis flow schematic diagram of a fusion analysis module is provided for an embodiment of the present application. DETAILED DESCRIPTION
[0023] The principles and characteristics of the present application are described below in combination with the drawings, and the listed embodiments are only used to explain the present application and are not used to limit the scope of the present application.
[0024] Referring to Figs. 1-3 The present embodiment provides an intelligent analysis system fusing fish body parameter measurement and disease diagnosis, which comprises: A data acquisition module: used for acquiring real-time aquaculture water environment data and fish image video data, specifically comprising: The data acquisition module acquires data in real time through a high-definition industrial camera, a water temperature sensor, a pH sensor and a dissolved oxygen sensor, so as to ensure the comprehensiveness and accuracy of the data; The high-definition industrial camera can be a Hikvision DS-2CD3320D-1, which is used for acquiring fish image video data; the water temperature sensor can be a DS18B20 model, which is used for monitoring the temperature of the aquaculture water in real time; the pH sensor can be an analog pH sensor module, which is used for detecting the pH of the aquaculture water; and the dissolved oxygen sensor can be a JPSJ-605F model, which is used for monitoring the dissolved oxygen content of the water.
[0025] A fish body measurement module: used for separating fish body moving targets and static backgrounds according to fish image video data, extracting fish body contours according to the fish body moving targets and static backgrounds, and calculating fish biological characteristic parameters, wherein the fish biological characteristic parameters include body length, body width and fatness, and specifically comprising: The fish body moving targets and static backgrounds are separated according to the fish image video data, specifically comprising the following steps: S11, performing a preprocessing operation on the fish image video data, wherein the preprocessing operation includes gray scale conversion, Gaussian filter denoising and contrast enhancement, which are used for the same image video format, and reduce noise and uneven light interference; S12, obtaining all pixel points of the fish image video data after the preprocessing operation, and constructing a background model from sample values randomly selected from the pixel point neighborhood, which is expressed as follows:
[0026] Wherein, is the background model, represents the i-th sample, the pixel value (gray scale or RGB) of each sample, =20; S13. Obtain the pixels of the preprocessed fish image video data, and calculate the distance between the pixel values and the samples in the background model, as shown below:
[0027] in, These are the pixel values of the pre-processed fish image / video data; If at least m preset samples are less than the preset distance threshold R, they are judged as static background (such as aquatic plants, pond walls, etc.), otherwise they are fish moving targets. Generally, m=2 and R=20 are set. A conservative update strategy and spatial propagation mechanism are used to update the background model. Only pixels classified as static background have the opportunity to update the background model. The updated background model sample has a 1 / 16 probability of propagating to the background model of the adjacent pixels.
[0028] The process of extracting the fish contour based on the moving target of the fish and the static background specifically includes the following steps: S21. Based on the moving target of the fish and the static background, an edge detection algorithm is used to extract candidate edges of the fish using a dual threshold (low threshold = 50, high threshold = 150), and a candidate edge point set is constructed as follows:
[0029] in, For the candidate edge point set, The number of candidate edge points. , for The first on the axis Point coordinates, for The first on the axis Point coordinates; S22. Obtain the candidate edge point set. shaft and axis coordinates, and The point with the smallest y-axis coordinate is taken as the pole (if...). If the axis has multiple identical minimum points, then take... (the point with the smallest axial coordinates) S23, Calculation shaft and The remaining points on the axis are sorted relative to the polar angles of the poles, and ordered in ascending order of polar angles. A cross-product method is used for sorting optimization, as shown below:
[0030] in, Polar angle, for The coordinates of the origin on the axis for The coordinates of the origin on the axis; S24. Calculate the convex hull of the fish body based on the ascending polar angle sorting, and fill the concave area of the contour to form a complete closed contour. S25. Obtain the outline area and aspect ratio, and filter out the final fish outline based on its area and aspect ratio.
[0031] The calculation of fish biological characteristic parameters uses a checkerboard calibration board (5mm squares) to pre-calibrate the fish body pixels, and the fish body pixel-physical size conversion coefficient is expressed as follows:
[0032]
[0033]
[0034] in, For body length, For body width, For plumpness, The length of the main axis of the fish's body outline in pixels. The maximum width in pixels perpendicular to the main axis of the fish's outline. The fish body pixel-physical conversion factor (unit: mm / pixel).
