Park landscape water body water quality perception early warning management method and system

Abstract

The application discloses a park landscape water body water quality sensing early warning management method and system, belongs to the technical field of water quality monitoring and landscape management, and comprises the following steps: collecting water surface multi-spectrum images and binocular stereo images of a park water body to form a time-synchronized multi-view image sequence; constructing a three-dimensional space coordinate of a water body surface according to the multi-view image sequence; mapping each band reflectivity on the three-dimensional coordinate to obtain a water body surface reflectivity spatial distribution map; extracting a water body appearance index vector according to the water body surface reflectivity spatial distribution map; establishing a correlation between reflectivity and water quality optical indexes, and inversely solving a water quality optical index vector; calculating a consistency deviation value of the water body appearance index vector and the water quality optical index vector, judging whether the indexes exceed a preset early warning threshold value, and outputting an early warning message if the indexes exceed the preset early warning threshold value; and outputting a management strategy according to the early warning message. The application avoids external interference through non-contact sensing, and reduces the false alarm rate through appearance and water quality optical double determination.

Description

Technical Field

[0001] This invention relates to the field of water quality monitoring and landscape management technology, and in particular to a method and system for water quality perception, early warning and management of park landscape water bodies. Background Technology

[0002] Park landscape water features are an important component of the urban ecosystem, serving the dual functions of ecological regulation and landscape beautification. Compared with large lakes, reservoirs, and other natural water bodies, park landscape water features have the following significant characteristics:

[0003] I. Irregular shape and high geometric complexity: In order to meet the needs of landscape design, the water bodies in the park are mostly irregular in shape, with winding and varied shorelines and a large number of concave and convex turns, making it difficult for traditional monitoring schemes based on regular grid layout to adapt.

[0004] Second, frequent and intense human interference: Tourists feeding, throwing foreign objects, trampling on the shore and other behaviors can cause strong instantaneous disturbances in local areas, resulting in short-term and violent fluctuations in water quality parameters. However, such fluctuations do not represent the true changes in the overall water quality.

[0005] Third, there is a conflict between ecological and landscape functions: water quality deterioration will affect ecological health, but some water quality improvement measures, such as the introduction of chemical agents and mechanical aeration, may damage the water surface landscape effect and affect the visitor experience.

[0006] Currently, water quality monitoring primarily employs methods such as immersion sensors for fixed-point monitoring, manual sampling and laboratory analysis, and remote sensing satellite or drone monitoring to achieve early warning of landscape water bodies. Regardless of the method, most suffer from a lack of focus on a single monitoring objective. For example, water quality monitoring focuses only on chemical indicators, while landscape monitoring focuses only on appearance indicators, lacking an effective correlation between the two. In reality, there is an inherent link between chemical and appearance indicators. For instance, increased chlorophyll concentration leads to a greenish hue and decreased transparency in water; increased turbidity causes cloudiness and shifts in chromaticity coordinates. When these two types of indicators are inconsistent—for example, normal chemical indicators but abnormal appearance indicators—it often indicates localized interference or measurement errors. Directly triggering an early warning and taking appropriate measures in such cases may lead to resource waste or even be counterproductive.

[0007] Therefore, traditional methods for sensing and early warning of water quality in park landscape water bodies still have certain shortcomings. Summary of the Invention

[0008] The purpose of this invention is to provide a method and system for water quality perception and early warning management of park landscape water bodies in order to solve the problem of low accuracy of traditional methods for water quality perception and early warning of park landscape water bodies. Through non-contact multi-view image acquisition, consistency analysis of water quality optical indicators and appearance indicators, and ecological landscape collaborative decision-making, it can achieve accurate perception, reliable early warning and reasonable handling of small irregular park water bodies.

[0009] Firstly, to achieve the above objectives, the technical solution adopted by the present invention is as follows:

[0010] The method for water quality sensing, early warning, and management of park landscape water bodies includes the following steps:

[0011] A time-synchronized multi-view image sequence was created by acquiring multispectral images of the park's water surface and binocular stereo images.

[0012] The three-dimensional spatial coordinates of the water surface are constructed based on the binocular stereo images in the multi-view image sequence, and the spatial distribution map of the water surface reflectance is obtained by mapping the reflectance of each band on the three-dimensional coordinates based on the multispectral images.

[0013] Based on the spatial distribution map of water surface reflectance, a vector of water appearance indicators is extracted. The indicators in the vector include transparency attenuation depth, chromaticity coordinates, water surface uniformity, and floating debris coverage. A water body optical correlation model is constructed to establish the correlation between reflectance and water quality optical indicators, including chlorophyll concentration, turbidity, and chemical oxygen demand. The water quality optical indicator vector is obtained by minimizing the objective function and inverting the solution.

[0014] Calculate the consistency deviation value between the water body appearance index vector and the water quality optical index vector, preset a deviation threshold, and when the consistency deviation value is less than the deviation threshold, determine whether either the water body appearance index vector or the water quality optical index vector exceeds a preset warning threshold. If it exceeds the threshold, output a warning message.

[0015] Management strategies are output based on early warning messages.

[0016] Preferably, the multispectral acquisition bands for collecting multispectral images of the park's water surface include at least the blue light band, green light band, red light band, and near-infrared band.

