Water surface floating object filtering data analysis system

Abstract

The invention relates to the technical field of floater filtering, in particular to a water surface floater filtering data analysis system which comprises a disturbance frequency extraction module, a target contour separation module, a pollution level judgment module, a space interception evaluation module and an efficiency index construction module. According to the invention, through analyzing the direction change rate and amplitude of the continuous motion vector of the pixel points in the water surface floating object image sequence, a disturbance area caused by water surface ripples or light changes is effectively filtered, the target identification precision is enhanced, and the target identification accuracy is improved by combining gradient changes and optical flow characteristic quantitative marks in the residual area. Accurate positioning of a floating object target area is achieved, the boundary contour of the floating object target area is determined, the floating object pollution degree is rapidly evaluated, the target distribution density in the unit area is corrected through the ratio of the image contour circumscribed rectangle scale to the floating object number, and the floating object filtering processing efficiency is comprehensively analyzed through the target average residence frame length and the unit area interception density. And a quantitative and operable efficiency evaluation index is formed.

Description

Technical Field

[0001] The present invention relates to the technical field of floating object filtering, in particular to a water surface floating object filtering data analysis system. Background Art

[0002] The field of floating debris filtration technology involves technical solutions for identifying, separating, intercepting, collecting, and removing various types of floating solid matter on liquid surfaces. This area encompasses the structural design of mechanical filtration devices, control mechanisms for diverting and sorting floating debris on the water surface, deployment of hydrodynamic auxiliary equipment, and digital processing methods such as image recognition, water quality monitoring, and data analysis. Technologies in this field are widely used in water environment management, urban water system maintenance, aquaculture purification, and water pollution prevention, significantly improving water cleanliness, reducing ecological damage, and enabling intelligent management.

[0003] The Surface Floating Object Filtration and Data Analysis System is an integrated system that automatically identifies and filters floating objects on the water surface, while simultaneously collecting and analyzing data. Its purpose is to efficiently, continuously, and unmannedly process floating objects on the water surface through visual perception and control. It also generates multi-dimensional data on the type, quantity, distribution, time, and space of floating objects, providing data support and operational basis for environmental management, governance decision-making, and pollution tracing.

[0004] Traditional analysis systems rely on visual perception and control to implement automated processing for the identification and filtering of floating objects on the water surface. However, in actual operations, they lack effective means to deal with pixel disturbances caused by ripples on the water surface and changes in lighting, which leads to misjudgments or missed judgments during identification, reducing the accuracy of floating object identification and subsequent filtering operations. Traditional systems usually make simple assessments of the pollution status of floating objects based on intuitive quantity and spatial distribution, without in-depth analysis of the apparent density or optical reflection characteristics of the pollutants themselves, making it difficult to comprehensively and accurately judge the degree of pollution, which in turn affects the targeted nature of pollution control. They do not comprehensively consider the relationship between target scale and spatial distribution density, making it difficult to accurately measure the actual spatial interception effect, causing the actual control effect to deviate from expectations. Summary of the Invention

[0005] The purpose of the present invention is to solve the shortcomings of the prior art and to propose a water surface floating object filtering data analysis system.

[0006] In order to achieve the above-mentioned purpose, the present invention adopts the following technical solution: a water surface floating object filtering data analysis system, the system comprising: The disturbance frequency extraction module obtains pixel points within a continuous time window in the water surface image sequence frame, performs attribution analysis on the direction change rate and amplitude value of each group of pixel vectors, excludes pixel groups within the image disturbance frequency threshold range, and generates disturbance frequency exclusion area data; The target contour separation module excludes the regional data based on the disturbance frequency, marks pixels whose gradient direction change rate is greater than the direction difference angle setting value and whose optical flow amplitude standard deviation is less than the amplitude fluctuation reference value, retains the regions with closed edge attributes, and generates a stable contour structure set of floating objects; The pollution level determination module calls the stable contour structure set of floating objects, obtains the average brightness of pixels in the corresponding area, and combines the apparent area and estimated mass value of the floating objects to calculate the pollution level of the floating objects and generate pollution level information; The spatial interception assessment module obtains the stable contour structure set of the floating objects, calculates the number of targets per unit area, extracts the circumscribed rectangle of the target's image contour, uses the diagonal length value as the scale feature, and performs correction using the scale value as the weighting factor to generate area interception density information.

[0007] As a further solution of the present invention, the disturbance frequency exclusion area data includes a non-periodic disturbance vector cluster, a frequency domain exclusion pixel index group and a frequency boundary outside judgment label; the floating object stable contour structure set is specifically a direction alienated edge set, a closed contour identification mapping and a continuous boundary graphic index; the pollution level information includes a light reflection intensity level index and an apparent density ratio group; the area interception density information is specifically a unit area target density value, a scale weighting coefficient and a standardized interception ratio.

