A method and system for detecting lake surface algae changes based on frequency domain adaptive filtering

Abstract

A method and system for detecting lake surface algae changes based on frequency domain adaptive filtering, which relates to the field of data recognition. The method includes: obtaining multiple frames of first and second images collected by a target camera, and generating an initial image pair based on the first and second images; extracting the lake surface area as the target to be monitored in the initial image pair to obtain a lake surface monitoring image pair; extracting multiple feature vectors of the lake surface monitoring image pair; performing Fourier transform on the feature vector to obtain frequency domain features and determine dynamic convolution coefficients; performing adaptive filtering, inverse Fourier transform and upsampling reconstruction on the frequency domain features based on the dynamic convolution coefficient to determine the regional contour of the algae change area; determining the area of ​​the algae change area corresponding to the regional contour according to the position parameter; and sending a cleaning prompt to the user terminal when the area of ​​the algae change area is greater than the preset area threshold. The implementation of this method can accurately identify the rapid growth of lake surface algae and improve the management efficiency of garden water bodies.

Description

Technical Field

[0001] The present application relates to the field of data recognition, and in particular to a method and system for detecting changes in lake algae based on frequency domain adaptive filtering. Background Art

[0002] With the acceleration of urbanization and the intensification of environmental pollution, the ecological balance of urban garden water bodies faces severe challenges. The water bloom caused by excessive algae growth not only affects the water landscape, but also produces harmful substances, endangering the aquatic ecosystem and human health. Therefore, garden water management departments urgently need a technical means to detect and monitor algae changes in a timely manner so as to take corresponding control measures.

[0003] Related technologies use a deep learning-based change detection framework for water monitoring. MetaChanger enhances the alignment of dual-phase features through feature interaction, and uses the interactive strategy of aggregation-distribution and feature exchange to detect change areas; GCD-DDPM uses a generative approach to directly generate change maps through difference conditional encoders and variational reasoning.

[0004] However, MetaChanger in related technologies is sensitive to image noise, and atmospheric conditions and sensor noise can affect feature interaction effects; GCD-DDPM is prone to produce pseudo-patterns under complex backgrounds, reducing detection accuracy. Especially in scenes with high dynamics and complex lighting changes such as water surfaces, the methods in related technologies are difficult to accurately identify the actual changing areas of algae. Summary of the invention

[0005] The present application provides a method and system for detecting lake surface algae changes based on frequency domain adaptive filtering, which is used to accurately identify the rapid growth of lake surface algae and improve the management efficiency of garden water bodies.

[0006] In the first aspect, the present application provides a method for detecting changes in lake algae based on frequency domain adaptive filtering, which is applied to a detection system, and the method includes: acquiring multiple frames of first images captured by a target camera at a first moment and multiple frames of second images captured at a second moment, and generating an initial image pair based on the first image and the second image; based on user settings, extracting the lake surface area to be monitored in the initial image pair to obtain a lake surface monitoring image pair; extracting multiple feature vectors of the lake surface monitoring image pair through multi-layer convolution downsampling; performing Fourier transform on the feature vector to obtain frequency domain features, and determining dynamic convolution coefficients based on the frequency domain features; performing adaptive filtering, inverse Fourier transform and upsampling reconstruction on the frequency domain features based on the dynamic convolution coefficients to determine the regional contour of the algae change area; determining the area of ​​the algae change area corresponding to the regional contour based on the position parameters of the target camera; when the area of ​​the algae change area is greater than a preset area threshold, sending a cleaning prompt to the user terminal.

[0007] In the above embodiment, the detection system extracts feature vectors from lake surface monitoring images through multi-layer convolution downsampling, and performs Fourier transform on the feature vectors to obtain frequency domain features, and uses the frequency domain features to determine the dynamic convolution coefficients, and performs adaptive filtering and reconstruction; it can effectively suppress the interference caused by water surface fluctuations and light changes, and improve the accuracy and robustness of detection of algae change areas.

[0008] In combination with some embodiments of the first aspect, in some embodiments, based on user settings, the step of extracting the lake surface area as the target to be monitored in the initial image pair to obtain the lake surface monitoring image pair specifically includes: in response to the user's click operation on the initial image on the user terminal, sending the area image corresponding to the click operation to the user terminal; in response to the area editing operation of the user terminal, performing range correction on the area image to obtain the target area image after editing; extracting the lake surface area as the target to be monitored in the target area image to obtain the lake surface monitoring image.

[0009] In the above embodiment, the detection system responds to the user's clicks and area editing operations, extracts the lake area and corrects the range of the initial image, ensures the accuracy of the monitoring area through interactive area selection, improves the flexibility of the system, and makes the monitoring task more in line with actual needs.

[0010] In combination with some embodiments of the first aspect, in some embodiments, the step of determining the area of ​​the algae change region corresponding to the regional contour according to the position parameters of the target camera specifically includes: determining the coordinate transformation matrix for mapping the regional contour from the image coordinate system to the standard coordinate system according to the position parameters of the target camera; mapping the coordinate points of the regional contour from the image coordinate system to the standard coordinate system according to the coordinate transformation matrix to obtain standard contour points; inputting the plane coordinate values ​​of the standard contour points into the area calculation formula to obtain the area of ​​the algae change region corresponding to the regional contour.

[0011] In the above embodiment, the detection system establishes a mapping relationship between the image coordinate system and the standard coordinate system based on the camera position parameters, and calculates the actual area using standard contour points. The area calculation method based on geometric transformation overcomes the influence of image perspective deformation and can accurately evaluate the algae coverage.

[0012] In conjunction with some embodiments of the first aspect, in some embodiments, the plane coordinate values ​​of the standard contour points are respectively , the area calculation formula is: ; Where n is the total number of standard contour points, is the curve closing point corresponding to the regional contour, .

[0013] In the above embodiment, the detection system adopts a specific area calculation formula to directly calculate the area of ​​the algae change region through the coordinate values ​​of the standard contour points, which can adapt to irregular area shapes and improve the efficiency and accuracy of area assessment.

[0014] In combination with some embodiments of the first aspect, in some embodiments, before the step of obtaining multiple frames of first images captured by the target camera at a first moment and multiple frames of second images captured at a second moment, and generating an initial image pair based on the first image and the second image, the method also includes: obtaining calibration parameters and installation parameters of the target camera; determining the location parameters and monitoring area of ​​the target camera according to the calibration parameters and the installation parameters; sending the monitoring areas of multiple target cameras to a user terminal, so that the user terminal displays an algae monitoring map; the algae monitoring map includes range identifiers corresponding to multiple target cameras.