[0017] Preferably, the method further includes setting the number of acquisition points in the multi-view image sequence and the observation angle of each acquisition point, wherein the setting method is as follows:

[0018] Extract the water area A, shoreline perimeter P, principal axis length L, and secondary axis length W, and calculate the shoreline morphological complexity K and aspect ratio E. The calculation formulas are as follows:

[0019] ;

[0020] ;

[0021] Simultaneously, the shoreline is discretized and sampled to obtain the local curvature at each discrete point. And calculate the mean curvature. and curvature variance The calculation formula is as follows:

[0022] ;

[0023] ;

[0024] in, This represents the total number of discrete points along the shoreline.

[0025] Calculate the total number of collection points The calculation formula is as follows:

[0026] ;

[0027] in, , , , The preset weighting coefficients, , This serves as a normalized baseline quantity.

[0028] Total number of collection points The distribution along the shoreline ensures that the spacing between each sampling point corresponds to the local curvature of its corresponding location. Inversely proportional, based on local curvature Calculate the first Observation angle of each collection point The calculation formula is as follows:

[0029] ;

[0030] in, This represents the observation angle corresponding to the normal direction of the shoreline at this sampling point. This represents the maximum curvature of the entire shoreline.

[0031] Preferably, the method for constructing the three-dimensional spatial coordinates of the water surface based on the binocular stereo images in the multi-view image sequence includes:

[0032] Map the left and right eye images of the binocular stereo image to the same epipolar direction;

[0033] Based on the epipolar constraint, for each pixel (x,y) in the left eye image, a matching window Ω of a preset size is selected on the corresponding epipolar line in the right eye image. The matching window Ω is a window with a preset size of 3-5 pixels centered on the pixel (x,y).

[0034] The pixel offset of the same spatial point in the left and right eye images is solved by minimizing the disparity cost function. The expression for the disparity cost function is as follows:

[0035] ;

[0036] in, and These are the pixel grayscale value matrices for the left and right eye images, respectively. This is the pixel offset. The smaller the value, the higher the matching degree between the pixels of the left and right images;

[0037] Calculate the three-dimensional spatial coordinates of each pixel on the water surface. , , The calculation formula is as follows:

[0038] ;

[0039] ;

[0040] ;

[0041] in, The coordinates of the pixel in the left eye image. , The pixel size of a stereo camera. The focal length of a binocular camera. This is the reference distance between the two eyes of a binocular camera.

[0042] Preferably, the method for extracting the water body appearance index vector based on the spatial distribution map of water surface reflectance is as follows:

[0043] The transparency attenuation depth index is extracted by estimating the reflectance of the blue and red light bands in the reflectance spatial distribution map using a logarithmic transformation. The formula for the logarithmic transformation estimation is as follows:

[0044] ;

[0045] in, The preset attenuation coefficient of the optical properties of water. The reflectivity is in the red light band. The reflectivity is in the blue light band;

[0046] The chromaticity coordinate index is extracted by calculating the chromaticity parameters of the water body through the RGB three-band reflectance in the reflectance spatial distribution map and mapping them to standard chromaticity coordinate values;

[0047] The water surface uniformity index is extracted by calculating the coefficient of variation of the reflectance of each pixel in the reflectance spatial distribution map, and taking the reciprocal of the coefficient of variation as the uniformity. The calculation formula is as follows:

[0048] ;

[0049] ;

[0050] Where the coefficient of variation , The average reflectance of the water surface. The standard deviation of reflectance, This represents the total number of pixels in the water body. For the first The reflectance of each pixel;

[0051] The floating debris coverage rate index is extracted by threshold segmentation of the reflectance spatial distribution map, identifying floating debris areas, and calculating the ratio of the floating debris area area to the total water area to obtain the floating debris coverage rate.

[0052] The four extracted indicators are integrated in a preset order to obtain the water body appearance indicator vector.

[0053] Preferably, the expression for the water body optical correlation model used to construct the correlation between reflectivity and water quality optical indicators from the inversion solution of the water quality optical index vector is as follows:

[0054] ;

[0055] in, For reflectivity, The reference reflectance of the clean water body in the corresponding wavelength band, The water absorption coefficient, The water scattering coefficient, Optical path length This is the error term;

[0056] Methods for obtaining the water quality optical index vector by minimizing the objective function include:

[0057] By minimizing the objective function, we find a set of chlorophyll concentration, turbidity, and chemical oxygen demand that minimizes the deviation between the reflectance calculated by the water body optical correlation model and the actual observed reflectance. The expression for the objective function is as follows:

[0058] ;

[0059] in, It is the sum of squares of the deviations between the reflectance calculated by the model and the actual observed reflectance. , , The regularization coefficient is . This is a regularization term.

[0060] Preferably, the formula for calculating the consistency deviation value between the water body appearance index vector and the water quality optical index vector is as follows:

[0061] ;

[0062] ;

[0063] ;

[0064] in, , , The preset weighting coefficients, The dispersion of water quality optical indicators. This represents the average value of each indicator in the water quality optical index. The standard deviation of each indicator is . This represents the dispersion of water body appearance indicators. This represents the average value of each indicator among the water body's appearance indicators. The standard deviation of each indicator is . is the correlation coefficient between the water body appearance index vector and the water quality optical index vector.