[0035] Disease diagnosis module: Used to automatically diagnose and identify fish diseases based on fish image and video data, locate diseased areas, and calculate the proportion of diseased area and disease distribution density on the fish body. Specifically, it includes the following steps: S31. Perform quality verification on the collected fish image and video data, preprocess the data that passes the quality verification, and input it into the YOLOv11 deep learning model. The quality verification is to ensure the clarity and contrast of the collected fish image and video data. The clarity and contrast must meet the preset requirements. Fish image and video data that do not meet the preset requirements will be resampled or a quality warning will be issued. The preprocessing operations include size normalization (converting the data into tensor format and then normalizing it), contrast enhancement, and other operations to improve the robustness of subsequent model recognition. The deep learning model is built on the CSPDarknet53 backbone network and includes multiple convolutional layers, batch normalization layers, and activation function layers. S32. Based on a deep learning model, an adaptive attention mechanism is introduced to enhance the feature extraction capability of diseased areas in fish, as shown below:
[0036] in, For attention weights, It is the Sigmoid activation function. It is a 1×1 convolutional layer. For feature splicing operations, For the input feature map, For global average pooling; S33. Using a deep learning model, multi-scale features of diseased areas in fish are extracted through feature pyramids of different levels (FPN+PAN structure). The multi-scale features are fully fused through the Feature Pyramid Network (FPN) and the Path Convergence Network (PAN). The multi-scale features include high-level features and low-level features. The low-level features are used to retain high-resolution spatial information and are suitable for detecting simple disease information. The high-level features have rich semantic information and are suitable for detecting complex disease information. S34. Based on fully fused multi-scale features, the prediction head of the deep learning model is used to make predictions at three different scales (20×20, 40×40, 80×80). The prediction head output includes bounding box coordinates, disease confidence, and disease category probability. The prediction head output is filtered to remove low-confidence predictions, and the NMS algorithm is applied to eliminate redundant bounding boxes, retaining the best prediction results. The disease confidence is the probability that the bounding box contains the disease, and the disease category probability is obtained using an independent logistic regression classifier. S35. Obtain existing data on diseases of farmed fish and integrate them into a fish disease dataset. Analyze the statistical characteristics of the dataset, optimize the size and proportion of the preset anchor frame based on the statistical characteristics, and use a clustering algorithm to analyze the size of the fish disease area in the dataset to generate the final fish disease detection anchor frame. S36. Calculate the offset of the bounding box coordinates output in step S34 relative to the fish disease detection anchor frame, align the bounding box coordinates with the fish disease detection anchor frame, and accurately locate the diseased area of the fish. The offset is expressed as follows:
[0037] in, and These are the bounding boxes. shaft and The center coordinates of the axis This is the width of the bounding box. The height of the bounding box. and They are respectively shaft and Offset on the axis and They are respectively shaft and The coordinates of the top left corner of the grid cell on the axis and These represent the width and height of the anchor frame for detecting fish diseases. The Sigmoid function constrains the coordinates within the current grid. The width scaling factor is determined by the logarithmic spatial offset t. w We obtain this by performing exponential operations. The height scaling factor is determined by the logarithmic spatial offset t. h Obtained by performing exponentiation; and The purpose of this is to ensure positive definiteness of dimensions. Through transformation of the logarithmic space, the model can learn disease areas at different scales more effectively.
[0038] Based on the diagnostic identification results of fish diseases and the location of disease areas, a multi-dimensional disease analysis is performed. The multi-dimensional disease analysis includes the calculation of the disease area ratio and the analysis of disease distribution density. Based on the results of the multi-dimensional disease analysis and combined with a preset expert knowledge base, a diagnostic report is generated that includes disease type, severity level, treatment suggestions, etc. For high-risk cases, the system immediately issues an early warning. The calculation of the percentage of diseased area on the fish body includes: Based on the bounding box coordinates aligned with the anchor frame for fish disease detection, the diseased area is cropped from the fish image and video data, and the percentage of the diseased area relative to the fish's surface area is calculated, as shown below:
[0039] in, This represents the percentage of the fish's body surface area affected by disease. This represents the image area of the affected region. The image area represents the fish's body.
