A method and system for identifying black and odorous water bodies based on image recognition processing

Abstract

The present invention relates to the field of image recognition and processing technology, and specifically to a method and system for identifying black and odorous water bodies based on image recognition and processing. The method comprises the following steps: based on black and odorous water body image data, adjusting the pixel value range by sampling, comparing the color channel distribution, extracting the gray value distribution characteristics, analyzing the partition brightness value, and generating a basic feature data set. In the present invention, by adjusting the pixel value range, extracting the gray value distribution characteristics and partition brightness value analysis, the characteristics of the water body pollution area are refined, the texture gradient change and the gray value distribution are analyzed, the pollution area boundary is divided by combining the cumulative analysis and edge intensity, the pollution type is classified by combining the geometric parameters, gray value and color channel changes, the diffusion direction and radius are analyzed by vector expansion, the coverage range is constructed by combining the density and path information, the cumulative pollution value is calculated, the pollution identification accuracy and dynamic tracking ability are enhanced, and the evaluation system is comprehensively improved.

Description

Technical Field

[0001] The present invention relates to the technical field of image recognition processing, and in particular to a method and system for identifying black and odorous water bodies based on image recognition processing. Background Art

[0002] The field of image recognition and processing technology uses computers to automatically identify and analyze the content in images or videos. This technology uses advanced algorithms such as deep learning, convolutional neural networks (CNNs), and pattern recognition to automatically process and understand images. Image recognition processing is widely used in various fields, including security monitoring, autonomous driving, industrial detection, etc. In image recognition, computers can learn to extract features from raw images, identify objects, classify, locate, detect, etc., and then realize automated decision-making or feedback. With the development of big data, cloud computing, and deep learning technology, the accuracy and application scope of image recognition processing have been continuously improved, promoting technological progress in various industries.

[0003] Among them, the black and odorous water body identification method is based on image recognition and processing technology, which aims to automatically identify and detect black and odorous water bodies in urban water bodies through deep learning and image processing technology. The main purpose of this method is to help environmental monitoring and governance departments to achieve early identification and continuous monitoring of black and odorous water bodies. Through high-precision identification of black and odorous water bodies, pollution sources can be discovered in time, water quality changes can be evaluated, and data support can be provided for water body governance measures. This method is of great significance in urban environmental protection, especially in promoting water quality improvement and protecting residents' health, and has broad application prospects.

[0004] When processing images of black and odorous water bodies, the existing technology only focuses on the overall color and texture changes, and does not adequately process pixel-level details, which can easily lead to the loss of pollution feature details and affect the accuracy of the data. In terms of edge detection and area division, the existing technology cannot accurately separate the boundaries of complex polluted areas, especially when multiple pollution types coexist, there are overlaps or misjudgments in boundary division. In pollution diffusion analysis, due to the lack of vectorized diffusion direction identification means, the dynamic tracking ability of the diffusion path is limited, resulting in distorted predictions of the distribution pattern of pollutants. In the assessment of pollution coverage, the existing technology is mainly based on simple area calculations, and is unable to combine distribution density and path information for comprehensive analysis, making it difficult to reflect the multi-level state of pollution, limiting the application of existing technology in coping with complex pollution environments, and easily resulting in insufficient accuracy in pollution identification and one-sidedness in assessment results. Summary of the invention

[0005] In order to solve the problem that the existing technology only focuses on the overall color and texture changes when processing black and odorous water images, and does not process pixel-level details enough, which easily leads to the loss of pollution feature details and affects the accuracy of the data. In terms of edge detection and area division, the existing technology cannot accurately separate the boundaries of complex pollution areas, especially when multiple pollution types coexist, there are overlaps or misjudgments in boundary division. In pollution diffusion analysis, due to the lack of vectorized diffusion direction identification means, the dynamic tracking ability of the diffusion path is limited, resulting in distorted predictions of the distribution law of pollutants. In the assessment of pollution coverage, the existing technology is mainly based on simple area calculations, and it is impossible to combine distribution density and path information for comprehensive analysis, and it is difficult to reflect the multi-level state of pollution, which limits the application of existing technology in dealing with complex pollution environments. It is easy to cause technical problems such as insufficient accuracy of pollution identification and one-sided evaluation results. The embodiment of the present invention provides a method and system for identifying black and odorous water bodies based on image recognition processing. The technical solution is as follows:

[0006] On the one hand, a method for identifying black and odorous water bodies based on image recognition processing is provided, the method comprising:

[0007] S1: Based on the black and odorous water image data, the pixel value range is adjusted by sampling, the color channel distribution is compared, the gray value distribution characteristics are extracted, the partition brightness value is analyzed, and the basic feature data set is generated;

[0008] S2: Based on the basic feature data set, analyzing the pixel gradient variation in the texture direction, extracting the neighborhood grayscale value directional distribution, accumulating the texture direction variation trend, dividing the pollution area boundary, and generating the pollution boundary characteristic parameter set;

[0009] S3: According to the pollution boundary characteristic parameter set, the morphological geometric parameters of the pollution area are measured, the target areas with the same morphology are screened, the pollution types are classified, and a pollution type information table is generated;

[0010] S4: using the location of the polluted area in the pollution type information table, analyzing the texture change rate and the color value similarity, identifying the diffusion direction and radius between regions, analyzing the distribution density of pollutants within the path range, measuring the distance and density between points in the polluted area, counting the coverage area, and constructing a pollution coverage range data set;

[0011] S5: Call the pollution coverage data set, measure the total coverage area, evaluate the pollutant density, combine the diffusion direction data, calculate the cumulative pollution value and coverage ratio, and generate the black and odorous water pollution assessment result.

[0012] Optionally, the basic feature data set includes pixel grayscale value distribution characteristics, brightness differentiation analysis results, and color channel ratio parameters; the pollution boundary characteristic parameter set includes boundary area intensity value, gradient change characteristic parameters, and texture direction cumulative data; the pollution type information table includes pollution area morphological parameters, grayscale value classification results, and color characteristic distribution data; the pollution coverage range data set includes distribution density statistics, point spacing parameters, and pollution coverage area values; the black and odorous water pollution assessment results include total coverage area, total pollutant density, diffusion cumulative value, and coverage ratio.