[0065] Preferably, the management strategy based on the early warning message output includes an ecological priority strategy, a landscape priority strategy, and an ecological-landscape synergy strategy, and the method for selecting the corresponding strategy includes:

[0066] Calculate the overall score of water body appearance The calculation formula is as follows:

[0067] ;

[0068] in, , , , These are the weighting coefficients for each indicator. , , , The values ​​are, in order, the transparency attenuation depth, chromaticity coordinates, water surface uniformity, and floating object coverage.

[0069] Set a judgment threshold, if If the water quality is below the judgment threshold and any of the water quality optical indicators does not exceed the warning threshold, then the landscape priority strategy is adopted.

[0070] like If the water quality is below the judgment threshold and any one of the water quality optical indicators exceeds the warning threshold, then the ecological landscape synergy strategy will be selected.

[0071] like If the water quality is not less than the judgment threshold and any one of the water quality optical indicators exceeds the warning threshold, then the ecological priority strategy shall be adopted.

[0072] Secondly, to achieve the above objectives, the technical solution adopted by the present invention is as follows:

[0073] The park landscape water quality sensing and early warning management system includes:

[0074] The image acquisition unit is used to acquire multispectral images of the water surface of the park's water bodies and binocular stereo images to form a time-synchronized multi-view image sequence.

[0075] The reflectance calculation unit is used to construct the three-dimensional spatial coordinates of the water surface based on the binocular stereo images in the multi-view image sequence, and to obtain the spatial distribution map of the water surface reflectance by mapping the reflectance of each band on the three-dimensional coordinates based on the multispectral images.

[0076] The index vector calculation unit is used to extract the water body appearance index vector based on the spatial distribution map of water surface reflectance. The indexes in the water body appearance index vector include transparency attenuation depth, chromaticity coordinates, water surface uniformity, and floating debris coverage. A water body optical correlation model is constructed to establish the correlation between reflectance and water quality optical indicators, including chlorophyll concentration, turbidity, and chemical oxygen demand. The water quality optical index vector is obtained by minimizing the objective function and inverting the solution.

[0077] The early warning judgment unit is used to calculate the consistency deviation value between the water body appearance index vector and the water quality optical index vector. A deviation threshold is preset. When the consistency deviation value is less than the deviation threshold, it is determined whether either the water body appearance index vector or the water quality optical index vector exceeds the preset early warning threshold. If it exceeds the threshold, an early warning message is output.

[0078] The strategy management unit is used to output management strategies based on early warning messages.

[0079] Thirdly, to achieve the above objectives, the technical solution adopted by the present invention is as follows:

[0080] A storage medium having a computer program stored thereon, which, when executed by a processor, implements the park landscape water quality perception and early warning management method as described in the first aspect.

[0081] Compared with the prior art, the present invention has the following beneficial effects:

[0082] This invention acquires water surface information by using a non-contact, shore-based multispectral and binocular stereo image acquisition scheme, eliminating the need for sensors to enter the water and avoiding inaccurate data acquisition due to interference. It extracts water body appearance indicators and water quality optical indicators from the acquired image data, avoiding misjudgments caused by relying on a single indicator. Furthermore, it calculates the consistency deviation between the water body appearance indicator vector and the water quality optical indicator vector; if the consistency deviation is low, it determines whether to output a warning message to avoid false alarms. This enables accurate perception, reliable early warning, and appropriate handling of small, irregularly shaped park water bodies. Attached Figure Description

[0083] Figure 1 This is a flowchart of the park landscape water quality perception and early warning management method of the present invention.

[0084] Figure 2 This is a schematic diagram of the composition of the park landscape water quality perception and early warning management system of the present invention. Detailed Implementation

[0085] To make the technical means, creative features, objectives and effects of this invention easier to understand, the invention will be further described below in conjunction with specific embodiments.

[0086] Example 1

[0087] like Figure 1 As shown, the park landscape water quality sensing and early warning management method includes the following steps:

[0088] Step S1 involves acquiring a time-synchronized multi-view image sequence composed of multispectral images of the park's water surface and binocular stereo images. Image acquisition is achieved by fixing a multispectral camera and a binocular stereo camera on the bank of the park's water features, ensuring that both devices are synchronized and that the field of view covers the entire water body. The multispectral acquisition bands for the acquired water surface images include at least the blue light band (450-490nm), the green light band (520-560nm), the red light band (630-690nm), and the near-infrared band (760-900nm). This is because the multispectral bands cover the water quality-sensitive spectral range; blue or red light reflects water transparency, green light reflects landscape color, and near-infrared light distinguishes water from floating objects. The binocular stereo camera, based on the parallax principle, enables three-dimensional reconstruction. This step selects the multi-view image sequence as a preliminary step in the entire early warning management method, employing non-contact acquisition to avoid water disturbance, equipment corrosion, and data interference caused by sensor immersion in water.