[0015] In the above embodiment, the detection system obtains the calibration and installation parameters of the camera, generates a monitoring area map to display the coverage of multiple cameras, and forms a network coverage for algae monitoring through a systematic monitoring layout, thereby improving the comprehensiveness and systematicness of the monitoring.

[0016] In combination with some embodiments of the first aspect, in some embodiments, after the step of sending a cleaning prompt to the user terminal when the area of ​​the algae change area is larger than a preset area threshold, the method also includes: establishing an algae growth prediction model based on historical monitoring data; inputting the area of ​​the algae change area into the algae growth prediction model to obtain the predicted value of the algae area within a preset time period in the future; when the predicted value of the algae area exceeds the warning threshold, generating a warning report including an algae distribution heat map, a change trend curve and governance suggestions.

[0017] In the above embodiment, the detection system establishes an algae growth prediction model, analyzes future trends in combination with historical data, and generates an early warning report including a distribution heat map and management suggestions, thereby improving the foresight of algae management.

[0018] In combination with some embodiments of the first aspect, in some embodiments, when the predicted value of the algae area exceeds the warning threshold, after the step of generating a warning report including an algae distribution heat map, a change trend curve and management suggestions, the method also includes: determining the optimal cleaning path based on the position parameters of the target camera, and sending the warning report and the cleaning path to the cleaning device terminal; receiving actual cleaning data transmitted back by the cleaning device terminal, and updating the display status of the warning report based on the actual cleaning data.

[0019] In the above embodiment, the detection system determines the optimal cleaning path according to the camera position and updates the warning status through the actual cleaning data, thereby improving the pertinence and effectiveness of algae control.

[0020] In a second aspect, an embodiment of the present application provides a detection system, which includes: one or more processors and a memory; the memory is coupled to the one or more processors, the memory is used to store computer program code, the computer program code includes computer instructions, and the one or more processors call the computer instructions to enable the detection system to perform the method described in the first aspect and any possible implementation method of the first aspect.

[0021] In a third aspect, an embodiment of the present application provides a computer program product comprising instructions, and when the above-mentioned computer program product runs on a detection system, the above-mentioned detection system executes the method described in the first aspect and any possible implementation method of the first aspect.

[0022] In a fourth aspect, an embodiment of the present application provides a computer-readable storage medium, comprising instructions, which, when executed on a detection system, enable the detection system to execute the method described in the first aspect and any possible implementation of the first aspect.

[0023] It is understandable that the detection system provided in the second aspect, the computer program product provided in the third aspect, and the computer storage medium provided in the fourth aspect are all used to execute the method provided in the embodiment of the present application. Therefore, the beneficial effects that can be achieved can refer to the beneficial effects in the corresponding method, which will not be repeated here.

[0024] One or more technical solutions provided in the embodiments of the present application have at least the following technical effects or advantages:

[0025] 1. Due to the adoption of an algae change detection method based on frequency domain adaptive filtering, image features are extracted through multi-layer convolution downsampling and adaptive filtering is performed in the frequency domain, so the interference of environmental factors such as water surface fluctuations and light changes can be effectively suppressed, and the algae change area can be accurately identified. This effectively solves the problem that deep learning methods in the prior art are sensitive to noise and prone to false detection under complex backgrounds, thereby achieving stable and reliable algae change detection.

[0026] 2. Due to the use of an interactive area selection method, the monitoring area is determined by responding to the user's click operation and area editing operation, so the system can flexibly adjust the monitoring range and key focus areas according to actual needs. It effectively solves the problem of fixed monitoring areas and difficulty in adapting to complex water environments in the existing technology, and thus realizes accurate target area extraction.

[0027] 3. The algae growth prediction model based on historical monitoring data is used to generate early warning reports in combination with multi-dimensional information such as algae distribution heat maps and change trend curves, so the risk of algae outbreaks can be predicted in a timely manner and governance suggestions can be provided. This effectively solves the problem of passive monitoring and lack of foresight in existing technologies, and thus realizes active prevention of algae governance. BRIEF DESCRIPTION OF THE DRAWINGS

[0028] Figure 1 It is a flow chart of a method for detecting lake surface algae changes based on frequency domain adaptive filtering in an embodiment of the present application;

[0029] Figure 2 is a schematic diagram of a model of algae change detection model in an embodiment of the present application;

[0030] Figure 3 (a) is a schematic diagram of a downsampling convolution module of an algae change detection model in an embodiment of the present application;

[0031] Figure 3 (b) is a schematic diagram of a frequency domain adaptive filtering module of the algae change detection model in an embodiment of the present application;

[0032] Figure 4 is another flow chart of the lake surface algae change detection method based on frequency domain adaptive filtering in an embodiment of the present application;

[0033] Figure 5 It is a schematic diagram of the structure of a physical device of the detection system in the embodiment of the present application. DETAILED DESCRIPTION

[0034] The terms used in the following embodiments of the present application are only for the purpose of describing specific embodiments, and are not intended to be used as limitations to the present application. As used in the specification of the present application, the singular expressions "one", "a kind of", "above", "the" and "this" are intended to also include plural expressions, unless there is a clear indication to the contrary in the context. It should also be understood that the term "and / or" used in the present application refers to any or all possible combinations comprising one or more of the listed items.

[0035] In the following, the terms "first" and "second" are used for descriptive purposes only and are not to be understood as suggesting or implying relative importance or implicitly indicating the number of the indicated technical features. Thus, a feature defined as "first" or "second" may explicitly or implicitly include one or more of the features, and in the description of the embodiments of the present application, unless otherwise specified, "plurality" means two or more.

[0036] For ease of understanding, the application scenarios of the embodiments of the present application are introduced below.

[0037] In a tourist attraction in Taihu Lake, lake water quality management has always been an important issue. In the summer of 2023, a large-scale algae outbreak broke out in the tourist lake area of ​​the scenic area, covering an area of ​​3,000 square meters. The traditional manual inspection method only inspects twice a day, and often misses the best time for treatment when large areas of algae are found. A serious algae outbreak forced the scenic area to close for 3 days for treatment, causing huge economic losses. In addition, the odor produced by the decay of algae affected the operation of the surrounding catering areas, and tourist satisfaction dropped significantly. This scene shows that traditional algae monitoring methods have problems such as limited coverage, delayed response, and unpredictability. An intelligent algae monitoring and early warning system is urgently needed.

[0038] In the related art, the recognition and area calculation of algae areas can be achieved by adopting image segmentation and change detection methods based on deep learning. The following introduces the scenario of using the lake surface algae change detection method based on frequency domain adaptive filtering in the related art.