[0013] Optionally, based on the black and odorous water image data, the steps of adjusting the pixel value range by sampling, comparing the color channel distribution, extracting the gray value distribution characteristics, analyzing the partition brightness value, and generating the basic feature data set are specifically as follows:

[0014] S101: Based on the black and odorous water body image data, extract the RGB color channel value of each pixel point, convert the RGB value into a gray value through a weighted formula, calculate the gray histogram of the entire image and record the gray distribution, and generate a gray value distribution data set;

[0015] S102: Divide the image into multiple analysis areas according to the gray value distribution data set, perform statistical analysis on the gray histogram of each area, calculate the average gray value and gray deviation in the area, and obtain regional brightness characteristic data;

[0016] S103: Based on the regional brightness characteristic data, perform brightness difference analysis on the image, compare the brightness distribution of each partition, record the brightness comparison data between the differentiated regions, and combine the data to generate a basic feature data set.

[0017] Optionally, based on the basic feature data set, analyzing the pixel gradient variation in the texture direction, extracting the neighborhood gray value directional distribution, accumulating the texture direction variation trend, dividing the pollution area boundary, and generating the pollution boundary characteristic parameter set are specifically as follows:

[0018] S201: Based on the basic feature data set, for each pixel, analyze the grayscale difference between adjacent pixels, determine the gradient direction and magnitude of each pixel, compare the grayscale values ​​of eight pixels around each pixel, record the gradient angle and magnitude, and generate pixel gradient direction data;

[0019] S202: using the pixel gradient direction data, performing direction statistics on the pixel points in each region, and analyzing key trends and patterns by aggregating the gradient directions in the same region, including frequency analysis of the gradient directions and evaluation of direction consistency, to obtain texture direction distribution characteristic data;

[0020] S203: Based on the texture direction distribution characteristic data and in combination with edge strength analysis, the boundary line of the polluted area is located by comparing the gradient strength and direction consistency of the area, and the high-intensity edge area is marked and connected to generate a polluted boundary characteristic parameter set.

[0021] Optionally, according to the pollution boundary characteristic parameter set, the morphological geometric parameters of the pollution area are measured, target areas of the same morphology are screened, pollution types are classified, and the steps of generating the pollution type information table are specifically as follows:

[0022] S301: based on the pollution boundary characteristic parameter set, extract the contour coordinates of the pollution area, measure the boundary length and calculate the area of ​​the area through boundary point interpolation, convert the measured data into morphological parameters, identify the regional morphological features, and obtain the regional morphological geometric parameter table;

[0023] S302: Based on the regional morphological geometric parameter table, the regional numbers in the table are compared with the morphological characteristic parameters, and regions with similar characteristics are screened by setting a numerical difference threshold to obtain a target region set;

[0024] S303: Based on the target area set, the K-means clustering algorithm is used to extract the grayscale value and color channel data in the target area, and the grayscale gradient change and color distribution direction analysis classification of the pixel position in the area are used to match the classification rules and assign the pollution type. The pollution type information table is generated by combining the area number and the pollution type.

[0025] Optionally, the formula of the K-means clustering algorithm is as follows:

[0026]

[0027] Calculates the distance between features in a region ;

[0028] in, Represents the average value of the gray value change rate of pixels in the target area. Represents the standard deviation of the directionality of the color channel distribution of pixels in the target area. Represents the overall deviation of the pixel position in the target area, is the grayscale change weight coefficient, is the color distribution weight coefficient, is the pixel position deviation weight coefficient.

[0029] Optionally, the steps of using the pollution area position in the pollution type information table, analyzing the texture change rate and color value similarity, identifying the diffusion direction and radius between regions, analyzing the distribution density of pollutants within the path range, measuring the distance and density between pollution area points, and counting the coverage area to construct the pollution coverage range data set are specifically as follows:

[0030] S401: Based on the location of the polluted area in the pollution type information table, determine the boundary points of each area, determine the change amplitude of the gray value change rate of adjacent boundary points, analyze the mean difference of color values, evaluate the color similarity of adjacent areas, mark the boundary vector and extract the diffusion direction, and obtain the pollution diffusion direction vector data;

[0031] S402: Based on the pollution diffusion direction vector data and the morphological feature data of the pollution area, the coordinates of the pollution points within the path range are extracted, the spacing within the path and the density of points within the unit path are analyzed, the path and distribution density data are integrated, the distribution of points within the area is marked, and the pollutant distribution density data is obtained;

[0032] S403: Based on the pollutant distribution density data, the coverage radius of the polluted area is identified, and the coverage range is corrected by gradually superimposing the point coverage range and combining the diffusion path direction to generate a pollution coverage range data set.

[0033] Optionally, the steps of calling the pollution coverage data set, measuring the total coverage area, evaluating the pollutant density, combining the diffusion direction data, calculating the cumulative pollution value and the coverage ratio, and generating the black and odorous water pollution assessment result are as follows:

[0034] S501: Based on the pollution coverage data set, extract the boundary coordinates of each pollution area, calculate the area of ​​the area by connecting the coordinate points, accumulate the area values ​​item by item, and perform data number correspondence and format integration to obtain the total area data of the coverage range;

[0035] S502: Based on the total area data of the coverage area, extract the area value of the polluted area, combine the regional pollutant distribution density, summarize the total amount of pollutants data by number, evaluate the relationship between the total amount of pollutants and the distribution density in sections, and obtain pollutant density evaluation data;

[0036] S503: Based on the pollutant density assessment data, the weighted distribution density accumulation method is used, combined with the pollution path diffusion direction data, to analyze the pollutant accumulation value of the path coverage area section by section, and the accumulated pollution value of the weighted relationship between the coverage area and the pollutant density distribution is statistically calculated. The pollutant density and area ratio within the path coverage area are integrated to construct the black and odorous water pollution assessment result.

[0037] Optionally, the formula of the weighted distribution density accumulation method is as follows:

[0038]

[0039] Calculate the pollution accumulation value of the area covered by the pollution path ,in, Represents the path coverage area The area of ​​the segment, Represents the path coverage area The pollutant density of the segment, Representative The weight coefficient of the segment, Representative The diffusion coefficient of the segment, is the number of path segments.