[0008] As a further solution of the present invention, the disturbance frequency extraction module includes: The vector direction extraction submodule obtains pixel points within a continuous time window in the water surface image sequence frame, detects the grayscale value change of the corresponding position of the pixel point in the continuous frame, calculates the two-dimensional coordinate offset value of the pixel between adjacent frames, and performs differential processing on the offset value in the time dimension to extract the motion direction vector, calls the pixel motion direction vector result, and generates a pixel vector direction sequence value; The disturbance frequency screening submodule obtains the direction change amplitude and the number of change cycles of each group of pixel vectors in consecutive frames based on the pixel vector direction sequence value, determines whether the number of change cycles is between the lower limit threshold of the image disturbance frequency and the upper limit threshold of the image disturbance frequency, screens and marks the pixel labels that meet the judgment conditions, and generates a frequency interference pixel label set; The regional pixel exclusion submodule excludes the pixel positions corresponding to the label index in the image frame according to the frequency interference pixel label set, retains the available area of ​​the current frame image, and establishes disturbance frequency exclusion area data.

[0009] As a further solution of the present invention, the target contour separation module includes: The direction change extraction submodule excludes the regional data based on the disturbance frequency, obtains the grayscale value change and coordinate position offset of the remaining pixel area in continuous frames, calculates the gradient direction angle difference and the mean square error of the optical flow amplitude of the regional pixels, combines the gradient direction angle difference and the mean square error of the optical flow amplitude, and classifies them to establish the direction amplitude distribution interval value; The edge pixel marking submodule calls the direction amplitude distribution interval value, determines the pixel points whose gradient direction angle difference value is greater than the direction difference angle setting value and whose optical flow amplitude mean square error value is less than the amplitude fluctuation reference value, and records the position of the pixels that meet the dual conditions, and counts the spatial clustering distribution of the pixels in the image frame to obtain the edge feature aggregation value; The closed area screening submodule screens pixel clusters with closed boundary characteristics in the continuous area according to the edge feature aggregation value, calculates the connectivity index and edge closure degree of the boundary contour, and integrates the areas that meet the closed feature judgment criteria into a boundary structure group to generate a stable contour structure set of floating objects.

[0010] As a further solution of the present invention, the pollution level determination module includes: The reflectivity recognition submodule calls the stable contour structure set of floating objects, extracts the grayscale maximum value and grayscale average value per unit area of ​​each contour target area based on the pixel distribution range, determines whether the grayscale maximum per unit area is within a valid interval within the light reflectivity grade segment, and records the segment position as a grade location index to establish a reflectivity grade location value; The density value calculation submodule obtains the apparent area and mass estimation value of the corresponding target according to the reflectivity level positioning value, calculates the average brightness value per unit area as the brightness index, and obtains the pollution level mapping value by calculation; The level classification submodule divides the value range into pollution level distribution segments based on the pollution level mapping value, identifies the level number according to the pollution intensity type corresponding to each segment, counts the level number and the number of contour targets, and establishes pollution level information.

[0011] As a further solution of the present invention, the formula for calculating and obtaining the pollution level mapping value is specifically: ; in, Represents the pollution level mapping value, represents the normalized value of the estimated quality, represents the normalized value of the apparent area, Indicates the brightness value per unit area, represents the average brightness value of grayscale, Indicates the maximum grayscale value.

[0012] As a further solution of the present invention, the space interception assessment module includes: The contour area determination submodule obtains the stable contour structure set of the floating object, extracts the pixel boundary of each contour based on the image position of the target within the projection area of ​​the interception device, calculates the corresponding image projection area, partitions the actual projection area in the image and the image area within the interception area, counts the number of contours and records the blocks they are located in, and generates regional target statistics; The target density extraction submodule calculates the number of target contours within the unit image area based on the regional target statistics, extracts the image contour circumscribed rectangle corresponding to the contour, calculates the diagonal length, uses the length value as the image scale indicator of the target, and generates a scale distribution feature value; The weighted density correction submodule calls the scale distribution characteristic value, calculates the ratio of the number of targets in each unit area to the total number of targets, multiplies the product with the diagonal length as a weighting factor, calculates the interception adjustment density value per unit area, and generates area interception density information.

[0013] As a further embodiment of the present invention, the system further comprises: The efficiency index construction module calculates the difference between the average residence frame length and the average interception frame length of the target based on the area interception density information and the pollution level information, combined with the frame sequence residence length of each target in the filtering area, and calculates the unit area efficiency index value based on the unit area interception density to generate unit filtering effect evaluation information; The unit filtration effect evaluation information specifically refers to the average treatment response difference, the area interception correction factor and the time-effect harmonic parameter.

[0014] As a further solution of the present invention, the efficiency index building module includes: The resident difference determination submodule extracts the frame number of the same target in each frame based on the area interception density information and the pollution level information, calculates the difference between the minimum and maximum numbers as the resident frame length, obtains the minimum value of the corresponding frame number of the same target being intercepted, calculates the target resident frame length and the interception frame length, and generates a resident difference index value; The density value superposition submodule calls the residence difference index value, the level number corresponding to each pollution level, takes the normalized value of the level number value as the pollution factor weight, and jointly calculates the efficiency expression value of floating objects per unit area by combining the three parameters to obtain the unit area efficiency index value; The efficiency value generation submodule performs mapping statistics according to the unit area efficiency index value in the image space area according to the projection block of the filtering device, extracts the mean of the corresponding index value in the block and outputs it as the basis for regional filtering efficiency evaluation, and establishes unit filtering effect evaluation information.