[0039] A scenic spot introduced an algae detection system based on deep learning and used the MetaChanger framework for monitoring. The system performed well on sunny days, but in rainy weather, the false alarm rate increased significantly due to increased image noise caused by changes in lighting. For example, during a rainfall in April 2024, the system misjudged water surface ripples as algae, resulting in multiple ineffective dispatches of cleaning personnel. Another scenic spot used the GCD-DDPM method for detection, which often resulted in misidentification in complex backgrounds (such as aquatic plants, fallen leaves, etc.). Once the system misjudged floating leaves as algae groups, triggering an unnecessary cleaning warning. These problems caused managers to question the reliability of the system, and they had to retain the manual patrol mechanism, which increased management costs.

[0040] The lake surface algae change detection method based on frequency domain adaptive filtering in the embodiment of the present application can effectively suppress environmental noise interference and accurately evaluate the algae coverage area through multi-layer convolution downsampling feature extraction, frequency domain conversion and adaptive filtering processing. The following introduces the scenario of using the lake surface algae change detection method based on frequency domain adaptive filtering in the present application.

[0041] After adopting the application scheme, the algae monitoring effect of a certain scenic spot has been significantly improved. The system effectively distinguishes between water surface fluctuations and actual algae through frequency domain adaptive filtering technology. In a rainy day, despite a large number of ripples on the water surface, the system still accurately identified an algae group with an area of ​​about 200 square meters and promptly notified the cleaning staff to deal with it. Through the precise calibration of the camera position parameters, the area error calculated by the system is controlled within 5%, providing a reliable basis for the allocation of cleaning resources.

[0042] It can be seen that the lake surface algae change detection method based on frequency domain adaptive filtering in the embodiment of the present application can not only realize the detection of algae changes, but also effectively solve the problems of susceptibility to environmental interference and insufficient detection accuracy in the prior art, thereby realizing stable and reliable algae monitoring.

[0043] For ease of understanding, the following describes the process of the method provided by this implementation in combination with the above scenario. Figure 1 , is a flow chart of a method for detecting lake surface algae changes based on frequency domain adaptive filtering in an embodiment of the present application.

[0044] S101, acquiring a plurality of frames of first images acquired by a target camera at a first moment and a plurality of frames of second images acquired at a second moment, and generating an initial image pair based on the first image and the second image.

[0045] Among them, the target camera refers to a fixed camera device installed in the lake monitoring area; the first moment and the second moment refer to two different image acquisition time points, and the time interval between the two is the detection interval, which can be configured by the staff; multi-frame images refer to a sequence of multiple pictures collected continuously at the same moment; the initial image pair refers to an image group formed by pairing the images of the first moment and the second moment in time.

[0046] When the detection system starts the algae monitoring task, it needs to obtain image data at two time points for change detection. Specifically, the detection system first determines the sampling time interval of the target camera, then collects 3-5 frames of images at two time points respectively, and combines the images at the two time points into an initial image pair. Figure 2 , Figure 2 This is a schematic diagram of the algae change detection model in the embodiment of the present application. It should be noted that the algae change detection model AFF-CD in the present application adopts the UNet structure as a whole, first continuously downsampling compression coding (based on Figure 3 (a) downsampling convolution module), and then fusion frequency domain adaptive filtering (based on Figure 3 (b) frequency domain adaptive filtering module) followed by upsampling decoding. Figure 2 As shown in Figure 2 ImagesA and ImagesB, which are used as model input 1 and input 2 respectively, are the initial image pairs, which include multiple images collected at two time nodes.

[0047] In some embodiments, image acquisition and image pair generation can be achieved in a variety of ways: optionally, set a fixed sampling time interval, such as collecting images every 4 hours; perform motion blur detection and brightness evaluation on the collected images; select the image with the highest score to form an image pair. Optionally, dynamically adjust the sampling interval according to weather conditions so that the previous and next images are in the same weather conditions to avoid image interference caused by weather changes; use multi-frame image superposition and average noise reduction; align image pairs based on image registration algorithms. It is understandable that other methods can also be used to achieve image acquisition and processing, which are not limited here.

[0048] S102: based on user settings, extract the lake surface area to be monitored in the initial image pair to obtain a lake surface monitoring image pair.

[0049] Among them, user settings refer to the monitoring area configuration performed by the operator through the system interface; the target to be monitored refers to the lake water area that needs to be focused on; the lake surface monitoring image pair refers to the image pair that only contains the target water area after area extraction.

[0050] After obtaining the initial image pair, the detection system needs to determine the specific monitoring area according to the actual monitoring needs. Specifically, the system first displays the initial image pair for the user to view, then receives the area of ​​interest defined by the user through the mouse or touch, and finally accurately extracts and optimizes the boundaries of the area to generate a monitoring image pair that only contains the target lake area.

[0051] In some embodiments, the monitoring area can be extracted in a variety of ways: optionally, using the Segment Anything Model (SAM) algorithm to automatically extract the lake area; applying morphological processing to optimize the boundary; optionally, using a deep learning model to automatically segment the water area; combining polygon editing tools for manual correction; and refining the boundary based on color and texture features. It is understandable that other methods can also be used to achieve area extraction and optimization, which are not limited here.

[0052] S103. Extract multiple feature vectors of the lake surface monitoring image pair through multi-layer convolution down-sampling.

[0053] Among them, multi-layer convolution refers to a neural network structure composed of multiple convolutional layers; downsampling refers to the operation of reducing the image resolution; feature vector refers to a numerical array that represents image features.

[0054] After obtaining the lake monitoring image pair, the detection system needs to extract the deep feature representation of the image. Specifically, the system uses a designed multi-layer convolutional neural network to extract multi-scale features of the image, performs downsampling operations after each layer of convolution to reduce the size of the feature map, and finally obtains a low-dimensional feature vector that can represent the image content. Figure 2 As shown in Figure 2 Schematic diagram of the algae change detection model in the embodiment of the present application; ImagesA and ImagesB as model input 1 and input 2 will be used to extract the lake area ( Figure 2 Omitted in the figure), and then input to the ConCat layer for splicing, that is Figure 2 The C module in . After that, the encoding process will be performed. In the encoding stage, there are i layers of downsampling. The convolution modules of each layer are represented as {CONV_DOWN_A1, CONV_DOWN_A2, ..., CONV_DOWN_Ai} and {CONV_DOWN_B1, CONV_DOWN_B2, ..., CONV_DOWN_Bi}. The convolution modules of the two inputs share parameters between the same layer. Figure 3 (a), Figure 3 (a) is a module schematic diagram of the downsampling convolution module of the algae change detection model in the embodiment of the present application; the downsampling convolution module will first pass through two 3x3 convolutions, including conv3x3 (indicating that a 3x3 convolution kernel is used to perform a convolution operation on the input feature map to extract local features), BN (i.e., Batch Normalization, which accelerates model convergence and improves training stability by normalizing each batch of feature maps), and ReLU (i.e., Rectified Linear Unit, a commonly used activation function that performs nonlinear transformation on the convolution result to enhance the expression ability of the model), and after the maximum pooling layer is downsampled, max pool2x2 is generated. max pool2x2 means that a 2x2 maximum pooling operation is used to downsample the feature map to reduce the size of the feature map.