[0040] On the other hand, a black and odorous water body recognition system based on image recognition processing is provided, and the black and odorous water body recognition system based on image recognition processing is used to perform the above-mentioned black and odorous water body recognition method based on image recognition processing, and the system includes:

[0041] The pixel distribution characteristic analysis module analyzes the red, green and blue channel values ​​of each pixel based on the black and odorous water image data, analyzes the distribution of channel values, groups the pixels according to the gray value interval, and performs statistical analysis on the brightness mean of the group. It analyzes the change trend of the pixel gradient value in the texture direction, combines the group brightness mean with the gradient change for mapping, and generates a partition texture brightness feature set.

[0042] The polluted area boundary analysis module screens the distribution interval of the brightness value and identifies the boundary points based on the partition texture brightness feature set, extracts the spatial coordinates of the gradient jump points, calculates the connectivity and position change between the boundary points, extracts the closed area through the connectivity parameters, analyzes the geometric properties of the closed area, extracts the contour characteristics and fits the morphological data, and generates a set of polluted area contour parameters;

[0043] The pollution form and type discrimination module analyzes the geometric attribute data of the pollution area based on the pollution area contour parameter set, extracts the variation range of the internal pixel grayscale value and the color channel value, analyzes the grayscale and color variation combination of the differentiated area, classifies according to the category characteristics of the geometric parameters and the internal variation parameters, summarizes the category labels and marks the pollution area, and obtains the pollution form and type annotation data;

[0044] The pollution diffusion and scope estimation module extracts the diffusion direction and path vectors between adjacent areas based on the pollution morphology and type annotation data, estimates the coverage range in combination with the pollutant diffusion radius, counts the pollution density parameters within the path and accumulates the area of ​​the coverage range, summarizes the distribution intensity and area information of the polluted area, and generates a black and odorous water pollution assessment result.

[0045] The beneficial effects brought about by the technical solution provided by the embodiment of the present invention include at least:

[0046] By adjusting the pixel value range and extracting the gray value distribution characteristics, combined with the differential partition brightness value analysis, the basic data processing capabilities of water body images are strengthened, making the characteristics of water pollution areas more refined and suitable for subsequent analysis. When analyzing the gradient change in texture direction and the gray value distribution in direction, the cumulative analysis is combined with the edge strength to accurately divide the boundaries of the pollution area and improve the level of detail in the identification of the pollution area. On the basis of measuring the morphological geometric parameters, the classification is carried out by combining the internal gray value and the color channel change, making the pollution type identification more diversified and accurate. In the process of identifying the regional diffusion direction and radius, the vector expansion method is combined with the texture and color characteristics of the adjacent areas to solve the problem of fuzziness of traditional technologies in the identification of pollutant diffusion direction. On the basis of the analysis of pollutant distribution density and coverage area, the path information is superimposed to construct the pollution coverage range, and the cumulative pollution value is calculated in combination with the diffusion path, making the pollution assessment results more comprehensive. This processing logic of gradual accumulation and refinement analysis significantly enhances the accuracy of pollution identification, the dynamic tracking ability of pollutant diffusion, and the systematic nature of the overall pollution assessment, meeting the diverse needs of water pollution identification in complex environments. BRIEF DESCRIPTION OF THE DRAWINGS

[0047] In order to more clearly illustrate the technical solutions in the embodiments of the present invention, the following briefly introduces the drawings required for use in the description of the embodiments. Obviously, the drawings described below are only some embodiments of the present invention. For ordinary technicians in this field, other drawings can be obtained based on these drawings without creative work.

[0048] Figure 1 It is a schematic diagram of the workflow of the present invention;

[0049] Figure 2 This is a detailed flow chart of S1 of the present invention;

[0050] Figure 3 This is a detailed flow chart of S2 of the present invention;

[0051] Figure 4 This is a detailed flow chart of S3 of the present invention;

[0052] Figure 5 This is a detailed flow chart of S4 of the present invention;

[0053] Figure 6 This is a detailed flow chart of S5 of the present invention;

[0054] Figure 7 It is a system flow chart of the present invention. DETAILED DESCRIPTION

[0055] The technical solution of the present invention is described below in conjunction with the accompanying drawings.

[0056] In the embodiments of the present invention, words such as "exemplarily" and "for example" are used to indicate examples, illustrations or explanations. Any embodiment or design described as "example" in the present invention should not be interpreted as being more preferred or more advantageous than other embodiments or designs. Specifically, the use of the word "example" is intended to present the concept in a specific way. In addition, in the embodiments of the present invention, the meaning expressed by "and / or" can be both, or it can be either of the two.

[0057] In order to make the technical problems, technical solutions and advantages to be solved by the present invention more clear, a detailed description will be given below with reference to the accompanying drawings and specific embodiments.

[0058] See also Figure 1 The embodiment of the present invention provides a method for identifying black and odorous water bodies based on image recognition processing, comprising the following steps:

[0059] S1: Based on the black and odorous water image data, the range of each pixel value is adjusted by sampling, the color channel distribution in the image is compared, the distribution characteristics of the pixel gray value are extracted, and the differential partition brightness value is analyzed to generate a basic feature data set;

[0060] S2: Based on the basic feature data set, by analyzing the gradient change of each pixel in the texture direction, the directional distribution characteristics of the neighborhood grayscale value are extracted, and the trend of texture direction change is cumulatively analyzed. The boundary of the polluted area is divided according to the edge intensity value to generate a pollution boundary characteristic parameter set;

[0061] S3: According to the pollution boundary characteristic parameter set, the morphological geometric parameters of each pollution area are measured, the target areas with similar morphological characteristics are screened, and the pollution type is classified according to the internal gray value and color channel change trend to generate a pollution type information table;

[0062] S4: Using the location of the polluted area in the pollution type information table, analyze the texture change rate and color value similarity of adjacent areas, perform vector recognition on the diffusion direction between regions, determine the diffusion direction and diffusion radius of the pollutants, and analyze the distribution density of pollutants within the path range in combination with the morphological feature data. By measuring the distance and density between points in the polluted area, estimate the pollution coverage range, and count the area of ​​the polluted area by superimposing the path and coverage information, and construct a pollution coverage range data set;

[0063] S5: Call the pollution coverage data set, measure the total area of ​​the coverage, evaluate the pollutant density in the coverage area, combine the pollution path diffusion direction data, calculate the cumulative pollution value and coverage ratio, and generate the black and odorous water pollution assessment results.