[0089] This application is applicable to park landscape water bodies that are small in area and have a high boundary ratio, such as those with a large ratio of shoreline length to area and irregular shape. Because these water bodies have complex geometric shapes and are frequently disturbed by humans, traditional fixed-point monitoring is insufficient to comprehensively perceive water quality conditions, and contact sensors can introduce local disturbances, leading to data distortion. Therefore, this method also includes setting the number of acquisition points for the multi-view image sequence and the observation angle of each acquisition point. The setting method is as follows:

[0090] Extract the water area A, shoreline perimeter P, principal axis length L, and secondary axis length W, and calculate the shoreline morphological complexity K and aspect ratio E. The calculation formulas are as follows:

[0091] ;

[0092] ;

[0093] The aforementioned A, P, L, and W parameters can be obtained by acquiring shoreline contour images of the target water body using shore-side camera devices, such as fixed monitoring cameras or drone aerial photography, and then segmenting and extracting the water body from the contour images. The shoreline morphological complexity K reflects the morphological complexity of the water body; the closer the water body shape is to a circle, the more complex it becomes. The closer the value is to 1, the more complex the shape, such as a winding shoreline with many bumps and depressions. The larger the value, the greater the aspect ratio E, which reflects the elongation of the water body. The larger the E value, the more the water body resembles a river channel. When E is close to 1, the water body resembles a lake. These two parameters together determine the degree of spatial heterogeneity of the water body and are the basis for determining the sampling density.

[0094] Simultaneously, the shoreline is discretized and sampled to obtain the local curvature at each discrete point. And calculate the mean curvature. and curvature variance The calculation formula is as follows:

[0095] ;

[0096] ;

[0097] in, The total number of discrete points on the shoreline; the variance of curvature. It reflects the dramatic undulation of the shoreline. A higher value indicates a more tortuous shoreline, requiring a denser network of monitoring points.

[0098] Calculate the total number of collection points The calculation formula is as follows:

[0099] ;

[0100] in, , , , The preset weighting coefficients, , The normalized baseline quantity; the total number of data collection points. The calculation formula takes into account the size, morphological complexity, elongation, and shoreline curvature of the water body. For water bodies with large area, complex shape, and curvature, the number of sampling points is automatically increased; while for regular small water bodies, the number of sampling points is reduced. This allocation method ensures the optimal allocation of monitoring resources and avoids the problems of waste caused by overly dense sampling points in small water bodies or missed detection caused by overly sparse sampling points in complex water bodies.

[0101] Total number of collection points The distribution along the shoreline ensures that the spacing between each sampling point corresponds to the local curvature of its corresponding location. Inversely proportional, based on local curvature Calculate the first Observation angle of each collection point The calculation formula is as follows:

[0102] ;

[0103] in, This represents the observation angle corresponding to the normal direction of the shoreline at this sampling point. The maximum curvature of the entire shoreline is found in areas where the shoreline is straight, i.e., local curvature. When the angle approaches 0, the observation angle approaches the normal direction. ≈ At this point, the imaging resolution is high, suitable for detailed monitoring; while the local curvature of the shoreline... When the angle is large, the observation angle is shifted obliquely to avoid areas with strong specular reflection on the water surface, while expanding the field of view coverage and reducing monitoring blind spots. Therefore, this step uses this adaptive angle setting to solve the technical problem that some areas of the park's water body cannot be effectively imaged or are overexposed due to specular reflection caused by irregular shorelines.

[0104] Step S2: Construct the three-dimensional spatial coordinates of the water surface based on the binocular stereo images in the multi-view image sequence, and obtain the spatial distribution map of water surface reflectance by mapping the reflectance of each band on the three-dimensional coordinates based on the multispectral images. This step uses the principle of binocular stereo vision to reconstruct the three-dimensional structure of the water surface and maps the multispectral information onto the three-dimensional spatial coordinates to form a high-resolution spatial distribution map of water surface reflectance. Compared with traditional single-point monitoring, this distribution map retains the spatial heterogeneity information of the water body and can identify the water quality differences of different functional areas such as boundary retention areas and central open areas.

[0105] The method involves constructing the three-dimensional spatial coordinates of the water surface based on the binocular stereo images in the multi-view image sequence. Binocular stereo vision, based on the parallax principle, simulates human visual perception by acquiring two images of the same scene from different perspectives (left and right eye images) and calculating the three-dimensional coordinates of spatial points using triangulation principles. The method includes:

[0106] Mapping the left and right images of the stereo image to the same epipolar direction is the purpose of keeping the corresponding epipolar lines of the two images horizontally aligned, which simplifies the subsequent pixel matching search and reduces the two-dimensional search to a one-dimensional search.

[0107] Based on the epipolar constraint, for each pixel (x,y) in the left eye image, a matching window Ω of a preset size is selected on the corresponding epipolar line in the right eye image. The matching window Ω is a window with a preset size of 3-5 pixels centered on the pixel (x,y).

[0108] The pixel offset of the same spatial point in the left and right eye images is solved by minimizing the disparity cost function. The expression for the disparity cost function is as follows:

[0109] ;

[0110] in, and These are the pixel grayscale value matrices for the left and right eye images, respectively. This is the pixel offset. The smaller the value, the higher the matching degree between the pixels of the left and right images; this step involves finding the optimal value among all possible disparity values. Minimum pixel offset ;

[0111] Calculate the three-dimensional spatial coordinates of each pixel on the water surface. , , The calculation formula is as follows:

[0112] ;

[0113] ;

[0114] ;

[0115] in, The coordinates of the pixel in the left eye image. , The pixel size of a stereo camera. The focal length of a binocular camera. This is the reference distance between the two eyes of a binocular camera.