[0055] S104, performing Fourier transform on the feature vector to obtain frequency domain features, and determining dynamic convolution coefficients according to the frequency domain features.

[0056] Among them, Fourier transform refers to the mathematical operation of converting signals from the spatial domain to the frequency domain; frequency domain features refer to the representation of images in the frequency space; and dynamic convolution coefficients refer to filtering parameters that are adaptively adjusted according to frequency domain features.

[0057] After obtaining the feature vector, the detection system needs to convert it to the frequency domain for analysis and processing. Specifically, the system first performs a two-dimensional Fourier transform on the feature vector to obtain a spectrum representation, then analyzes the amplitude and phase characteristics of the spectrum, and finally dynamically calculates the convolution kernel parameters based on the distribution characteristics of the frequency domain features to adapt to different image change patterns. Figure 2 As shown in Figure 2This is a schematic diagram of the model of the algae change detection model in the embodiment of the present application; following the above step S103, after the maximum pooling layer downsampling, each layer output (frequency domain feature) of the coding layer will synchronously pass through the frequency domain adaptive filtering module (AFF). Figure 3 (b), Figure 3 (b) is a schematic diagram of a frequency domain adaptive filtering module (AFF) of the algae change detection model in an embodiment of the present application; Figure 3 In (b), the input feature map x (i.e., feature vector) is Fourier transformed to the frequency domain space FFT(x) (i.e., frequency domain feature), and then through avg pool (i.e., Average Pooling, global average pooling, which compresses the feature map into a single value by averaging the entire feature map and reducing the feature dimension), FC (i.e., Fully Connected Layer, which maps the pooled feature vector to another feature space through linear transformation and nonlinear activation), ReLU (i.e., Rectified Linear Unit, which performs nonlinear transformation on the output of the fully connected layer to enhance the feature representation capability), FC (i.e., another fully connected layer, which further maps the feature vector to the input dimension of the softmax function), softmax (i.e., Softmax Function, which normalizes the input vector to a probability distribution and obtains the weight distribution of the dynamic convolution coefficient), and finally, conv_1 (the convolution corresponding to the input image pair, corresponding to the representation) is calculated according to the frequency domain distribution. Figure 2 CONV_DOWN_A1 and CONV_DOWN_B1 in , and the following conv_2, ..., conv_k are expressed similarly), the dynamic convolution coefficients π_1, π_2, ..., π_k corresponding to conv_2, ..., conv_k.

[0058] In some embodiments, frequency domain feature processing can be implemented in a variety of ways: optionally, using a fast Fourier transform algorithm for frequency domain conversion; calculating the amplitude and phase information of the spectrum; estimating the optimal convolution parameters based on the frequency distribution. Optionally, using wavelet transform to obtain time-frequency features; constructing a frequency domain attention module; using an adaptive threshold to determine the convolution coefficient. It is understandable that other methods can also be used to implement frequency domain feature analysis and parameter estimation, which are not limited here.

[0059] S105, performing adaptive filtering, inverse Fourier transform and up-sampling reconstruction on the frequency domain features based on the dynamic convolution coefficients to determine the regional contour of the algae change area.

[0060] Among them, adaptive filtering refers to the filtering operation that is dynamically adjusted according to the signal characteristics; inverse Fourier transform refers to the operation of converting the frequency domain signal back to the spatial domain; upsampling reconstruction refers to the process of restoring the original resolution of the image.

[0061] The detection system needs to filter the frequency domain features and reconstruct the image. Specifically, the system first uses dynamic convolution coefficients to adaptively filter the frequency domain features to suppress noise and irrelevant information, then converts the filtered features back to the spatial domain through inverse Fourier transform, and finally reconstructs the feature map of the original resolution through upsampling operations and extracts the contour information of the algae change area. Figure 3 (b), Figure 3 (b) is a schematic diagram of a frequency domain adaptive filtering module (AFF) of the algae change detection model in an embodiment of the present application; Figure 3 In (b), for the dynamic convolution coefficients π_1, π_2, ..., π_k,

[0062] Through y = IFFT(FFT(x)*CONV(π_1,π_2,...,π_k)), we can get the frequency domain filtering result, that is, the dynamic convolution coefficient is calculated through conv (i.e. Convolution, convolution operation, using dynamic convolution coefficient as convolution kernel, adaptive filtering of frequency domain features), BN (i.e. Batch Normalization, batch normalization, normalizing the convolution result, accelerating convergence, and improving training stability), and activate (i.e. Activation Function, activation function, performing nonlinear transformation on the normalized feature map to enhance feature representation capabilities. Common activation functions include ReLU, sigmoid, tanh, etc.), and generate iFFT (i.e. Inverse Fast Fourier Transform, inverse Fourier transform, converting the frequency domain filtering result back to the spatial domain to obtain the feature map after adaptive filtering) as the y output of the adaptive filtering module (AFF). Then return to Figure 2 , Figure 2 Schematic diagram of the algae change detection model in the embodiment of the present application; the detection system will enter the decoding stage based on the frequency domain filtering result. The decoding stage is the inverse process of the encoding stage, and the downsampling is replaced by up-convolution. Figure 2 In the figure, it is specifically represented by CONV_UP_O1, CONV_UP_O2, …CONV_UP_Oi, etc. After upsampling and reconstruction, a SegMask is generated as the model output, that is, the regional outline of the algae change area (specifically the coordinate data of all points on the outline).

[0063] In some embodiments, feature reconstruction and contour extraction can be achieved in a variety of ways: optionally, design an adaptive Wiener filter to process frequency domain features; use inverse fast Fourier transform to restore spatial features; use bilinear interpolation for upsampling. Optionally, construct a multi-scale adaptive filter group; combine phase information to optimize reconstruction quality; use active contour model to extract boundaries. It is understandable that feature reconstruction and contour extraction can also be achieved in other ways, which are not limited here.

[0064] S106. Determine the area of ​​the algae change region corresponding to the regional contour according to the position parameters of the target camera.

[0065] Among them, the position parameter represents the geometric information of the camera such as the installation height and angle; the regional contour refers to the boundary curve of the algae change area; and the area of ​​the change area refers to the actual area size of the algae coverage area.