[0064] The basic feature data set includes pixel grayscale value distribution characteristics, brightness differentiation analysis results, and color channel ratio parameters. The pollution boundary characteristic parameter set includes boundary area intensity value, gradient change characteristic parameters, and texture direction cumulative data. The pollution type information table includes pollution area morphological parameters, grayscale value classification results, and color characteristic distribution data. The pollution coverage range data set includes distribution density statistics, point spacing parameters, and pollution coverage area values. The black and odorous water pollution assessment results include total coverage area, total pollutant density, diffusion accumulation value, and coverage ratio.

[0065] Specifically, if Figure 2 As shown, based on the black and odorous water image data, the steps of adjusting the pixel value range by sampling, comparing the color channel distribution, extracting the gray value distribution characteristics, analyzing the partition brightness value, and generating the basic feature data set are as follows:

[0066] S101: Based on the black and odorous water body image data, extract the RGB color channel value of each pixel point, convert the RGB value into a gray value through a weighted formula, calculate the gray histogram of the entire image and record the gray distribution, and generate a gray value distribution data set;

[0067] In the analysis of black and odorous water image data, we first need to extract the RGB color channel value of each pixel. The RGB color mode consists of three basic colors: red, green, and blue. For each pixel, the RGB value represents the intensity of the three channels, red, green, and blue. In order to convert the RGB value into a grayscale value, a weighted formula is used. The common weighted formula is:

[0068]

[0069] Among them, R, G, and B represent the red, green, and blue color channel values ​​of each pixel in the image. For example, assuming that the RGB value of a pixel is R=120, G=150, and B=200, then the grayscale value of the pixel can be calculated according to the above formula:

[0070]

[0071]

[0072] Therefore, the grayscale value of this pixel is 146.718;

[0073] In this way, the RGB value of each pixel can be converted into a corresponding grayscale value to obtain a set of grayscale values ​​for the entire image. The grayscale histogram of the entire image can be counted, that is, the frequency of each grayscale value in the image is recorded. For example, if the image contains 10,000 pixels and the pixel with a grayscale value of 146.718 appears 300 times, then the frequency of the grayscale value 146.718 is 300 / 10,000=0.03. By counting the frequency of occurrence of all grayscale values, the grayscale distribution of the image can be understood, providing a basis for subsequent image analysis.

[0074] After completing this conversion, the generated grayscale histogram and distribution data set will provide an effective basis for analyzing the brightness characteristics of different areas in the image. The results of the process will provide the required basic data for subsequent operations such as image region division and brightness difference analysis.

[0075] S102: Divide the image into multiple analysis areas according to the gray value distribution data set, perform statistical analysis on the gray histogram of each area, calculate the average gray value and gray deviation in the area, and obtain regional brightness characteristic data;

[0076] The image can be divided into multiple analysis areas. The division of the areas is based on different grayscale value intervals to ensure that the characteristics of each area in the image are fully reflected. The grayscale histogram of each area is statistically analyzed. By calculating the average grayscale value and grayscale deviation in each area, the brightness characteristic data of the area can be obtained. The average grayscale value refers to the average of the grayscale values ​​of all pixels in the area, which represents the overall brightness level of the area. The grayscale deviation reflects the discrete degree of pixel brightness in the area. The larger the deviation, the more drastic the brightness change in the area. This step is of great significance for further refining the analysis of image brightness distribution characteristics, and can help identify brightness differences in different areas.

[0077] S103: Based on the regional brightness characteristic data, perform brightness difference analysis on the image, compare the brightness distribution of each partition, record the brightness comparison data between the differentiated regions, and combine the data to generate a basic feature data set;

[0078] The image can be analyzed for brightness difference. By comparing the brightness distribution of each partition, the area with large brightness difference in the image can be identified. Brightness difference analysis can help reveal potential abnormal areas in the image, such as pollution sources or uneven brightness distribution in black and smelly water bodies. By comparing the brightness data between each area, the brightness comparison data between the differentiated areas can be recorded. The data will be combined into a basic feature data set for subsequent image feature extraction and further analysis. This data set is of great value for identifying and analyzing key areas in the image, especially in environmental monitoring and pollution analysis.

[0079] Specifically, if Figure 3 As shown, based on the basic feature data set, the steps of analyzing the pixel gradient change in the texture direction, extracting the neighborhood gray value direction distribution, accumulating the texture direction change trend, dividing the pollution area boundary, and generating the pollution boundary characteristic parameter set are as follows:

[0080] S201: Based on the basic feature data set, for each pixel, analyze the grayscale difference between adjacent pixels, determine the gradient direction and magnitude of each pixel, compare the grayscale values ​​of the eight pixels around each pixel, record the gradient angle and magnitude, and generate pixel gradient direction data;

[0081] Compare the grayscale values ​​of the current pixel with those of the eight pixels around it, and further clarify the direction of the pixel grayscale change by calculating the grayscale difference between the current pixel and the adjacent pixels. After selecting a pixel point, calculate the difference with the grayscale values ​​of the eight pixels above, below, left, right and diagonally. Determine the angle of the pixel gradient by judging the positive and negative directions of the grayscale value change, and divide the angle into a range of 0-180 degrees. At the same time, use the absolute value of the grayscale difference as an indicator of the gradient amplitude. The gradient direction angle and amplitude of each pixel point are stored in a matrix. The matrix dimension is consistent with the input image. By scanning the complete image multiple times, the gradient direction and amplitude data of all pixels are accumulated to generate comprehensive pixel gradient direction data to provide support for subsequent steps.

[0082] S202: using pixel gradient direction data, performing direction statistics on pixel points in each region, and by aggregating gradient directions in the same region, analyzing key trends and patterns, including frequency analysis of gradient directions and evaluation of direction consistency, to obtain texture direction distribution characteristic data;

[0083] First, according to the boundary range of the region, the pixel gradient direction data is divided into multiple subsets by region. The frequency of the direction data in each subset is counted, the proportion of different directional gradients is calculated, and the direction with the highest proportion is recorded and marked as the main direction trend in the region. At the same time, the consistency of the gradient direction in the region is evaluated, and the consistency index is obtained by calculating the difference between the main direction and the direction. When counting the direction frequency, the direction is divided into several ranges at fixed angle intervals, such as 0-20 degrees, 20-40 degrees, etc., and the pixel direction data is classified and accumulated. The final result includes the main direction trend in the region, the frequency distribution of each direction, and the consistency index, laying the foundation for the subsequent extraction of texture direction distribution characteristic data.