[0116] After obtaining the three-dimensional coordinates of the water surface, the reflectance information of each band of the multispectral image is mapped onto the corresponding three-dimensional coordinates to form a spatial distribution map of the water surface reflectance. The mapping method is as follows:

[0117] Pixels in a multispectral image can be mapped to three-dimensional points in a binocular coordinate system through projection transformation. , , On the ), for each three-dimensional point, its reflectance values ​​in the four bands of blue, green, red, and near-infrared are recorded to form a four-dimensional data volume. , , , , , , ) .

[0118] Traditional monitoring can only obtain water quality data at a single point or a few discrete points, while the reflectance spatial distribution map generated in this embodiment is planar and high-resolution, so it can clearly show the spatial heterogeneity of reflectance on the water surface. For example, the low reflectance in the boundary retention area represents water turbidity, the uniform reflectance in the central area, and the high reflectance in the floating object covered area provide data support for the subsequent extraction of spatial heterogeneity indicators such as transparency, color, and uniformity.

[0119] Step S3: Extract water body appearance index vectors based on the spatial distribution map of water surface reflectance. These vectors include indicators such as transparency attenuation depth, chromaticity coordinates, water surface uniformity, and floating debris coverage. Construct a water body optical correlation model to establish the relationship between reflectance and water quality optical indicators, including chlorophyll concentration, turbidity, and chemical oxygen demand (COD). The water quality optical index vectors are obtained by minimizing the objective function through inversion. In this step, on the one hand, appearance indicators such as transparency, chromaticity, uniformity, and floating debris coverage are directly extracted from the spatial distribution map of reflectance to reflect the aesthetic state of the landscape. On the other hand, based on the water body optical radiation transmission mechanism, a correlation model is established between reflectance and water quality optical indicators such as chlorophyll concentration, turbidity, and COD. Water quality parameters are solved through inversion algorithms. This dual-indicator extraction method, combining appearance and water quality, avoids the limitations of relying on a single indicator.

[0120] The method for extracting water body appearance index vectors based on the spatial distribution map of water surface reflectance is as follows:

[0121] The transparency attenuation depth index is extracted by estimating the reflectance of the blue and red light bands in the reflectance spatial distribution map using a logarithmic transformation. The formula for the logarithmic transformation estimation is as follows:

[0122] ;

[0123] in, The preset attenuation coefficient of water body optical properties is related to the water body type and is used for cleaning water bodies. Low value, turbid water body The value is relatively large. The reflectivity is in the red light band. The reflectance is the blue light band. Red light has weaker penetrating power in water than blue light. The ratio of their reflectances is negatively correlated with the turbidity of the water. Taking the logarithm can linearize the transparency.

[0124] The extraction of chromaticity coordinates involves calculating the chromaticity parameters of the water body using the RGB three-band reflectance in the reflectance spatial distribution map and mapping them to standard chromaticity coordinate values. Chromaticity coordinates can quantify the color bias of the water body, such as blue-green indicating cleanliness, yellow-green indicating algal proliferation, and yellow-brown indicating organic pollution.

[0125] The water surface uniformity index is extracted by calculating the coefficient of variation of the reflectance of each pixel in the reflectance spatial distribution map, and taking the reciprocal of the coefficient of variation as the uniformity. The calculation formula is as follows:

[0126] ;

[0127] ;

[0128] Where the coefficient of variation , The average reflectance of the water surface. The standard deviation of reflectance, This represents the total number of pixels in the water body. For the first Reflectance of individual pixels; uniformity A larger size indicates a more uniform water surface and a better landscape effect. The smaller the value, the more likely there is local turbidity or accumulation of floating objects, resulting in a poor landscape effect;

[0129] The floating debris coverage rate index is extracted by threshold segmentation of the reflectance spatial distribution map, identifying floating debris areas, and calculating the ratio of the floating debris area area to the total water area to obtain the floating debris coverage rate.

[0130] The four extracted indicators are integrated in a preset order to obtain the water body appearance indicator vector. .

[0131] Water quality optical indicators: chlorophyll concentration turbidity and chemical oxygen demand Since direct observation is not possible, it is necessary to invert the water quality optical index by establishing a water body optical correlation model. Therefore, the expression of the water body optical correlation model for establishing the correlation between reflectivity and water quality optical indexes in the water quality optical index vector obtained by the inversion is as follows:

[0132] ;

[0133] in, For reflectivity, The reference reflectance of the clean water body in the corresponding wavelength band, The water absorption coefficient is related to chlorophyll concentration. Chemical oxygen demand Relatedly, the worse the water quality and the more pollutants, the greater the absorption coefficient. The water scattering coefficient, and turbidity Correlation is relevant; the higher the turbidity, the greater the scattering coefficient. Optical path length represents the distance a light ray travels in a body of water. This is the error term;

[0134] Methods for obtaining the water quality optical index vector by minimizing the objective function include:

[0135] By minimizing the objective function, we find a set of chlorophyll concentration, turbidity, and chemical oxygen demand that minimizes the deviation between the reflectance calculated by the water body optical correlation model and the actual observed reflectance. The expression for the objective function is as follows:

[0136] ;

[0137] in, It is the sum of squares of the deviations between the reflectance calculated by the model and the actual observed reflectance. , , The regularization coefficient is . This is the regularization term. The introduction of the regularization term is based on the assumption of spatial continuity. In small-scale park water bodies, the water quality parameters of adjacent areas should not change abruptly. By applying spatial smoothing constraints, the inversion bias caused by strong local disturbances in the boundary areas of small water bodies, such as feeding by tourists on the shore, can be reduced, thereby improving the robustness of the inversion results.