[0040] The calculation of disease distribution density in fish specifically includes the following steps: S41. Divide the fish body region into an N×N uniform grid (usually N=5) according to the fish body outline, and obtain the total number of pixels in the fish body region. S42. Based on the divided uniform grid, count the number of defective pixels in each grid cell; S43. Based on the number of diseased pixels in each grid cell, calculate the disease distribution density of the fish body. The disease distribution density includes the local density and the global density variation coefficient, expressed as follows:
[0041]
[0042] in, For local density, The global density variation coefficient. The number of pixels affected by the defect. This represents the total number of pixels. This represents the local density average. This represents the local density standard deviation.
[0043] Fusion Analysis Module: This module integrates data from the fish body measurement module, disease diagnosis module, and aquatic environment to comprehensively analyze fish health status and make early warning decisions. Specifically, it includes the following steps: S51. Spatial and temporal alignment of fish biological characteristic parameters, fish disease diagnosis results and aquatic environment data, so that different data have the same timestamp and spatial reference system. S52. Based on various types of data that have been aligned in time and space, calculate the daily growth rate and morphological change rate of fish based on fish biological characteristic parameters, calculate the disease spread rate based on fish disease diagnosis results, and calculate the environmental pH gradient and dissolved oxygen change rate based on aquatic environmental data. The spatiotemporal alignment includes time synchronization and spatial coordinate unification. The time synchronization is a unified processing of time windows, and the spatial coordinate unification includes establishing a three-dimensional coordinate system (length × width × water depth) for the aquaculture pond, using relative coordinates for fish positioning (with the fish center as the origin), and mapping the aquaculture water environment data to a physical coordinate grid.
[0044] S53. Based on the rate of change of various data obtained in step S52, analyze the correlation between fish biological characteristic parameters, fish disease diagnosis results, and aquatic environmental data, and score the health status of fish according to the correlation, as shown below: Health score = W1 × growth status score + W2 × disease status score + W3 × environmental adaptability score Among them, W1 is the weight of fish growth status, W2 is the weight of fish disease status, and W3 is the weight of environmental adaptation. An integrated rule-based reasoning engine is used to assess risk based on health scores and specific rules. Green and safe zone (score ≥ 80): Normal farming, regular monitoring; Yellow alert zone (score 60 ≤ rating < 80): Monitor changes and adjust management accordingly; Red Alert Zone (Score < 60): Immediate intervention and emergency treatment.
[0045] Based on the experience of aquaculture experts in the knowledge base, personalized prevention and control suggestions are generated according to the risk assessment results. A fish analysis report is also generated based on the risk assessment and the corresponding prevention and control suggestions. The prevention and control suggestions include water quality adjustment plans, feed feeding suggestions, disease treatment measures, and stocking density adjustments.
[0046] The above description is only a preferred embodiment of the present invention and is not intended to limit the present invention. Any modifications, equivalent substitutions, improvements, etc., made within the spirit and principles of the present invention should be included within the protection scope of the present invention.
Claims
1. An intelligent analysis system that integrates fish body parameter measurement and disease diagnosis, characterized in that, The system comprises: a data acquisition module for acquiring real-time aquaculture water environment data and fish image and video data; a fish body measurement module for separating fish body moving targets and static backgrounds from the fish image and video data, extracting fish body contours from the fish body moving targets and static backgrounds, and calculating fish biological characteristic parameters including body length, body width, and fatness; a disease diagnosis module for automatically diagnosing and identifying fish body diseases based on the fish image and video data, positioning disease areas, and calculating fish body disease area proportions and disease distribution densities; a fusion analysis module for comprehensively analyzing the fish body measurement module, the disease diagnosis module, and the water environment data to achieve fish body health state evaluation and early warning decision-making.
2. The intelligent analysis system for measuring fish body parameters and diagnosing diseases according to claim 1, wherein, The data acquisition module acquires real-time data through a high-definition industrial camera, a water temperature sensor, a pH sensor, and a dissolved oxygen sensor.
3. The intelligent analysis system for measuring fish body parameters and diagnosing diseases according to claim 1, wherein, The fish body moving targets and static backgrounds are separated from the fish image and video data, specifically including the following steps: S11, performing a preprocessing operation on the fish image and video data; S12, obtaining all pixel points of the fish image and video data after the preprocessing operation, randomly selecting sample values from the pixel point neighborhood to construct a background model; S13, obtaining pixels of the fish image and video data after the preprocessing operation, calculating the distance between the pixel values and the samples in the background model, and if the distance of the preset sample number is less than the preset distance threshold, determining it as a static background, otherwise as a fish body moving target.