[0084] S203: Based on the texture direction distribution characteristic data and combined with edge strength analysis, the boundary line of the polluted area is located by comparing the gradient strength and direction consistency of the area, and the high-intensity edge area is marked and connected to generate a pollution boundary characteristic parameter set;

[0085] By comparing the gradient strength and directional consistency region by region, the boundary line of the polluted area can be accurately located. During the execution process, the gradient intensity data of each region is thresholded, the pixels with intensity values ​​higher than the set threshold are marked, and the points are connected to form a preliminary boundary line. In order to improve the accuracy of the boundary, the directional consistency of the pixels around the boundary line is further analyzed, the continuity and smoothness of the boundary are confirmed, and the discontinuous or noisy areas are corrected and filled. During the boundary analysis process, the texture directional distribution characteristics and edge strength of each region jointly determine the final shape of the polluted boundary. The generated polluted boundary characteristic parameter set includes indicators such as boundary position coordinates, boundary line length, and average intensity. The parameters are organized into a structured data set for subsequent pollution range analysis and result display.

[0086] Specifically, if Figure 4 As shown in the figure, according to the pollution boundary characteristic parameter set, the morphological geometric parameters of the pollution area are measured, the target areas of the same morphology are screened, the pollution types are classified, and the steps of generating the pollution type information table are as follows:

[0087] S301: based on the pollution boundary characteristic parameter set, extract the contour coordinates of the pollution area, measure the boundary length and calculate the area of ​​the area through boundary point interpolation, convert the measured data into morphological parameters, identify the regional morphological features, and obtain the regional morphological geometric parameter table;

[0088] By traversing the coordinate data of the boundary points one by one, the interpolation method is used to calculate the continuous curve between two adjacent boundary points to fill the gap caused by the sampling boundary points, and the length of the boundary curve after interpolation is measured segment by segment, and the overall boundary length of the polluted area is accumulated. By scanning all the pixel points within the boundary, the total number of pixels covered is gradually counted, and the actual physical area is converted according to the pixel spacing. In order to fully describe the morphological characteristics of the polluted area, geometric characteristic parameters such as the ratio of perimeter to area, circularity, and aspect ratio are further calculated, and the results are associated with the contour morphology of the area to form a regional morphological geometric parameter table. Each parameter corresponds to the area number and is recorded in the table, laying the foundation for subsequent feature comparison and classification between regions.

[0089] S302: Based on the regional morphological geometry parameter table, the regional numbers in the table are compared with the morphological characteristic parameters, and regions with similar characteristics are screened by setting a numerical difference threshold to obtain a target region set;

[0090] By setting the numerical difference threshold, we screen areas with similar morphological characteristics. First, we select a target area as a reference object, calculate the difference with the parameter value of the area one by one, and record the area number with the difference value lower than the set threshold as a candidate target. For multiple candidate target areas, we further verify whether the areas can be classified into the same set through multi-dimensional homogeneity analysis of parameters, such as shape consistency and area ratio. Based on the screening and verification results, we generate a target area set, which contains the number, morphological characteristics and homogeneity analysis results of each area, to provide support for the identification of pollution types in subsequent steps.

[0091] S303: Based on the target area set, the K-means clustering algorithm is used to extract the grayscale value and color channel data in the target area, and the grayscale gradient change and color distribution direction analysis classification of the pixel position in the area are used to match the classification rules and assign the pollution type, and the pollution type information table is generated by combining the area number and the pollution type;

[0092] The formula of K-means clustering algorithm is as follows:

[0093]

[0094] Calculates the distance between features in a region ;

[0095] in, Represents the average value of the gray value change rate of pixels in the target area. Represents the standard deviation of the directionality of the color channel distribution of pixels in the target area. Represents the overall deviation of the pixel position in the target area, is the grayscale change weight coefficient, is the color distribution weight coefficient, is the pixel position deviation weight coefficient;

[0096] Detailed explanation of the formula and the process of formula calculation and derivation:

[0097] Grayscale change rate ( ): It is obtained by calculating the average value of the change rate of the grayscale values ​​of all pixels in the target area, calculating the grayscale difference between adjacent pixels, obtaining the grayscale gradient, and obtaining the average value of these gradients as the grayscale change rate. For example, if the grayscale gradient values ​​in a target area are 10, 15, and 20, then ;

[0098] Color channel distribution directionality ( ): Obtained by calculating the standard deviation of the color channel distribution direction of all pixels in the target area, converting the color channel (such as RGB) into an appropriate color space (such as HSV), extracting the hue component, and calculating the standard deviation of the hue value as the color distribution direction. For example, if the hue values ​​in a target area are 30, 35, and 40, then ;

[0099] Pixel position deviation ( ): Obtained by calculating the overall deviation of all pixel positions in the target area, determining the center of gravity of the target area, calculating the distance from each pixel to the center of gravity, and taking the average of these distances as the pixel position deviation For example, if the distance from the pixel point to the centroid in a target area is 2, 3, 4, then ;

[0100] Weight coefficient ( ): The coefficient is used to adjust the influence of each feature on the distance calculation. Its setting is based on the analysis of pollution type characteristics and experimental results. For example, if the grayscale change has a greater impact on the classification, you can set , color distribution is second, setting , pixel position has the least influence, setting , the value can be adjusted according to the specific situation and experimental results;

[0101] Substitute the above example values ​​into the formula to calculate the feature distance The process is as follows:

[0102]

[0103]

[0104]

[0105]

[0106]

[0107] The result shows that the characteristic distance of the target area is 10.98. In the K-means clustering algorithm, this distance is used to measure the similarity between the target area and the cluster center, so as to determine the type of pollution to which it belongs.

[0108] Specifically, if Figure 5As shown in the figure, using the location of the polluted area in the pollution type information table, analyzing the texture change rate and color value similarity, identifying the diffusion direction and radius between regions, analyzing the distribution density of pollutants within the path range, measuring the distance and density between points in the polluted area, and counting the coverage area, the steps of constructing the pollution coverage range data set are as follows:

[0109] S401: Based on the location of the polluted area in the pollution type information table, the boundary points of each area are determined, the change amplitude is determined by the gray value change rate of adjacent boundary points, the mean difference of the color value is analyzed, the color similarity of adjacent areas is evaluated, the boundary vector is marked and the diffusion direction is extracted, and the pollution diffusion direction vector data is obtained;

[0110] First, the specific coordinates of each polluted area are identified through the positioning information of each area, and then the grayscale value data of adjacent boundary points are collected. By analyzing the change amplitude of the data, the mean difference of the color value is calculated. The execution of this step requires precise image processing technology and edge detection algorithm to ensure the accuracy of the data. Next, the color similarity of adjacent areas is evaluated, the boundary vector is marked, and the diffusion direction is extracted. The image analysis software and technology used in the process ensure the reliability of the results. Finally, the pollution diffusion direction vector data is obtained to ensure the efficiency of monitoring and management of polluted areas.