[0138] Step S4: Calculate the consistency deviation value between the water body appearance index vector and the water quality optical index vector. A deviation threshold is preset. When the consistency deviation value is less than the threshold, determine whether either the water body appearance index vector or the water quality optical index vector exceeds a preset warning threshold. If it does, an warning message is output. Frequent human interference in park water bodies, such as visitors feeding the water, throwing objects, and trampling on the banks, can cause data to appear abnormal during collection, but this does not necessarily indicate actual water quality deterioration. By calculating the consistency deviation value between the two types of index vectors (appearance index and water quality optical index), false anomalies caused by external interference can be identified. Only after eliminating external interference can a genuine water quality event be determined, effectively solving the problem of false alarms caused by human interference.

[0139] In monitoring water features in parks, two types of inconsistencies are frequently encountered:

[0140] Scenario A: The water quality optical indicators show abnormally high values, but the appearance indicators are normal. This may be due to local disturbances, such as false positives caused by tourists feeding the water near the camera.

[0141] Scenario B: Appearance indicators show a significant decline, but water quality optical indicators are normal. This may be due to monitoring blind spots caused by soluble organic matter or colloidal particles.

[0142] To identify the aforementioned inconsistencies, this embodiment incorporates a consistency deviation value calculation. Taking into account the variability and correlation of the two types of indicators, the formula for calculating the consistency deviation value between the water body appearance indicator vector and the water quality optical indicator vector is as follows:

[0143] ;

[0144] ;

[0145] ;

[0146] in, , , The preset weighting coefficients, The dispersion of water quality optical indicators. This represents the average value of each indicator in the water quality optical index. The standard deviation of each indicator is . This represents the dispersion of water body appearance indicators. This represents the average value of each indicator among the water body's appearance indicators. The standard deviation of each indicator is . The correlation coefficient between the water body appearance index vector and the water quality optical index vector; when When the value is small, it indicates that the two types of indicators change synchronously and have a consistent spatial distribution, making the data reliable. However, when... When the value is large, it indicates that there is a significant difference between the two types of indicators, so it may be interference. Therefore, when outputting the warning message in this step, the warning judgment is only made when the consistency deviation value is less than the deviation threshold. The warning message is then output after determining whether any one of the water body appearance index vector and the water quality optical index vector exceeds the preset warning threshold.

[0147] Step S5: Output management strategies based on the warning message.

[0148] The management strategies output based on early warning messages include ecological priority strategies, landscape priority strategies, and ecological-landscape synergy strategies. The methods for selecting the corresponding strategies include:

[0149] Calculate the overall score of water body appearance , A higher value indicates a better landscape condition. The calculation formula is as follows:

[0150] ;

[0151] in, , , , These are the weighting coefficients for each indicator. , , , The values ​​are, in order, the transparency attenuation depth, chromaticity coordinates, water surface uniformity, and floating object coverage.

[0152] Set a judgment threshold, if If the water quality is below the judgment threshold and any of the water quality optical indicators does not exceed the warning threshold, then the landscape priority strategy is adopted. At this time, the water quality is still acceptable but the landscape effect has deteriorated, such as mild algae accumulation and increased floating objects. Direct drug administration may damage the ecology and affect the landscape. Physical measures such as mechanical algae removal, manual removal of floating objects, and the addition of aeration fountains are adopted to quickly restore the landscape effect while avoiding chemical intervention.

[0153] like If the water quality is below the judgment threshold and any one of the water quality optical indicators exceeds the warning threshold, an ecological landscape synergy strategy is adopted. At this point, both water quality and landscape deteriorate, requiring a balance between purification and landscape protection. Biological control measures such as introducing algae-eating fish, planting aquatic plants, ecological floating beds, and partial water exchanges are implemented to improve water quality and create an ecological landscape, avoiding the side effects of chemical agents.

[0154] like If the water quality is not less than the judgment threshold and any of the water quality optical indicators exceeds the warning threshold, then the ecological priority strategy is adopted. At this time, the water quality has deteriorated but the landscape impact has not yet appeared. It is necessary to quickly control the water quality to prevent deterioration and take chemical or physical measures such as adding algaecides, forced aeration, and local enclosure to prioritize ecological safety. Even if there is a slight impact on the landscape in the short term, such as temporary turbidity, it is acceptable.

[0155] Example 2

[0156] like Figure 2 As shown, the park landscape water quality sensing and early warning management system includes:

[0157] The image acquisition unit is used to acquire multispectral images of the water surface of the park's water bodies and binocular stereo images to form a time-synchronized multi-view image sequence.

[0158] The reflectance calculation unit is used to construct the three-dimensional spatial coordinates of the water surface based on the binocular stereo images in the multi-view image sequence, and to obtain the spatial distribution map of the water surface reflectance by mapping the reflectance of each band on the three-dimensional coordinates based on the multispectral images.

[0159] The index vector calculation unit is used to extract the water body appearance index vector based on the spatial distribution map of water surface reflectance. The indexes in the water body appearance index vector include transparency attenuation depth, chromaticity coordinates, water surface uniformity, and floating debris coverage. A water body optical correlation model is constructed to establish the correlation between reflectance and water quality optical indicators, including chlorophyll concentration, turbidity, and chemical oxygen demand. The water quality optical index vector is obtained by minimizing the objective function and inverting the solution.