[0066] The detection system needs to convert the contour information in the image into actual area data. Specifically, the system first reads the calibration parameters and installation parameters of the camera, then establishes a mapping relationship from the image coordinate system to the actual geographic coordinate system, and finally calculates the actual coverage area based on the coordinates of the contour points.

[0067] In some embodiments, area calculation can be implemented in a variety of ways: optionally, establishing a homography matrix based on camera parameters; projecting contour points to a ground coordinate system; using a polygon area formula to calculate the actual area. Optionally, using depth map-assisted 3D reconstruction; combining GIS data for georeferencing; using an integral method to calculate the area of ​​an irregular region. It is understandable that other methods can also be used to implement area calculation, which are not limited here.

[0068] S107: When the area of ​​the algae change region is larger than a preset area threshold, a cleaning prompt is sent to the user terminal.

[0069] Among them, the preset area threshold represents the area judgment standard for triggering the cleaning warning; the cleaning prompt refers to the algae control notification sent to the user; and the user terminal refers to the mobile device or management platform that receives the warning information.

[0070] The detection system needs to decide whether to issue a cleanup notice based on the area calculation results. Specifically, the system compares the calculated algae coverage area with the pre-set warning threshold. When the coverage area exceeds the threshold, the system pushes a cleanup notice to the terminal device of the relevant user through the communication network to remind the user to clean up the algae in time.

[0071] In some embodiments, early warning push can be implemented in a variety of ways: optionally, setting multi-level area early warning thresholds; generating early warning information including location and area; sending notifications through message push services. Optionally, dynamically adjusting thresholds based on historical data; automatically generating cleaning suggestions; and synchronously pushing early warning information through multiple channels. It is understandable that other methods can also be used to implement early warning management and information push, which are not limited here.

[0072] In the above embodiment, the detection system realizes accurate detection of algae changes and area calculation through frequency domain adaptive filtering technology. In practical applications, in order to further improve the practicality of the system, the detection system also needs to support user interaction operations to make the selection and editing of monitoring areas more flexible and accurate. Therefore, in the subsequent steps, the area selection and editing functions based on user interaction are added.

[0073] The following is a more detailed description of the process of the method provided by this implementation. Figure 4 , is another flow chart of the lake surface algae change detection method based on frequency domain adaptive filtering in an embodiment of the present application.

[0074] S401, acquiring a plurality of frames of first images acquired by a target camera at a first moment and a plurality of frames of second images acquired at a second moment, and generating an initial image pair based on the first image and the second image.

[0075] Referring to step S101 , the detection system generates an initial image pair.

[0076] In some embodiments, the detection system obtains the calibration parameters and installation parameters of the target camera; determines the location parameters and monitoring area of ​​the target camera based on the calibration parameters and installation parameters; sends the monitoring areas of multiple target cameras to the user terminal, so that the user terminal displays an algae monitoring map; the algae monitoring map includes range identifiers corresponding to multiple target cameras.

[0077] Among them, calibration parameters refer to the set of internal parameters that describe the imaging characteristics of the camera; installation parameters refer to the spatial position and orientation information of the camera in the actual scene; position parameters refer to the spatial geometric relationship of the camera after comprehensive consideration of calibration and installation; monitoring area refers to the effective coverage of a single camera; algae monitoring map refers to the visual interface that displays the monitoring areas of multiple cameras; range identification refers to the graphic mark of the camera coverage range marked on the map.

[0078] Before starting the monitoring task, the detection system needs to complete the camera parameter configuration and monitoring area visualization. Specifically, the system first obtains the camera's internal parameter matrix and distortion coefficient through calibration, and records the camera's installation height, inclination and other installation information, and then calculates the actual monitoring range of each camera. Finally, the monitoring areas of all cameras are integrated into the electronic map to form an intuitive monitoring network layout diagram.

[0079] In some embodiments, the generation and display of the monitoring map can be achieved in a variety of ways: optionally, using chessboard calibration to obtain camera parameters; determining the installation location based on GPS positioning; calculating the projection range to generate the coverage area. Optionally, using an electronic map as the base map; using semi-transparent polygons to mark the coverage area; adding camera information tags. It is understandable that other methods can also be used to achieve the visual display of the monitoring network, which is not limited here.

[0080] S402 : In response to a click operation of the user on the initial image at the user terminal, a region image corresponding to the click operation is sent to the user terminal.

[0081] Among them, the user terminal refers to the computer or mobile device used by the operator; the click operation refers to the interactive behavior of the user selecting the target location through the mouse or touch; the regional image refers to the local image area centered on the click location.

[0082] The detection system needs to respond to the user's interactive operation to determine the area of ​​interest. Specifically, the system first displays the initial image on the user terminal for the user to browse. When the user's click operation is detected, the system automatically captures the local area image centered on the click position and transmits the area image back to the user terminal in real time for further editing.

[0083] In some embodiments, interactive response and image transmission can be achieved in a variety of ways: optionally, calculating the pixel coordinates of the click position; capturing the regional image according to the preset window size; and achieving real-time transmission through WebSocket. Optionally, supporting multi-touch to select multiple areas; adaptively adjusting the capture window size; and using image compression to speed up transmission. It is understandable that other methods can also be used to achieve user interaction and data transmission, which are not limited here.

[0084] S403: In response to the region editing operation of the user terminal, the region image is corrected to obtain the edited target region image.

[0085] Among them, the region editing operation refers to the set of operations performed by the user to adjust the region range; the range correction refers to the process of updating the region boundary according to the user editing; the target region image refers to the final monitoring region image determined after editing.

[0086] The detection system needs to optimize the area range according to the user's editing operations. Specifically, the system provides regional editing tools for users to adjust the area boundaries, including zooming in, zooming out, panning and deformation, and updates the display editing effects in real time, and finally generates a target area image that meets user needs.

[0087] In some embodiments, the region editing function can be implemented in a variety of ways: optionally, a rectangular frame adjustment tool is provided; polygon vertex dragging is supported; region translation and rotation are implemented. Optionally, Bezier curves are used to edit boundaries; eraser and brush tools are provided; undo and redo functions are added. It is understood that other methods can also be used to implement region editing and range correction, which are not limited here.

[0088] S404: extracting the lake surface area as the target to be monitored in the target area image to obtain a lake surface monitoring image.

[0089] Among them, the target area image refers to the image area edited by the user; the target to be monitored refers to the water area that needs to be monitored; the lake surface monitoring image refers to the processing result image that only contains the water surface area.

[0090] The detection system needs to accurately extract the water surface area from the target area image. Specifically, the system first segments the image to distinguish the water surface and non-water surface areas, then applies morphological operations to optimize the boundaries, and finally extracts the pure water surface area through mask operations to generate a lake surface image for subsequent monitoring.