[0015] As a further solution of the present invention, the formula for calculating the unit area efficiency index value is specifically: ; in, Indicates the unit area efficiency index value, represents the normalized value of the area interception density information, Indicates the normalized value of the pollution level number, represents the average length of the resident frame, Indicates the average intercept frame length.

[0016] Compared with the prior art, the advantages and positive effects of the present invention are: In the present invention, by analyzing the directional change rate and amplitude of the continuous motion vector of pixels in the image sequence of floating objects on the water surface, the disturbance area caused by water surface ripples or light changes is effectively filtered out, the target recognition accuracy is enhanced, and the gradient change and optical flow feature quantification mark in the remaining area are combined to achieve accurate positioning of the floating object target area and clarify its boundary contour, thereby improving the reliability of floating object contour extraction. Relying on the light reflectance peak value and area-mass estimation of the contour area, the degree of floating object pollution is quickly assessed, and the pertinence and accuracy of data analysis are improved. The target distribution density per unit area is corrected by using the scale of the circumscribed rectangle of the image contour and the ratio of the number of floating objects, making the spatial interception efficiency evaluation more objective. The floating object filtering processing efficiency is comprehensively analyzed by the average target residence frame length and the interception density per unit area, forming a quantitative and operational efficiency evaluation index, providing a clear and quantitative decision-making basis for environmental governance, and effectively enhancing the reliability and timeliness of environmental management and governance decision-making data. BRIEF DESCRIPTION OF THE DRAWINGS

[0017] 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.

[0018] Figure 1 is a system flow chart of the present invention; Figure 2 Schematic diagram of the system framework of the present invention; Figure 3 This is a flow chart of the disturbance frequency extraction module of the present invention; Figure 4 This is a flow chart of the target contour separation module of the present invention; Figure 5 This is a flow chart of the pollution level determination module of the present invention; Figure 6This is a flow chart of the space interception assessment module of the present invention; Figure 7 Flowchart of the efficiency index building module of the present invention. DETAILED DESCRIPTION

[0019] The technical solution of the present invention is described below in conjunction with the accompanying drawings.

[0020] 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 an "exemplary" in the present invention should not be interpreted as being preferred or advantageous over other embodiments or designs. Rather, the use of the word "exemplary" is intended to present concepts in a concrete manner. Furthermore, in the embodiments of the present invention, "and / or" can mean both or either of the two.

[0021] In the embodiments of the present invention, the terms "image" and "picture" may sometimes be used interchangeably. It should be noted that, when the distinction is not emphasized, the meanings they convey are the same. The terms "of," "corresponding," and "corresponding" may sometimes be used interchangeably. It should be noted that, when the distinction is not emphasized, the meanings they convey are the same.

[0022] In the embodiments of the present invention, sometimes a subscript such as W1 may be written as a non-subscript such as W1. When the difference is not emphasized, the meanings to be expressed are the same.

[0023] In order to make the technical problems, technical solutions and advantages to be solved by the present invention clearer, a detailed description will be given below with reference to the accompanying drawings and specific embodiments.