[0160] The early warning judgment unit is used to calculate the consistency deviation value between the water body appearance index vector and the water quality optical index vector. A deviation threshold is preset. When the consistency deviation value is less than the deviation threshold, it is determined whether either the water body appearance index vector or the water quality optical index vector exceeds the preset early warning threshold. If it exceeds the threshold, an early warning message is output.

[0161] The strategy management unit is used to output management strategies based on early warning messages.

[0162] Since Embodiment 1 and Embodiment 2 are essentially the same, the units in Embodiment 2 will not be explained in detail.

[0163] The present invention also discloses a storage medium storing a computer program, which, when executed by a processor, implements the park landscape water quality perception and early warning management method as described in Example 1.

[0164] It will be apparent to those skilled in the art that the present invention is not limited to the details of the exemplary embodiments described above, and that the invention can be implemented in other specific forms without departing from its spirit or essential characteristics. Therefore, the embodiments should be considered in all respects as exemplary and non-limiting, and the scope of the invention is defined by the appended claims rather than the foregoing description. Thus, all variations falling within the meaning and scope of equivalents of the claims are intended to be included within the present invention. No reference numerals in the claims should be construed as limiting the scope of the claims.

[0165] Furthermore, it should be understood that although this specification describes embodiments, not every embodiment contains only one independent technical solution. This narrative style is merely for clarity. Those skilled in the art should consider the specification as a whole, and the technical solutions in each embodiment can also be appropriately combined to form other embodiments that can be understood by those skilled in the art.

Claims

1. A method for water quality sensing, early warning, and management of park landscape water bodies, characterized in that: Includes the following steps: A time-synchronized multi-view image sequence was created by acquiring multispectral images of the park's water surface and binocular stereo images. The three-dimensional spatial coordinates of the water surface are constructed based on the binocular stereo images in the multi-view image sequence, and the spatial distribution map of the water surface reflectance is obtained by mapping the reflectance of each band on the three-dimensional coordinates based on the multispectral images. Based on the spatial distribution map of water surface reflectance, a vector of water appearance indicators is extracted. The indicators in the vector include transparency attenuation depth, chromaticity coordinates, water surface uniformity, and floating debris coverage. A water body optical correlation model is constructed to establish the correlation between reflectance and water quality optical indicators, including chlorophyll concentration, turbidity, and chemical oxygen demand. The water quality optical indicator vector is obtained by minimizing the objective function and inverting the solution. Calculate the consistency deviation value between the water body appearance index vector and the water quality optical index vector, preset a deviation threshold, and when the consistency deviation value is less than the deviation threshold, determine whether either the water body appearance index vector or the water quality optical index vector exceeds a preset warning threshold. If it exceeds the threshold, output a warning message. Management strategies are output based on early warning messages.

2. The method for water quality sensing, early warning, and management of park landscape water bodies according to claim 1, characterized in that, The multispectral acquisition bands for collecting multispectral images of the park's water surface include at least the blue light band, green light band, red light band, and near-infrared band.

3. The method for water quality sensing, early warning, and management of park landscape water bodies according to claim 1, characterized in that, The method also includes setting the number of acquisition points in the multi-view image sequence and the observation angle of each acquisition point, as follows: Extract the water area A, shoreline perimeter P, principal axis length L, and secondary axis length W, and calculate the shoreline morphological complexity K and aspect ratio E. The calculation formulas are as follows: ； ； Simultaneously, the shoreline is discretized and sampled to obtain the local curvature at each discrete point. And calculate the mean curvature. and curvature variance The calculation formula is as follows: ； ； in, This represents the total number of discrete points along the shoreline. Calculate the total number of collection points The calculation formula is as follows: ； in, , , , The preset weighting coefficients, , This serves as a normalized baseline quantity. Total number of collection points The distribution along the shoreline ensures that the spacing between each sampling point corresponds to the local curvature of its corresponding location. Inversely proportional, based on local curvature Calculate the first Observation angle of each collection point The calculation formula is as follows: ； in, This represents the observation angle corresponding to the normal direction of the shoreline at this sampling point. This represents the maximum curvature of the entire shoreline.

4. The method for water quality sensing, early warning, and management of park landscape water bodies according to claim 1, characterized in that, The method for constructing the three-dimensional spatial coordinates of the water surface based on the binocular stereo images in the multi-view image sequence includes: Map the left and right eye images of the binocular stereo image to the same epipolar direction; Based on the epipolar constraint, for each pixel (x,y) in the left eye image, a matching window Ω of a preset size is selected on the corresponding epipolar line in the right eye image. The matching window Ω is a window with a preset size of 3-5 pixels centered on the pixel (x,y). The pixel offset of the same spatial point in the left and right eye images is solved by minimizing the disparity cost function. The expression for the disparity cost function is as follows: ； in, and These are the pixel grayscale value matrices for the left and right eye images, respectively. This is the pixel offset. The smaller the value, the higher the matching degree between the pixels of the left and right images; Calculate the three-dimensional spatial coordinates of each pixel on the water surface. , , The calculation formula is as follows: ； ； ； in, The coordinates of the pixel in the left eye image. , The pixel size of a stereo camera. The focal length of a binocular camera. This is the reference distance between the two eyes of a binocular camera.