[0111] S402: Based on the pollution diffusion direction vector data and the morphological feature data of the pollution area, the coordinates of the pollution points within the path range are extracted, the spacing within the path and the density of points within the unit path are analyzed, the path and distribution density data are integrated, the distribution of points within the area is marked, and the pollutant distribution density data is obtained;

[0112] First, determine the morphological characteristics of the polluted area, then conduct a detailed analysis of the coordinates of the polluted points within the path range, and perform a distribution analysis on the spacing within the path and the density of points within the unit path. Through data integration, the distribution of points within the area can be identified. The data analysis technology and methods involved in this process must be accurate and reliable to ensure the accuracy of the obtained pollutant distribution density data. The statistical methods and data processing techniques used in the process must meet the standards of environmental monitoring.

[0113] S403: Based on the pollutant distribution density data, the coverage radius of the polluted area is identified, and the coverage range is corrected by gradually superimposing the point coverage range and combining the diffusion path direction to generate a pollution coverage range data set;

[0114] Based on the pollutant distribution density data, identify the coverage radius of the polluted area according to the formula:

[0115]

[0116] Calculate the coverage radius, where Represents the coverage radius of the pollution area, Represents the total contaminated area;

[0117] Considering that the contaminated area can be approximated as a circular area, according to the area formula , the pollution coverage radius can be derived The formula is used to set the measured pollution area is 452 square meters, then the pollution coverage radius The calculation is as follows:

[0118]

[0119] The results show that the pollution coverage radius is 12.0 meters, indicating that the pollution area is relatively concentrated, providing important data for environmental management and monitoring.

[0120] Specifically, if Figure 6 As shown in the figure, the steps of calling the pollution coverage data set, measuring the total coverage area, evaluating the pollutant density, combining the diffusion direction data, calculating the cumulative pollution value and coverage ratio, and generating the black and odorous water pollution assessment results are as follows:

[0121] S501: Based on the pollution coverage data set, extract the boundary coordinates of each pollution area, calculate the area of ​​the area by connecting the coordinate points, accumulate the area values ​​item by item, and perform data number correspondence and format integration to obtain the total area data of the coverage range;

[0122] By traversing the coordinate data of the boundary points one by one, the interpolation method is used to calculate the continuous curve between two adjacent boundary points to fill the gap caused by the sampled boundary points. The length of the boundary curve after interpolation is measured segment by segment, and the overall boundary length of the polluted area is accumulated. By scanning all pixel points within the boundary, the total number of pixels covered is gradually counted, and the actual physical area is converted according to the pixel spacing. In order to fully describe the morphological characteristics of the polluted area, geometric feature parameters such as the ratio of perimeter to area, circularity, and aspect ratio are further calculated. The results are associated with the contour morphology of the area to form a regional morphological geometric parameter table. Each parameter corresponds to the regional number and is recorded in the table, laying the foundation for subsequent feature comparison and classification between regions.

[0123] S502: Based on the total area data of the coverage area, extract the area value of the polluted area, combine the regional pollutant distribution density, summarize the total amount of pollutants data by number, evaluate the relationship between the total amount of pollutants and distribution density in sections, and obtain pollutant density evaluation data;

[0124] By setting the numerical difference threshold, we screen areas with similar morphological characteristics. First, we select a target area as a reference object, calculate the difference with the parameter value of the area one by one, and record the area number with the difference value lower than the set threshold as a candidate target. For multiple candidate target areas, we further verify whether the areas can be classified into the same set through multi-dimensional homogeneity analysis of parameters, such as shape consistency and area ratio. Based on the screening and verification results, we generate a target area set, which contains the number, morphological characteristics and homogeneity analysis results of each area, to provide support for the identification of pollution types in subsequent steps.

[0125] S503: Based on the pollutant density assessment data, the weighted distribution density accumulation method is used, combined with the pollution path diffusion direction data, to analyze the pollutant accumulation value of the path coverage area section by section, and the accumulated pollution value of the weighted relationship between the coverage area and the pollutant density distribution is calculated. The pollutant density and area ratio within the path coverage area are integrated to construct the black and odorous water pollution assessment results;

[0126] The formula of weighted distribution density accumulation method is as follows:

[0127]

[0128] Calculate the pollution accumulation value of the area covered by the pollution path ,in, Represents the path coverage area The area of ​​the segment, Represents the path coverage area The pollutant density of the segment, Representative The weight coefficient of the segment, Representative The diffusion coefficient of the segment, is the number of path segments;

[0129] Detailed explanation of the formula and the process of formula calculation and derivation:

[0130] Determine the area of ​​each segment of the path coverage area :Use geographic information system (GIS) technology, remote sensing images or field measurement data to obtain the area of ​​each section of the pollution path coverage area, and set the first The area of ​​the section is 500 square meters;

[0131] Get the pollutant density of each segment :Through on-site sampling and laboratory analysis, the pollutant concentration in each area is determined and the The pollutant density of each segment is 200 mg / m2; determine the weight coefficient of each segment :According to the land use type, population density and other factors of each section, set the weight coefficient and set the The segment is a residential area, and the weight coefficient is set to 1.2; calculate the diffusion coefficient of each segment :Use the pollutant diffusion model, combine meteorological data (such as wind speed and direction) and terrain characteristics, calculate the diffusion coefficient of pollutants in each section, and set the first The diffusion coefficient of the segment is 0.8;

[0132] Substituting the above values ​​into the formula, calculate the Weighted pollutant accumulation value of the segment:

[0133] ;

[0134] Therefore, the The weighted pollutant accumulation value of the segment is approximately 107,280 mg;

[0135] Assume that there are 3 regions in total, and calculate the weighted pollutant accumulation value of each region:

[0136] Phase 1: 107,280 mg;

[0137] Section 2: Set the area to 600 square meters, the pollutant density to 180 mg / square meter, the weight coefficient to 1.0, and the diffusion coefficient to 0.9;

[0138] calculate:

[0139] ;

[0140] Section 3: Set the area to 400 square meters, the pollutant density to 220 mg / square meter, the weight coefficient to 1.1, and the diffusion coefficient to 0.85;

[0141] calculate:

[0142] ;

[0143] Add up the weighted pollutant accumulation values ​​of each segment to get the total:

[0144] ;

[0145] At the same time, calculate the sum of the products of the area of ​​each segment and the weight coefficient:

[0146] ;

[0147] Substitute the above results into the formula to calculate the weighted cumulative pollution assessment value:

[0148] ;

[0149] The results show that the weighted cumulative pollution assessment value within the path coverage area is 182.3 mg / m2, reflecting the combined impact of pollutant density and area ratio.