[0024] See also Figure 1 , Water surface floating object filtration data analysis system, the system includes: The disturbance frequency extraction module obtains pixel points within a continuous time window in the water surface image sequence frame, detects the vector motion direction of the pixel points in the continuous frames, and performs attribution analysis on the direction change rate and amplitude value of each group of pixel vectors to determine whether they are within the image disturbance frequency threshold range. Pixel groups within the frequency threshold range are excluded to generate disturbance frequency exclusion area data; The image disturbance frequency threshold usually refers to the periodic frequency generated by the change of pixel position in the video sequence frame. The trend of pixel position change in consecutive frames can be analyzed by Fourier transform to obtain its spectrum peak range, which serves as the upper and lower limits of the frequency range corresponding to the fluctuation interference. The target contour separation module excludes regional data based on the disturbance frequency, obtains the gradient direction change rate and optical flow amplitude standard deviation of the remaining pixel area, marks the pixels whose gradient direction change rate is greater than the direction difference angle setting value and whose optical flow amplitude standard deviation is less than the amplitude fluctuation reference value, compares the marked areas for spatial continuity, retains areas with closed edge properties, and generates a stable contour structure set of floating objects; The gradient direction change rate is the degree of angular change of the image gradient direction within a spatial neighborhood. It is often extracted using the Sobel or Scharr operator, and then the angular offset is calculated. The optical flow amplitude standard deviation measures the degree of change in pixel motion speed between consecutive frames and is usually obtained based on the Horn-Schunck or Farneback optical flow method. The pollution level determination module calls the stable contour structure set of floating objects. Based on the peak light reflectance per unit area of ​​the target in the area, it obtains the average pixel brightness of the corresponding area and combines the apparent area and estimated mass value of the floating objects to calculate the apparent density ratio. It determines the location of the target reflectance peak within the light reflectance level segment, calculates the floating object pollution level, and generates pollution level information. The peak value of the light reflectance per unit area can be calculated by the maximum value of the grayscale histogram. The apparent density ratio is the estimated mass divided by the apparent area, which can be inferred by referring to the force characteristics of the floating body in the static water state. The spatial interception assessment module obtains a stable outline structure set of floating objects, counts the number of identified floating object outlines based on the image projection area of ​​the outlines in the interception device's working area, calculates the number of targets per unit area, and extracts the circumscribed rectangle of the target image outline. The diagonal length value is used as the scale feature, and the ratio of the target number value to the total number of identified targets is calculated. The comparison value is corrected using the scale value as a weighting factor to generate area interception density information. The diagonal length of the rectangle circumscribing the image outline is the diagonal of the rectangle surrounding the two-dimensional outline of the floating object, which is often used to approximate the geometric size of the target and facilitate the standardization of measurement scale features. The efficiency index construction module calculates the difference between the average target residence frame length and the average interception frame length based on the area interception density information and pollution level information, combined with the frame sequence residence length of each target in the filtering area. Combined with the unit area interception density, the unit area efficiency index value is calculated to generate unit filtering effect evaluation information; Frame sequence dwell length refers to the number of image frames in which the target appears continuously in the filter area, measured in frames, and is used to evaluate the duration between the target being identified and being intercepted. The disturbance frequency exclusion area data includes non-periodic disturbance vector clusters, frequency domain exclusion pixel index groups and frequency boundary judgment labels. The floating object stable contour structure set is specifically the direction alienation edge set, closed contour identification mapping and continuous boundary graphic index. The pollution level information includes the light reflection intensity level index and apparent density ratio group. The area interception density information is specifically the unit area target density value, scale weighting coefficient and standardized interception ratio. The unit filtering effect evaluation information specifically refers to the average treatment response difference, area interception correction factor and time-effect harmonization parameter.

[0025] See also Figure 2 and Figure 3 ,The disturbance frequency extraction module includes a vector direction extraction submodule, a disturbance frequency ,screening submodule, and a regional pixel exclusion submodule; The vector direction extraction submodule obtains pixel points within a continuous time window in the water surface image sequence frame, detects the grayscale value change of the corresponding position of the pixel point in the continuous frame, calculates the two-dimensional coordinate offset value of the pixel between adjacent frames, and performs differential processing on the offset value in the time dimension to extract the motion direction vector, calls the pixel motion direction vector result, and generates a pixel vector direction sequence value; Obtain the pixel points in the continuous time window of the water surface image sequence frame. First, set a fixed sampling window in each frame of the image. For example, taking a 1920×1080 pixel image as an example, set the continuous time window to 10 frames and the acquisition frame rate to 30fps, which means that the continuous time window span is 0.33 seconds. Then, extract the grayscale value of each pixel point in a row and column cycle. The extraction method uses frame-by-frame comparison of the grayscale changes of the same coordinate points, records the difference in grayscale values ​​of each pixel point between frame t and frame t+1, and constructs an inter-frame grayscale change sequence. Set the grayscale value of pixel p to 128 in frame t and 122 in frame t+1, then the difference is -6. Similarly, obtain the grayscale change difference sequence within the entire window period, and then identify the position offset of the same pixel between adjacent frames, that is, according to the grayscale value change trend and the grayscale value difference, the grayscale value difference is obtained. Gradient edge detection calculates its offset step size in the x-axis and y-axis directions. For example, if point p shifts from (300, 200) to (302, 201), the x-axis offset is +2, the y-axis offset is +1, and the two-dimensional coordinate offset vector is (2, 1). Subsequently, the offset vectors of all frames are subjected to temporal difference processing according to the frame sequence number to extract the inter-frame direction change trend within the entire window. For example, if (2, 1) is compared with the previous frame (1, 1), the calculated direction angle change is approximately 26.56°. This change value is merged into the direction sequence, and the operation is repeated to complete the construction of the full-frame pixel direction vector sequence. The final generated sequence value needs to store the pixel number, timestamp, x-axis and y-axis offset, and the current direction angle value to form a data set containing the pixel motion trend of the entire image, and obtain the pixel vector direction sequence value.