[0091] In some embodiments, water surface area extraction can be achieved in a variety of ways: optionally, using color thresholds to segment the water surface; applying a region growing algorithm to expand the region; performing boundary smoothing and hole filling. Optionally, using a deep learning model for semantic segmentation; combining multi-feature fusion optimization results; using image gradients to constrain boundaries. It is understandable that other methods can also be used to achieve water surface area extraction, which are not limited here.

[0092] S405. Extract multiple feature vectors of the lake surface monitoring image pair through multi-layer convolution down-sampling.

[0093] Referring to step S103 , the detection system determines a plurality of feature vectors.

[0094] S406: Perform Fourier transform on the feature vector to obtain frequency domain features, and determine dynamic convolution coefficients according to the frequency domain features.

[0095] Referring to step S104, the detection system determines the dynamic convolution coefficient.

[0096] S407, performing adaptive filtering, inverse Fourier transform and up-sampling reconstruction on the frequency domain features based on the dynamic convolution coefficients to determine the regional contour of the algae change area.

[0097] Referring to step S105 , the detection system generates a region outline.

[0098] S408. Determine a coordinate conversion matrix for mapping the region outline from the image coordinate system to the standard coordinate system according to the position parameters of the target camera.

[0099] Among them, the position parameter represents the three-dimensional spatial position and orientation information of the camera; the image coordinate system refers to the pixel coordinate system with the upper left corner of the image as the origin; the standard coordinate system represents the coordinate system of the actual geographic space; and the coordinate transformation matrix refers to the mathematical transformation matrix that realizes the transformation between two coordinate systems.

[0100] After obtaining the area contour, the detection system needs to convert the coordinates of the image space into the actual space coordinates. Specifically, the system first reads the camera's installation height, tilt angle and other position parameters, then establishes the projection equation based on the pinhole camera model, and finally solves the homography matrix as the coordinate transformation matrix for subsequent coordinate mapping.

[0101] In some embodiments, the coordinate transformation matrix can be solved in a variety of ways: optionally, the calibration plate image is collected to obtain the intrinsic parameter matrix; the PnP algorithm is used to solve the extrinsic parameter matrix; the internal and external parameters are combined to calculate the homography matrix. Optionally, the perspective transformation matrix is ​​calculated using known control point pairs; the least squares method is used to optimize the transformation parameters; and distortion correction is introduced to improve accuracy. It is understandable that other methods can also be used to achieve coordinate system transformation, which are not limited here.

[0102] S409 , mapping the coordinate points of the region contour from the image coordinate system to the standard coordinate system according to the coordinate conversion matrix to obtain standard contour points.

[0103] Among them, the coordinate points represent discrete sampling points on the area contour; the mapping operation refers to the mathematical operation of coordinate transformation using the transformation matrix; the standard contour point refers to the coordinates of the contour point in the actual geographic space.

[0104] The detection system needs to transform the contour points in the image into a standard coordinate system. Specifically, the system first samples the contour of the area at equal intervals to obtain a sequence of key points, then multiplies the pixel coordinates of each point by the coordinate transformation matrix, and finally obtains the actual position coordinates of these points in the standard coordinate system.

[0105] In some embodiments, the coordinate mapping of contour points can be achieved in a variety of ways: optionally, contour points are acquired by equal arc length sampling; homogeneous coordinate transformation calculation is performed; coordinate normalization is performed. Optionally, a B-spline curve is used to fit the contour; feature points are extracted for sparse mapping; and dense corresponding points are calculated by interpolation. It is understandable that other methods can also be used to achieve contour point transformation, which are not limited here.

[0106] S410, inputting the plane coordinate values ​​of the standard contour points into the area calculation formula to obtain the area of ​​the algae change region corresponding to the regional contour.

[0107] Among them, the plane coordinate value refers to the position value of the standard contour point on the two-dimensional plane; the area calculation formula represents the mathematical formula for calculating the area of ​​the polygon; the area of ​​the algae change region refers to the numerical result of the actual coverage area.

[0108] The detection system needs to calculate the actual coverage area based on the standard contour points. Specifically, the system first organizes the standard contour points into closed polygons in order, and then uses the polygon area calculation formula to calculate the actual algae coverage area through the plane coordinate values ​​of the contour points.

[0109] In some embodiments, area calculation can be implemented in a variety of ways: optionally, using the upper triangle point sequence to calculate the area; applying the Gaussian area formula to solve; performing unit conversion to output the result. Optionally, using the integral method to calculate the area of ​​the irregular region; introducing density weighting to calculate the effective area; estimating the uncertainty of the area calculation. It is understandable that other methods can also be used to implement area calculation, which are not limited here. In some embodiments, the plane coordinate values ​​of the standard contour points are respectively , the area calculation formula is: ;

[0110] Where n is the total number of standard contour points, is the curve closing point corresponding to the regional contour, During the calculation process, for the coordinate point The area of ​​the corresponding curve is calculated as:

[0111] ;

[0112] Since the coordinate points correspond to each other, there will be , ,… ,Right now A type of formula, because the curve is closed, ,all The formulas of are added together to cancel each other out. Considering the positive and negative values, the final formula is .

[0113] S411. When the area of ​​the algae change region is larger than a preset area threshold, a cleaning prompt is sent to the user terminal.

[0114] Referring to step S107, the detection system will send a cleaning prompt to the user terminal.

[0115] S412. Establish an algae growth prediction model based on historical monitoring data.

[0116] Among them, historical monitoring data refers to the record of changes in algae area collected over a period of time in the past; the growth prediction model refers to a mathematical model that describes the growth laws of algae; and the prediction model parameters refer to the various coefficients and variables in the model.

[0117] The detection system needs to build a prediction model for algae growth. Specifically, the system first collects and preprocesses historical monitoring data, then analyzes the relationship between algae growth and environmental factors, and finally trains a machine learning model to predict the growth trend of algae.

[0118] In some embodiments, the construction of the prediction model can be achieved in a variety of ways: optionally, using time series analysis to extract periodic features; establishing the association between environmental factors and growth rate; training LSTM neural network model. Optionally, using ensemble learning methods to fuse multiple sub-models; introducing meteorological data to enhance prediction capabilities; using cross-validation to evaluate model performance. It is understandable that other methods can also be used to achieve growth prediction, which are not limited here.

[0119] It should be noted that the input of the algae growth prediction model includes historical monitoring data (algae area change records) and environmental parameters (such as water temperature, pH value, etc.), and outputs the predicted value of algae area in the future preset time period. The model establishes a mathematical model of the growth law of algae by analyzing the time series characteristics and environmental factor correlation in historical data. The training criteria include the mean square error between the predicted value and the actual value, the trend accuracy within the prediction time length, and the model's adaptability to seasonal changes. The model uses deep learning structures such as LSTM to capture time series features, and integrates multiple prediction sub-models through ensemble learning methods to improve prediction accuracy.