4. The intelligent analysis system for fusion of fish body parameter measurement and disease diagnosis according to claim 3, characterized in that, The fish body contours are extracted from the fish body moving targets and static backgrounds, specifically including the following steps: S21, based on the fish body moving targets and static backgrounds, using an edge detection algorithm to extract candidate edges of the fish body with double thresholds, and constructing a candidate edge point set; S22, obtaining the candidate edge point set axis and axis coordinates, and taking the point with the minimum axis coordinates as the pole point; S23, calculating axis and the polar angle of the rest of the points on the axis relative to the polar point, and sorted in ascending order of polar angle; S24, calculating the fish body convex hull based on the polar angle ascending order sorting, and filling the contour concave area to form a complete closed contour; S25, obtaining the contour area and aspect ratio, and selecting the final fish body contour according to the area and aspect ratio.
5. The intelligent analysis system for measuring fish body parameters and diagnosing diseases according to claim 4, characterized in that, The fish biological characteristic parameters are calculated as follows: wherein, is the body length, is the body width, is the roundness, is the length of the main axis of the fish body profile in pixels, is the maximum width perpendicular to the main axis of the fish body profile in pixels, is the pixel-physical conversion factor of the fish body.
6. The intelligent analysis system for measuring fish body parameters and diagnosing diseases according to claim 1, wherein, The disease diagnosis module specifically includes the following steps: S31, performing quality verification on the collected fish image and video data, performing preprocessing operation on the data that passes the quality verification, and inputting it into a deep learning model; S32, based on the deep learning model, introducing an adaptive attention mechanism, and using the adaptive attention mechanism to enhance the feature extraction capability of the fish disease area; S33, using the deep learning model to extract multi-scale features of the fish disease area through different levels of feature pyramids, and fully fusing the multi-scale features through a feature pyramid network and a path aggregation network, the multi-scale features including high-level features and low-level features; S34, based on the fully fused multi-scale features, using the prediction head of the deep learning model to make predictions at different scales, filtering the prediction head output, and retaining the best prediction result, the prediction head output including bounding box coordinates, disease confidence, and disease category probability; S35, acquire existing fish disease data, integrate them into a fish disease dataset, analyze the statistical characteristics of the dataset, optimize the size and proportion of the preset anchor box according to the statistical characteristics, and analyze the size of the fish disease area in the dataset using a clustering algorithm to generate a final fish disease detection anchor box; S36, calculate the offset of the boundary box coordinates output in step S34 relative to the fish disease detection anchor box, align the boundary box coordinates with the fish disease detection anchor box, and accurately locate the disease area of the fish.
7. The intelligent analysis system for measuring fish body parameters and diagnosing diseases according to claim 6, wherein, The calculation of the disease area ratio of the fish includes: Based on the boundary box coordinates aligned with the fish disease detection anchor box, the disease area is cropped from the fish image video data, and the percentage of the disease area to the fish surface area is calculated, represented as follows: wherein, is the percentage of the fish body surface area occupied by the diseased area, is the image area of the diseased area, is the image area of the fish body.
8. The intelligent analysis system for fusion of fish parameter measurement and disease diagnosis according to claim 6, characterized in that, The calculation of the disease distribution density of the fish specifically includes the following steps: S41, divide the fish area into uniform grids according to the fish contour, and obtain the total number of pixels in the fish area; S42, based on the divided uniform grid, count the number of disease pixels in each grid unit; S43, based on the number of disease pixels in each grid unit, calculate the disease distribution density of the fish, which includes local density and global density variation coefficient, represented as follows: wherein, is the local density, is the global density coefficient of variation, is the number of diseased pixels, is the total number of pixels, is the local density average, is the local density standard deviation.
9. The intelligent analysis system for measuring fish body parameters and diagnosing diseases according to claim 1, wherein, The fusion analysis module specifically includes the following steps: S51, time and space align the fish biological feature parameters, fish disease diagnosis results and water environment data, so that different data have the same timestamp and spatial reference system; S52, based on the time and space aligned data, calculate the daily growth rate and morphological change rate of the fish according to the fish biological feature parameters, calculate the disease expansion speed according to the fish disease diagnosis results, and calculate the environmental pH gradient and dissolved oxygen change rate according to the water environment data; S53, based on the change rates of various data calculated in step S52, analyze the correlation between fish biological feature parameters, fish disease diagnosis results and water environment data, and evaluate the health status of fish according to the correlation.