[0026] The disturbance frequency screening submodule obtains the directional change amplitude and the number of change cycles of each group of pixel vectors in consecutive frames based on the pixel vector direction sequence value, determines whether the number of change cycles is between the lower limit threshold of the image disturbance frequency and the upper limit threshold of the image disturbance frequency, and filters and marks the pixel labels that meet the judgment conditions to generate a frequency interference pixel label set; Based on the pixel vector direction sequence value, firstly, the direction angle values ​​of each group of pixels in the continuous frames are counted, and the direction change amplitude is extracted, that is, the difference between the direction angles of the previous and next two frames is used as the amplitude benchmark. For example, if the direction angle change sequence of a certain pixel point from the 1st frame to the 10th frame is [15°, 30°, 20°, 35°, 10°], then the direction change amplitude sequence is [15°, -10°, 15°, -25°]. The average amplitude value is obtained by taking the average absolute value or standard deviation of the sequence. The lower limit threshold of the image disturbance frequency is set to 2 times / second and the upper limit threshold is 6 times / second. Then, it is necessary to judge the number of complete cycle fluctuations of the pixel in 10 frames. Specifically, the number of positive and negative intersections is divided by the total number of positive and negative intersections. Frame duration: if the fluctuation frequency is 3 times / 0.33 seconds, it is converted to 9.09 times / second, which exceeds the upper threshold and should be excluded; if it is 2 times / 0.33 seconds, that is, 6.06 times / second, it is just at the upper limit of the interval; if it is 1 time / 0.33 seconds, that is, 3.03 times / second, it is in the valid interval and is retained; the judgment is based on the logical comparison operation, that is, if the frequency f satisfies 2≤f≤6, the pixel is labeled as a valid point in the current frame, otherwise it is marked as an invalid point, and finally a frequency interference pixel label set is obtained. Each label records the pixel position index, judgment time, direction amplitude index, number of fluctuation cycles, and screening status value (0 or 1), thereby generating a frequency interference pixel label set.

[0027] The regional pixel exclusion submodule excludes the pixel position corresponding to the label index in the image frame according to the frequency interference pixel label set, retains the available area of ​​the current frame image, and establishes the disturbance frequency exclusion area data; According to the frequency interference pixel label set, all pixels with invalid labels in the image frame are mapped to their positions. The pixel coordinate index is used as the judgment basis to remove the pixel data of the corresponding coordinates from the frame image, that is, the grayscale value at the corresponding position in the image matrix is ​​set to 0 or set to an invisible value. The remaining area is then extracted and divided into a new valid pixel area. The spatial continuity of the remaining pixels is judged and their spatial coverage is calculated. If the total number of initial pixels of a single frame image is 2073600 (1920×1080) and the number of removed pixels is 350000, the remaining pixel ratio is 82.5%. These pixels are combined into a valid area set of continuous frame images, while retaining the time label and spatial position index. The boundary rectangle and index table structure of the valid image block are marked. Finally, the new area set composed of the remaining pixels is used as the valid processing range in the input image to establish the disturbance frequency exclusion area data.

[0028] See also Figure 2 and Figure 4 ,The target contour separation module includes the direction change extraction ,submodule, the edge pixel marking submodule, and the closed area screening ,submodule; The direction change extraction submodule excludes regional data based on the disturbance frequency, obtains the grayscale value change and coordinate position offset of the remaining pixel area in consecutive frames, calculates the gradient direction angle difference and the mean square error of the optical flow amplitude of the regional pixels, combines the gradient direction angle difference and the mean square error of the optical flow amplitude, and classifies them to establish the direction amplitude distribution interval value; Based on the disturbance frequency exclusion area data, the remaining pixels in the continuous frames are first extracted, and the grayscale value of each pixel between the t frame and the t+1 frame is recorded. The change in grayscale value is obtained by the difference in pixel values ​​between the two frames. For example, the grayscale value of pixel q is 140 in the 5th frame and 132 in the 6th frame, then the grayscale change is -8, and the change is stored in the inter-frame change sequence. At the same time, the coordinate position change of the pixel in the same time window is extracted, that is, the x and y coordinate difference of the pixel is located in the continuous frames. If the coordinate of point q changes from (400, 220) to (403, 224), the offset vector is (3, 4). Based on the grayscale change sequence and the coordinate offset, the directional feature of the pixel is extracted. Specifically, the gradient vector is constructed to calculate its direction angle and the angle difference between each frame is taken as the gradient direction angle difference. For example, if point q is in frame t The direction is 60° and the t+1 frame is 80°, then the difference is 20°. Then, the mean square error of the optical flow amplitude is calculated. The optical flow velocity of the pixel in all frames (that is, the displacement amplitude between adjacent frames) needs to be extracted and the average of its mean and the square value of the deviation of each frame is calculated. For example, the velocity values ​​in 10 frames are [2, 3, 2, 2, 3, 2, 1, 3, 2, 2], and the average value is 2.2. The mean square error is about 0.49. The direction angle difference and the mean square error constitute a two-dimensional vector feature group. All pixels are classified accordingly and classified into an interval table in the form of a two-dimensional matrix. The direction difference interval is set at 5° and the optical flow amplitude variance interval is set at 0.1. A two-dimensional mapping table is formed by combining them. For example, the angle difference of 20° corresponds to the 5th gear, and the optical flow difference of 0.49 corresponds to the 5th gear. It is finally classified into the (5, 5) position. After classification, the direction amplitude distribution interval value is established.