[0150] like Figure 7 As shown, a black and odorous water body identification system based on image recognition processing, the system includes:

[0151] The pixel distribution characteristic analysis module analyzes the red, green and blue channel values ​​of each pixel based on the black and odorous water image data, analyzes the distribution of channel values, groups the pixels according to the gray value interval, and performs statistical analysis on the brightness mean of the group. It analyzes the change trend of the pixel gradient value in the texture direction, combines the group brightness mean with the gradient change for mapping, and generates a partition texture brightness feature set.

[0152] The pollution area boundary analysis module is based on the partition texture brightness feature set, screens the distribution interval of brightness values ​​and identifies boundary points, extracts the spatial coordinates of gradient jump points, calculates the connectivity and position change between boundary points, extracts closed areas through connectivity parameters, analyzes the geometric properties of closed areas, extracts contour characteristics and fits morphological data to generate a set of pollution area contour parameters;

[0153] The pollution form and type discrimination module analyzes the geometric attribute data of the pollution area based on the pollution area contour parameter set, extracts the variation range of the internal pixel grayscale value and color channel value, analyzes the grayscale and color change combination of the differentiated area, classifies according to the category characteristics of the geometric parameters and internal variation parameters, summarizes the category labels and marks the pollution area, and obtains the pollution form and type annotation data;

[0154] The pollution diffusion and scope estimation module extracts the diffusion direction and path vectors between adjacent areas based on the pollution morphology and type annotation data, estimates the coverage range based on the pollutant diffusion radius, counts the pollution density parameters within the path and accumulates the area of ​​the coverage range, summarizes the distribution intensity and area information of the polluted area, and generates black and odorous water pollution assessment results.

[0155] The above are only specific embodiments of the present invention, but the protection scope of the present invention is not limited thereto. Any person skilled in the art who is familiar with the technical field can easily think of changes or substitutions within the technical scope disclosed by the present invention, which should be included in the protection scope of the present invention. Therefore, the protection scope of the present invention should be based on the protection scope of the claims.

Claims

1. A method for identifying black and odorous water bodies based on image recognition processing, characterized in that: The following steps are involved: Based on the black and odorous water image data, the pixel value range is adjusted by sampling, the color channel distribution is compared, the gray value distribution characteristics are extracted, the partition brightness value is analyzed, and the basic feature data set is generated; Based on the basic feature data set, the pixel gradient change in the texture direction is analyzed, the neighborhood gray value direction distribution is extracted, the texture direction change trend is accumulated, the pollution area boundary is divided, and the pollution boundary characteristic parameter set is generated; According to the pollution boundary characteristic parameter set, the morphological geometric parameters of the pollution area are measured, the target areas with the same morphology are screened, the pollution types are classified, and a pollution type information table is generated; Using the location of the polluted area in the pollution type information table, analyze the texture change rate and color value similarity, identify the diffusion direction and radius between regions, analyze the distribution density of pollutants within the path range, measure the distance and density between points in the polluted area, count the coverage area, and construct a pollution coverage data set; The steps of calling the pollution coverage data set, measuring the total coverage area, evaluating the pollutant density, combining the diffusion direction data, calculating the cumulative pollution value and the coverage ratio, and generating the black and odorous water pollution assessment result; based on the basic feature data set, analyzing the pixel gradient change in the texture direction, extracting the neighborhood gray value directional distribution, accumulating the texture direction change trend, dividing the pollution area boundary, and generating the pollution boundary characteristic parameter set are as follows: Based on the basic feature data set, for each pixel, analyze the grayscale difference between adjacent pixels, determine the gradient direction and magnitude of each pixel, compare the grayscale values ​​of eight pixels around each pixel, record the gradient angle and magnitude, and generate pixel gradient direction data; Using the pixel gradient direction data, the direction statistics of the pixels in each area are performed, and by aggregating the gradient directions in the same area, key trends and patterns are analyzed, including frequency analysis of the gradient directions and evaluation of the direction consistency, to obtain texture direction distribution characteristic data; Based on the texture direction distribution characteristic data and combined with edge strength analysis, the boundary line of the polluted area is located by comparing the gradient strength and direction consistency of the area, and the high-intensity edge area is marked and connected to generate a pollution boundary characteristic parameter set; The steps of using the pollution area position in the pollution type information table, analyzing the texture change rate and color value similarity, identifying the diffusion direction and radius between regions, analyzing the distribution density of pollutants within the path range, measuring the distance and density between pollution area points, and counting the coverage area to construct the pollution coverage range data set are as follows: Based on the location of the polluted area in the pollution type information table, the boundary points of each area are determined, the change amplitude is determined by the gray value change rate of adjacent boundary points, the mean difference of the color value is analyzed, the color similarity of adjacent areas is evaluated, the boundary vector is marked and the diffusion direction is extracted, and the pollution diffusion direction vector data is obtained; Based on the pollution diffusion direction vector data, combined with the morphological feature data of the pollution area, the coordinates of the pollution points within the path range are extracted, the spacing within the path and the density of points within the unit path are analyzed, the path and distribution density data are integrated, the distribution of points inside the area is marked, and the pollutant distribution density data is obtained; Based on the pollutant distribution density data, the coverage radius of the polluted area is identified, and the pollution coverage data set is generated by gradually superimposing the point coverage and correcting the coverage in combination with the diffusion path direction.