5. The method for water quality sensing, early warning, and management of park landscape water bodies according to claim 2, characterized in that, The method for extracting water body appearance index vectors based on the spatial distribution map of water surface reflectance is as follows: The transparency attenuation depth index is extracted by estimating the reflectance of the blue and red light bands in the reflectance spatial distribution map using a logarithmic transformation. The formula for the logarithmic transformation estimation is as follows: ； in, The preset attenuation coefficient of the optical properties of water. The reflectivity is in the red light band. The reflectivity is in the blue light band; The chromaticity coordinate index is extracted by calculating the chromaticity parameters of the water body through the RGB three-band reflectance in the reflectance spatial distribution map and mapping them to standard chromaticity coordinate values; The water surface uniformity index is extracted by calculating the coefficient of variation of the reflectance of each pixel in the reflectance spatial distribution map, and taking the reciprocal of the coefficient of variation as the uniformity. The calculation formula is as follows: ； ； Where the coefficient of variation , The average reflectance of the water surface. The standard deviation of reflectance, This represents the total number of pixels in the water body. For the first The reflectance of each pixel; The floating debris coverage rate index is extracted by threshold segmentation of the reflectance spatial distribution map, identifying floating debris areas, and calculating the ratio of the floating debris area area to the total water area to obtain the floating debris coverage rate. The four extracted indicators are integrated in a preset order to obtain the water body appearance indicator vector.

6. The method for water quality sensing, early warning, and management of park landscape water bodies according to claim 1, characterized in that, The water quality optical index vector obtained through inversion is used to construct a water body optical correlation model to establish the relationship between reflectivity and water quality optical indexes. The expression of the water body optical correlation model is as follows: ； in, For reflectivity, The reference reflectance of the clean water body in the corresponding wavelength band, The water absorption coefficient, The water scattering coefficient, Optical path length This is the error term; Methods for obtaining the water quality optical index vector by minimizing the objective function include: By minimizing the objective function, we find a set of chlorophyll concentration, turbidity, and chemical oxygen demand that minimizes the deviation between the reflectance calculated by the water body optical correlation model and the actual observed reflectance. The expression for the objective function is as follows: ； in, It is the sum of squares of the deviations between the reflectance calculated by the model and the actual observed reflectance. , , The regularization coefficient is . This is a regularization term.

7. The method for water quality sensing, early warning, and management of park landscape water bodies according to claim 6, characterized in that, The formula for calculating the consistency deviation between the water body appearance index vector and the water quality optical index vector is as follows: ； ； ； in, , , The preset weighting coefficients, The dispersion of water quality optical indicators. This represents the average value of each indicator in the water quality optical index. The standard deviation of each indicator is . This represents the dispersion of water body appearance indicators. This represents the average value of each indicator among the water body's appearance indicators. The standard deviation of each indicator is . is the correlation coefficient between the water body appearance index vector and the water quality optical index vector.

8. The method for water quality sensing, early warning, and management of park landscape water bodies according to claim 1, characterized in that, The management strategies output based on early warning messages include ecological priority strategies, landscape priority strategies, and ecological-landscape synergy strategies. The methods for selecting the corresponding strategies include: Calculate the overall score of water body appearance The calculation formula is as follows: ； in, , , , These are the weighting coefficients for each indicator. , , , The values ​​are, in order, the transparency attenuation depth, chromaticity coordinates, water surface uniformity, and floating object coverage. Set a judgment threshold, if If the water quality is below the judgment threshold and any of the water quality optical indicators does not exceed the warning threshold, then the landscape priority strategy is adopted. like If the water quality is below the judgment threshold and any one of the water quality optical indicators exceeds the warning threshold, then the ecological landscape synergy strategy will be selected. like If the water quality is not less than the judgment threshold and any one of the water quality optical indicators exceeds the warning threshold, then the ecological priority strategy shall be adopted.

9. A park landscape water quality sensing and early warning management system, characterized in that, include: The image acquisition unit is used to acquire multispectral images of the water surface of the park's water bodies and binocular stereo images to form a time-synchronized multi-view image sequence. The reflectance calculation unit is used to construct the three-dimensional spatial coordinates of the water surface based on the binocular stereo images in the multi-view image sequence, and to obtain the spatial distribution map of the water surface reflectance by mapping the reflectance of each band on the three-dimensional coordinates based on the multispectral images. The index vector calculation unit is used to extract the water body appearance index vector based on the spatial distribution map of water surface reflectance. The indexes in the water body appearance index vector include transparency attenuation depth, chromaticity coordinates, water surface uniformity, and floating debris coverage. A water body optical correlation model is constructed to establish the correlation between reflectance and water quality optical indicators, including chlorophyll concentration, turbidity, and chemical oxygen demand. The water quality optical index vector is obtained by minimizing the objective function and inverting the solution. The early warning judgment unit is used to calculate the consistency deviation value between the water body appearance index vector and the water quality optical index vector. A deviation threshold is preset. When the consistency deviation value is less than the deviation threshold, it is determined whether either the water body appearance index vector or the water quality optical index vector exceeds the preset early warning threshold. If it exceeds the threshold, an early warning message is output. The strategy management unit is used to output management strategies based on early warning messages.

10. A storage medium, characterized in that, It stores a computer program, which, when executed by a processor, implements the park landscape water quality perception and early warning management method as described in any one of claims 1-8.

Images