[0029] The edge pixel marking submodule calls the direction amplitude distribution interval value to determine the pixel points where the gradient direction angle difference is greater than the direction difference angle setting value and the optical flow amplitude mean square error is less than the amplitude fluctuation reference value. The position of the pixels that meet the dual conditions is recorded, and the spatial clustering distribution of the pixels in the image frame is counted to obtain the edge feature aggregation value. Call the direction amplitude distribution interval value, traverse the gradient direction angle difference and optical flow amplitude mean square error of each pixel one by one, and judge whether the set conditions are met, that is, whether the angle difference is greater than the direction difference angle setting value, and whether the optical flow amplitude mean square error is less than the amplitude fluctuation reference value. The direction difference angle setting value is controlled to 15° according to the detection stability, and the amplitude fluctuation reference value is set to 0.6. The former indicates that the pixel direction change must have sufficient difference, and the latter indicates that the pixel movement must remain relatively stable. For example, a pixel direction difference of 18° and an optical flow variance of 0.52 meet the judgment conditions, so It is marked as an edge candidate point. If the direction difference is 10° and the optical flow variance is 0.65, it will be eliminated due to the insufficiency of the previous item. The positions of all pixels that meet the conditions are recorded, and the density of the area in which they are located in the image is counted. For example, the number of edge candidate pixels in a 20×20 pixel unit block is calculated. If the number of candidate pixels in a block is 180 and the total number of pixels is 400, the aggregation degree is 180 / 400=0.45. If the aggregation value is greater than 0.3, it is recorded as an edge aggregation area. Finally, the number of aggregated pixels in each block and the pixel index to which they belong are extracted to obtain the edge feature aggregation value.

[0030] The closed area screening submodule selects pixel clusters with closed boundary characteristics in continuous areas based on the edge feature aggregation value, calculates the connectivity index and edge closure of the boundary contour, and integrates the areas that meet the closed feature judgment criteria into boundary structure groups to generate a stable contour structure set of floating objects; According to the edge feature aggregation value, a connectivity identification operation is performed on all edge candidate blocks. Specifically, each edge candidate pixel is divided into connected branch regions according to the adjacent position relationship. The boundary contour of each connected region is calculated and it is determined whether it constitutes a closed loop, that is, the proximity between the starting point and the end point on the two-dimensional plane is less than the set closing error threshold of 0.05×the boundary length. At the same time, the boundary connectivity index needs to be determined, that is, whether the pixels in the region are completely continuous in the 4-neighborhood or 8-neighborhood. When the connectivity index is 1.0, it indicates complete continuity and is included in the closed candidate region. For example, if the pixel cluster area is 100 pixels, the closing error is less than 5 pixels, and the connectivity index is 0.98, it is classified as a closed region. Multiple pixel clusters that meet the closing conditions are integrated, and their contour point coordinate sequence, boundary size information and image frame number are uniformly recorded. A structured closed region information table is established for subsequent target recognition, and a stable contour structure set of floating objects is generated.

[0031] See also Figure 2 and Figure 5 ,The pollution level determination module includes a reflectivity identification submodule, a density value calculation submodule, and a level classification submodule; The reflectivity recognition submodule calls the stable contour structure set of floating objects. Based on the pixel distribution range of each contour target area, it extracts the maximum grayscale value and the average grayscale value per unit area within the area. It determines whether the maximum grayscale value per unit area is within the valid interval of the light reflectivity grade segment, records the segment position as the grade location index, and establishes the reflectivity grade positioning value. Call the stable contour structure set of floating objects. According to the position boundary of the pixel distribution in each contour area, first obtain the two-dimensional coordinate range covered by the area and limit it to a closed polygon area. Extract the grayscale values ​​of all pixels in the area. The grayscale value is sampled at 8-bit depth, that is, the range is 0 to 255. Count the maximum grayscale value R in the area pixels and the arithmetic mean of the grayscale of all pixels. For example, if the pixel grayscale value in the area is [112, 115, 118, 113, 111], the maximum value is 118 and the average value is 113.8. The maximum grayscale value per unit area is used to determine whether it is in the valid range. The pre-set light reflectance level segment is used for comparison. The segment is divided as follows: 0–80 is level 1, 81–160 is level 2, 161–200 is level 3, and 201–255 is level 4. At this time, the maximum grayscale value 118 is in segment 2, so its corresponding segment number is recorded as 2, which is used as the level location index of the floating object target. To improve versatility, the grayscale value is normalized to the [0, 1] interval and numbered. With 255 as the denominator, the normalized grayscale maximum value is 0.462, which is linearly mapped to level 2 according to the above segment ratio. Finally, all contour targets are assigned their own reflectance level number, which is stored in the structure table in a one-to-one correspondence with the contour ID to establish the reflectance level positioning value.