2. According to the method for identifying black and odorous water bodies based on image recognition processing according to claim 1, it is characterized in that: The basic feature data set includes pixel grayscale value distribution characteristics, brightness differentiation analysis results, and color channel ratio parameters; the pollution boundary characteristic parameter set includes boundary area intensity value, gradient change characteristic parameters, and texture direction cumulative data; the pollution type information table includes pollution area morphological parameters, grayscale value classification results, and color characteristic distribution data; the pollution coverage range data set includes distribution density statistics, point spacing parameters, and pollution coverage area values; the black and odorous water pollution assessment results include total coverage area, total pollutant density, diffusion cumulative value, and coverage ratio.

3. The method for identifying black and odorous water bodies based on image recognition processing according to claim 1 is characterized in that: Based on the black and odorous water image data, the steps of adjusting the pixel value range by sampling, comparing the color channel distribution, extracting the gray value distribution characteristics, analyzing the partition brightness value, and generating the basic feature data set are as follows: Based on the black and odorous water image data, the RGB color channel value of each pixel is extracted, the RGB value is converted to the gray value through the weighted formula, the gray histogram of the entire image is counted and the gray distribution is recorded to generate the gray value distribution data set; According to the grayscale value distribution data set, the image is divided into multiple analysis areas, the grayscale histogram of each area is statistically analyzed, the average grayscale value and grayscale deviation in the area are calculated, and the regional brightness characteristic data is obtained; Based on the regional brightness characteristic data, the image is analyzed for brightness difference, the brightness distribution of each partition is compared, the brightness contrast data between the differentiated regions is recorded, and the data is combined to generate a basic feature data set.

4. The method for identifying black and odorous water bodies based on image recognition processing according to claim 1 is characterized in that: According to the pollution boundary characteristic parameter set, the morphological geometric parameters of the pollution area are measured, the target areas of the same morphology are screened, the pollution types are classified, and the steps of generating the pollution type information table are specifically as follows: Based on the pollution boundary characteristic parameter set, the contour coordinates of the pollution area are extracted, the boundary length is measured and the area of ​​the area is calculated through boundary point interpolation, the measurement data is converted into morphological parameters, the regional morphological features are identified, and the regional morphological geometric parameter table is obtained; Based on the regional morphological geometric parameter table, the regional numbers in the table are compared with the morphological characteristic parameters, and regions with similar characteristics are screened by setting a numerical difference threshold to obtain a target region set; Based on the target area set, the K-means clustering algorithm is used to extract the grayscale value and color channel data in the target area. The grayscale gradient change and color distribution direction analysis classification of the pixel position in the area are used to match the classification rules and attribute the pollution type. The pollution type information table is generated by combining the area number and the pollution type.

5. The method for identifying black and odorous water bodies based on image recognition processing according to claim 4 is characterized in that: The formula of the K-means clustering algorithm is as follows: ； Calculates the distance between features in a region ; in, Represents the average value of the gray value change rate of pixels in the target area. Represents the standard deviation of the directionality of the color channel distribution of pixels in the target area. Represents the overall deviation of the pixel position in the target area, is the grayscale change weight coefficient, is the color distribution weight coefficient, is the pixel position deviation weight coefficient.

6. The method for identifying black and odorous water bodies based on image recognition processing according to claim 1 is characterized in that: The steps of calling the pollution coverage data set, measuring the total coverage area, evaluating the pollutant density, combining the diffusion direction data, calculating the cumulative pollution value and coverage ratio, and generating the black and odorous water pollution assessment results are as follows: Based on the pollution coverage data set, the boundary coordinates of each pollution area are extracted, the area of ​​the area is calculated by connecting the coordinate points, the area values ​​are accumulated item by item, and the data numbers are matched and formatted to obtain the total area data of the coverage; Based on the total area data of the coverage area, extract the area value of the polluted area, combine the regional pollutant distribution density, summarize the total amount of pollutants data by number, evaluate the relationship between the total amount of pollutants and distribution density in sections, and obtain pollutant density assessment data; Based on the pollutant density assessment data, the weighted distribution density accumulation method is adopted, combined with the pollution path diffusion direction data, the pollutant accumulation value of the path coverage area is analyzed section by section, the cumulative pollution value of the weighted relationship between the coverage area and the pollutant density distribution is statistically calculated, the pollutant density and area ratio within the path coverage area are integrated, and the black and odorous water pollution assessment results are constructed.

7. The method for identifying black and odorous water bodies based on image recognition processing according to claim 6 is characterized in that: The formula of the weighted distribution density accumulation method is as follows: ； Calculate the pollution accumulation value of the area covered by the pollution path ,in, Represents the path coverage area The area of ​​the segment, Represents the path coverage area The pollutant density of the segment, Representative The weight coefficient of the segment, Representative The diffusion coefficient of the segment, is the number of path segments.

8. A black and odorous water body identification system based on image recognition processing, characterized in that: According to a method for identifying black and odorous water bodies based on image recognition processing according to any one of claims 1 to 7, the system comprises: The pixel distribution characteristic analysis module analyzes the red, green and blue channel values ​​of each pixel based on the black and odorous water image data, analyzes the distribution of channel values, groups the pixels according to the gray value interval, and performs statistical analysis on the brightness mean of the group. It analyzes the change trend of the pixel gradient value in the texture direction, combines the group brightness mean with the gradient change for mapping, and generates a partition texture brightness feature set. The polluted area boundary analysis module screens the distribution interval of the brightness value and identifies the boundary points based on the partition texture brightness feature set, extracts the spatial coordinates of the gradient jump points, calculates the connectivity and position change between the boundary points, extracts the closed area through the connectivity parameters, analyzes the geometric properties of the closed area, extracts the contour characteristics and fits the morphological data, and generates a set of polluted area contour parameters; The pollution form and type discrimination module analyzes the geometric attribute data of the pollution area based on the pollution area contour parameter set, extracts the variation range of the internal pixel grayscale value and the color channel value, analyzes the grayscale and color variation combination of the differentiated area, classifies according to the category characteristics of the geometric parameters and the internal variation parameters, summarizes the category labels and marks the pollution area, and obtains the pollution form and type annotation data; The pollution diffusion and scope estimation module extracts the diffusion direction and path vectors between adjacent areas based on the pollution morphology and type annotation data, estimates the coverage range in combination with the pollutant diffusion radius, counts the pollution density parameters within the path and accumulates the area of ​​the coverage range, summarizes the distribution intensity and area information of the polluted area, and generates a black and odorous water pollution assessment result.

Images