[0120] S413, inputting the area of ​​the algae change region into the algae growth prediction model to obtain the predicted value of the algae area within a future preset time period.

[0121] Among them, the area of ​​the algae change region represents the currently detected algae coverage area; the preset time period refers to the future time range that needs to be predicted; and the area prediction value refers to the future algae coverage area value predicted by the model.

[0122] The detection system needs to use a prediction model to estimate the future trend of algae changes. Specifically, the system takes the currently detected algae area data as input, performs time series prediction through the trained prediction model, and obtains the trend of algae area changes in the future, including the predicted area value at each time point.

[0123] In some embodiments, area prediction can be achieved in a variety of ways: optionally, setting a sliding time window; adding environmental parameters to assist prediction; calculating the confidence interval of the prediction results. Optionally, using multi-model integrated prediction; considering the influence of seasonal factors; dynamically updating prediction parameters. It is understandable that other methods can also be used to achieve algae area prediction, which is not limited here.

[0124] S414. When the predicted value of the algae area exceeds the warning threshold, a warning report including an algae distribution heat map, a change trend curve and management suggestions is generated.

[0125] Among them, the warning threshold represents the critical area value that triggers the warning; the heat map refers to a visual image showing the distribution density of algae; the trend curve represents a line graph showing the change of algae area over time; and the governance suggestion refers to the disposal plan given based on the prediction results.

[0126] The detection system needs to generate warning information when the predicted value exceeds the threshold. Specifically, the system compares the predicted area value with the warning threshold. When it exceeds the threshold, it automatically generates a warning report containing multi-dimensional information, including algae distribution heat map, area change trend chart, and governance suggestions generated based on empirical rules.

[0127] In some embodiments, early warning report generation can be achieved in a variety of ways: optionally, using color gradients to represent distribution density; drawing time series prediction curves; matching optimal governance solutions. Optionally, generating multi-scale heat maps; adding historical comparative analysis; formulating graded early warning strategies. It is understandable that other methods can also be used to achieve early warning report generation, which are not limited here.

[0128] In some embodiments, the detection system determines the optimal cleaning path based on the location parameters of the target camera, and sends the warning report and the cleaning path to the cleaning device terminal; receives the actual cleaning data sent back by the cleaning device terminal, and updates the display status of the warning report based on the actual cleaning data.

[0129] Among them, the optimal cleaning path refers to the best operating route of the cleaning equipment; the cleaning equipment terminal refers to the intelligent device that performs the algae cleaning task; the actual cleaning data refers to the operation information collected during the cleaning process; the display status indicates the real-time update status of each indicator in the early warning report.

[0130] After issuing an early warning, the detection system needs to coordinate the execution of the cleaning work. Specifically, the system first plans the shortest path covering all areas to be cleaned based on the positional relationship of multiple cameras and the distribution of algae, and then sends the early warning information containing the cleaning path to the cleaning equipment. During the cleaning process, the system receives the operation data returned by the equipment in real time, including information such as the cleaning location, area, and effect, and updates the relevant indicators in the early warning report accordingly.

[0131] In some embodiments, the coordination and monitoring of cleaning tasks can be achieved in a variety of ways: optionally, using a heuristic algorithm to plan the cleaning path; establishing real-time communication with the cleaning equipment; calculating the cleaning progress and efficiency indicators. Optionally, optimizing the cleaning sequence based on algae density; using blockchain to record cleaning data; generating a cleaning work report. It is understandable that other methods can also be used to achieve intelligent scheduling of cleaning work, which is not limited here.

[0132] In the embodiment of the present application, since frequency domain adaptive filtering technology is used to process the water surface image, image features are extracted through multi-layer convolution and adaptive filtering is performed in the frequency domain, it is possible to effectively suppress the interference of environmental factors such as water surface fluctuations and light changes, thereby improving the accuracy and robustness of detection of algae change areas, and effectively solving the problem in the prior art that deep learning methods are sensitive to noise and prone to false detection in complex backgrounds, thereby achieving stable and reliable detection of algae changes.

[0133] The following describes the detection system in the embodiment of the present invention from the perspective of hardware processing. Figure 5 , is a schematic diagram of a physical device structure of a detection system in an embodiment of the present application.

[0134] It should be noted that Figure 5 The structure of the detection system shown is only an example and should not bring any limitation to the functions and scope of use of the embodiments of the present invention.

[0135] like Figure 5 As shown, the detection system includes a central processing unit (CPU) 501, which can perform various appropriate actions and processes according to the program stored in the read-only memory (ROM) 502 or the program loaded from the storage part 508 to the random access memory (RAM) 503, such as executing the method described in the above embodiment. In the RAM 503, various programs and data required for system operation are also stored. The CPU 501, the ROM 502 and the RAM 503 are connected to each other through the bus 504. The input / output (I / O) interface 505 is also connected to the bus 504.

[0136] The following components are connected to the I / O interface 505: an input section 506 including an audio input device, a button switch, etc.; an output section 507 including a liquid crystal display (LCD) and an audio output device, an indicator light, etc.; a storage section 508 including a hard disk, etc.; and a communication section 509 including a network interface card such as a LAN (Local Area Network) card, a modem, etc. The communication section 509 performs communication processing via a network such as the Internet. A drive 510 is also connected to the I / O interface 505 as needed. A removable medium 511, such as a magnetic disk, an optical disk, a magneto-optical disk, a semiconductor memory, etc., is installed on the drive 510 as needed so that a computer program read therefrom is installed into the storage section 508 as needed.

[0137] In particular, according to an embodiment of the present invention, the process described above with reference to the flowchart can be implemented as a computer software program. For example, an embodiment of the present invention includes a computer program product, which includes a computer program carried on a computer-readable medium, and the computer program includes a computer program for executing the method shown in the flowchart. In such an embodiment, the computer program can be downloaded and installed from a network through a communication part 509, and / or installed from a removable medium 511. When the computer program is executed by a central processing unit (CPU) 501, various functions defined in the present invention are performed.

[0138] It should be noted that specific examples of computer-readable storage media may include, but are not limited to: an electrical connection with one or more wires, a portable computer disk, a hard disk, a random access memory (RAM), a read-only memory (ROM), an erasable programmable read-only memory (EPROM), a flash memory, an optical fiber, a portable compact disk read-only memory (CD-ROM), an optical storage device, a magnetic storage device, or any suitable combination thereof. In the present invention, a computer-readable storage medium may be any tangible medium containing or storing a program that may be used by or in combination with an instruction execution system, apparatus, or device.