[0032] The density value calculation submodule obtains the apparent area and mass estimation value of the corresponding target based on the reflectivity level positioning value, calculates the average brightness value per unit area as the brightness index, and obtains the pollution level mapping value through calculation; The specific formula for calculating the pollution level mapping value is: ; in, Represents the pollution level mapping value, represents the normalized value of the estimated quality, represents the normalized value of the apparent area, Indicates the brightness value per unit area, represents the average brightness value of grayscale, Indicates the maximum grayscale value; Pollution level mapping value It is a composite indicator used to quantify the degree of floating debris pollution. This value comprehensively considers the following three types of physical quantities: Estimated mass of floating objects ( ): reflects the possible load of pollutants; Apparent area ( ): Indicates the coverage of pollutants on the water surface; The pollution level mapping value is essentially a double-weighted combination of the pollutant's scale (area + mass) and reflectivity (optical characterization), which is used to support pollution identification and classification.

[0033] Item 1 : Measures the “mass density” of pollution per unit area, giving higher weight to “heavy and small” pollutants; Item 2 : Measures the ratio of brightness fluctuation (non-uniformity) within pollutants to peak brightness, enhancing the ability to identify highly reflective abnormal areas (such as plastic bottles); Denominator :Suppress objects with too large an area from dominating the pollution level calculation, avoid excessively high levels due to huge volume but low actual optical pollution, and play a harmonizing and normalizing role.

[0034] The pollution level mapping value provides a quantifiable decision value for the pollution level determination module; a unified standard is formed to classify different types of floating objects, such as "high-density and highly reflective type", "low-quality and weakly reflective type", etc.

[0035] According to the reflectivity level positioning value, the apparent area A and estimated mass m of the target corresponding to each contour are extracted. The apparent area is obtained by converting the number of pixels surrounded by the contour and the image resolution. For example, a contour contains 450 pixels and the area of ​​a single pixel in the image is 0.0025m2, then A=450×0.0025=1.125m2. The mass m sets the density range according to the type of floating objects and converts the area combined with the density. The density of this type of material is set to be approximately 80kg / m2, then m=A×80=90kg. The average brightness value per unit area B is the sum of the grayscale values ​​of the pixels in the area divided by the total number of pixels. For example, if the grayscale value is [112, 115, 118, 113, 111], then B=(112+115+118+113+111) / 5=113.8. The average grayscale value is known. The maximum grayscale value R is 113.8, and the maximum grayscale value R is 118. Substitute the above values ​​into the pollution level mapping value calculation formula: ; Substitution is worth: , The pollution level mapping value is 79.993, which is used as a floating reference index for the target pollution level and recorded in the structure table to generate the pollution level mapping value.

[0036] The level classification submodule divides the value range into pollution level distribution segments based on the pollution level mapping value, identifies the level number according to the pollution intensity type corresponding to each segment, counts the level number and the number of contour targets, and establishes pollution level information; Based on the pollution level mapping value, all pollution indices L are grouped according to the set level distribution segment. The preset segmentation rules are 0-50 for low pollution, 51-100 for medium pollution, 101-150 for heavy pollution, and above 151 for extremely heavy pollution. If the L value in the previous example is 79.993, it belongs to the medium pollution level, corresponding to the number 2. All pollution level mapping values ​​are numbered accordingly, and the number of contours corresponding to each number is counted to generate a pollution level number and quantity comparison table. For example, number 1 has 15 targets, number 2 has 28, number 3 has 9, and number 4 has 3. Each pollution level number is associated with the contour ID and mapped, and the output is a pollution level structured data set to establish pollution level information.

[0037] See also Figure 2 and Figure 6 ,The spatial interception evaluation module includes a contour area determination submodule, a target density extraction submodule, and a weighted density correction submodule; The contour area measurement submodule obtains a stable contour structure set of floating objects, extracts the pixel boundary of each contour based on the image position of the target within the projection area of ​​the interception device, calculates the corresponding image projection area, and maps the actual projection area in the image with the image area within the interception area. It counts the number of contours and records the blocks they are located in to generate regional target statistics. Obtain the image position of each contour area in the stable contour structure set of floating objects, extract the pixel boundary of each contour from the image coordinate system, and the boundary is composed of continuous pixel contour points. Use the extreme boundary method to record the coordinate points of the upper left corner and lower right corner of the minimum circumscribed rectangle. Assume that the horizontal and vertical coordinate range of a certain contour point is x=120 to x=160, and y=300 to y=340. Then the width and height of the contour enclosing rectangle are 40 and 40, and the number of pixels contained in the boundary is 1600. The image resolution is 0.001m2 / pixel, and the converted area is 1.6m2. Then perform the above area calculation for all contours in the image. Operation, obtain the list of contour image projection areas, and read the projection area boundary settings corresponding to the interception device in the image. For example, the interception device mapping boundary is x=0 to x=1000, y=0 to y=600, divided into 5 horizontal segments and 3 vertical segments, and the size of a single block is 200×200 pixels. Block mapping is performed on the boundary center point position of each contour, and it is determined in which row and column of the projection block the contour is located. The contour ID and its mapping block number are recorded, and the number of contours mapped in each block is counted. Finally, the number of contours and the corresponding position coordinates of each mapping block are output to establish the regional target statistics.