[0139] The flowcharts and block diagrams in the accompanying drawings illustrate the possible architecture, functions and operations of the systems, methods and computer program products according to various embodiments of the present invention. Each box in the flowchart or block diagram may represent a module, a program segment, or a part of a code, and the above-mentioned module, program segment, or a part of a code contains one or more executable instructions for implementing the specified logical functions. It should also be noted that in some alternative implementations, the functions marked in the box may also occur in an order different from that marked in the accompanying drawings.

[0140] Specifically, the detection system of this embodiment includes a processor and a memory. The memory stores a computer program. When the computer program is executed by the processor, the lake surface algae change detection method based on frequency domain adaptive filtering provided in the above embodiment is implemented.

[0141] As another aspect, the present invention further provides a computer-readable storage medium, which may be included in the detection system described in the above embodiment; or may exist independently without being assembled into the detection system. The above storage medium carries one or more computer programs, and when the above one or more computer programs are executed by a processor of the detection system, the detection system implements the lake surface algae change detection method based on frequency domain adaptive filtering provided in the above embodiment.

[0142] As described above, the above embodiments are only used to illustrate the technical solutions of the present application, rather than to limit them. Although the present application has been described in detail with reference to the aforementioned embodiments, those skilled in the art should understand that they can still modify the technical solutions described in the aforementioned embodiments, or make equivalent replacements for some of the technical features therein. However, these modifications or replacements do not cause the essence of the corresponding technical solutions to deviate from the scope of the technical solutions of the embodiments of the present application.

[0143] As used in the above embodiments, the term "when..." may be interpreted to mean "if..." or "after..." or "in response to determining..." or "in response to detecting...", depending on the context. Similarly, the phrases "upon determining..." or "if (the stated condition or event) is detected" may be interpreted to mean "if determining..." or "in response to determining..." or "upon detecting (the stated condition or event)" or "in response to detecting (the stated condition or event)", depending on the context.

[0144] Those skilled in the art can understand that to implement all or part of the processes in the above-mentioned embodiments, the processes can be completed by computer programs to instruct related hardware, and the programs can be stored in computer-readable storage media. When the programs are executed, they can include the processes of the above-mentioned method embodiments. The aforementioned storage media include: ROM or random access memory RAM, magnetic disk or optical disk and other media that can store program codes.

Claims

1. A method for detecting lake surface algae changes based on frequency domain adaptive filtering, characterized in that: Applied to a detection system, the method comprises: Acquire multiple frames of first images acquired by a target camera at a first moment and multiple frames of second images acquired at a second moment, and generate an initial image pair based on the first image and the second image; Based on user settings, extracting the lake surface area as the target to be monitored in the initial image pair to obtain a lake surface monitoring image pair; Extracting multiple feature vectors of the lake surface monitoring image pair through multi-layer convolution downsampling; Performing Fourier transform on the feature vector to obtain frequency domain features, and determining dynamic convolution coefficients according to the frequency domain features; Based on the dynamic convolution coefficient, the frequency domain features are adaptively filtered, inverse Fourier transformed and up-sampled to reconstruct, so as to determine the regional contour of the algae change area; Determine the area of ​​the algae change region corresponding to the regional contour according to the position parameters of the target camera; When the area of ​​the algae change region is greater than a preset area threshold, a cleaning prompt is sent to the user terminal.

2. The method according to claim 1, characterized in that The step of extracting the lake surface area as the target to be monitored in the initial image pair based on the user setting to obtain the lake surface monitoring image pair specifically includes: In response to a user clicking on the initial image at a user terminal, sending a region image corresponding to the clicking operation to the user terminal; In response to the region editing operation of the user terminal, performing range correction on the region image to obtain an edited target region image; A lake surface area as a target to be monitored in the target area image is extracted to obtain a lake surface monitoring image.

3. The method according to claim 1, characterized in that The step of determining the area of ​​the algae change region corresponding to the regional contour according to the position parameters of the target camera specifically includes: Determine, according to the position parameters of the target camera, a coordinate transformation matrix for mapping the area contour from the image coordinate system to the standard coordinate system; According to the coordinate conversion matrix, the coordinate points of the area contour are mapped from the image coordinate system to the standard coordinate system to obtain standard contour points; The plane coordinate values ​​of the standard contour points are input into the area calculation formula to obtain the area of ​​the algae change region corresponding to the regional contour.

4. The method according to claim 3, characterized in that The plane coordinate values ​​of the standard contour points are respectively , the area calculation formula is: ； Wherein, n is the total number of standard contour points, is the curve closing point corresponding to the contour of the region, .

5. The method according to claim 1, characterized in that Before the step of acquiring a plurality of frames of first images acquired by the target camera at a first moment and a plurality of frames of second images acquired at a second moment, and generating an initial image pair based on the first image and the second image, the method further includes: Obtaining calibration parameters and installation parameters of the target camera; Determining the location parameters and monitoring area of ​​the target camera according to the calibration parameters and the installation parameters; The monitoring areas of the multiple target cameras are sent to a user terminal, so that the user terminal displays an algae monitoring map; the algae monitoring map includes range identifiers corresponding to the multiple target cameras.

6. The method according to claim 1, characterized in that After the step of sending a cleaning prompt to a user terminal when the area of ​​the algae change region is greater than a preset area threshold, the method further includes: Establish an algae growth prediction model based on historical monitoring data; Inputting the area of ​​the algae change region into the algae growth prediction model to obtain the predicted value of the algae area within a future preset time period; When the predicted value of the algae area exceeds the warning threshold, an early warning report including an algae distribution heat map, a change trend curve and management suggestions is generated.

7. The method according to claim 6, characterized in that After the step of generating an early warning report including an algae distribution heat map, a change trend curve and management suggestions when the algae area prediction value exceeds the early warning threshold, the method further includes: Determine an optimal cleaning path according to the position parameters of the target camera, and send the warning report and the cleaning path to a cleaning device terminal; Receive actual cleaning data sent back by the cleaning device terminal, and update the display status of the early warning report based on the actual cleaning data.

8. A detection system, characterized in that: The detection system comprises: one or more processors and a memory; the memory is coupled to the one or more processors, the memory is used to store computer program code, the computer program code comprises computer instructions, and the one or more processors call the computer instructions so that the detection system executes the method as described in any one of claims 1-7.

9. A computer-readable storage medium comprising instructions, characterized in that: When the instructions are executed on a detection system, the detection system is caused to execute the method according to any one of claims 1 to 7.

10. A computer program product, characterized in that When the computer program product is run on a detection system, the detection system is caused to perform the method according to any one of claims 1 to 7.

Images