[0038] The target density extraction submodule calculates the number of target contours per unit image area based on the regional target statistics, extracts the image contour circumscribed rectangle corresponding to the contour, calculates the diagonal length, uses the length value as the image scale indicator of the target, and generates the scale distribution feature value; According to the regional target statistics, read the contour target values ​​contained in each image partition and calculate the area of ​​each partition. For example, the area of ​​the aforementioned 200×200 pixel area is 40,000 pixels, which is converted to 40m2. If the number of statistical contours in this block is 8, the contour density per unit area is 8 / 40=0.2 / m2. Then traverse each contour target and obtain the leftmost, rightmost, topmost and bottommost coordinate points of the boundary from its contour point set. Construct the circumscribed rectangle and calculate the Euclidean distance between the diagonal points of the rectangle as the scale indicator. The formula is , for example, if the x coordinate is from 120 to 160 and the y coordinate is from 300 to 340, then the diagonal length is Pixel, if the actual length corresponding to the image pixel is 0.01m, then the scale value is 0.5657m, which represents the typical geometric dimension of the contour. Repeat the process for all contours and finally form a record table of the one-to-one correspondence between the diagonal scale value of each contour and the partition number to which it belongs. The output is the scale distribution feature value.

[0039] The weighted density correction submodule calls the scale distribution eigenvalue, calculates the ratio of the number of targets in each unit area to the total number of targets, multiplies it with the diagonal length as the weighting factor, calculates the interception adjustment density value per unit area, and generates area interception density information; Call the scale distribution eigenvalue and calculate the proportion of contours in all unit areas by dividing the number of contours in the entire image. For example, if the number of regional contours is 8 and the total number of contours is 50, the proportion is 0.16. Read the scale index of all contours in the region and set it to [l1, l2, …, l8] = [0.45, 0.48, 0.52, 0.43, 0.50, 0.47, 0.49, 0.46]. The mean is 0.475, and the weighted index of the region is calculated as 0.16 × 0.475 = 0.076. Process all partitions in this way and obtain the interception adjustment density value of each unit area. Finally, output the weighted density result value corresponding to each image region number to establish the area interception density information.

[0040] See also Figure 2 and Figure 7 ,The efficiency index building module includes a residence difference determination submodule, a density value superposition submodule, and an efficiency value generation submodule; The residence difference measurement submodule extracts the frame number of the same target in each frame based on the area interception density information and pollution level information, calculates the difference between the minimum and maximum numbers as the residence frame length, obtains the minimum value of the corresponding frame number of the same target being intercepted, and calculates the target residence frame length and interception frame length to generate the residence difference index value; According to the area interception density information and pollution level information, first extract the timestamps of all identified floating objects in each frame from the image sequence, and record the first and last numbers of the target in the consecutive frames as the start and end positions of the appearance. For example, if a target appears continuously in the frame number sequence [21, 22, 23, 24, 25], then its residence frame length is 25–21=4 frames minus the start frame from the last frame. If the frame rate is set to 30fps, the corresponding residence time is 4 / 30=0.133 seconds. Then obtain the interception frame number, that is, the image frame number where the target is intercepted. If the interception occurs in the 23rd frame, the interception frame length is 23–21. =2 frames, i.e. 0.067 seconds; all targets are processed separately in this way to form corresponding residence time and interception time data sets, and then the residence frame lengths and interception frame lengths of all targets are statistically averaged. For example, the residence frame lengths of 5 targets are [4, 6, 5, 7, 4] frames, and the interception frame lengths are [3, 5, 4, 5, 3] frames, then the average residence frame length is 5.2 frames, and the average interception frame length is 4 frames. Finally, the difference operation 5.2–4=1.2 frames is performed, and the residence frame length, interception frame length and their difference of each target are recorded. Combined with the density data of each unit area in the area interception density information, for example, the density of a certain area is 0.34 / m 2 The level number in the pollution level information is set to an interval of 1–4, and the pollution level number is normalized to a decimal value between 0.25–1. For example, level 2 is numbered 0.5. An association table between the three variables is established to provide a numerical basis for the subsequent calculation of the efficiency expression value, and finally generate the residence difference index value.

[0041] The density value superposition submodule calls the resident difference index value, the level number corresponding to each pollution level, takes the normalized value of the level number value as the pollution factor weight, and combines the three parameters to calculate the efficiency expression value of floating objects per unit area, and calculates the efficiency index value per unit area; The specific formula for calculating the unit area efficiency index value is: ; in, Indicates the unit area efficiency index value, represents the normalized value of the area interception density information, Indicates the normalized value of the pollution level number, represents the average length of the resident frame, represents the average intercept frame length; Unit area efficiency index value It is a comprehensive indicator to evaluate the filtration capacity of the filtration system in a fixed area; It is composed by integrating the following three types of parameters: Area interception density ( ): represents the density of the number of targets detected and captured by the system per unit area; Pollution level ( ): Normalized value based on pollution level number, reflecting the pollution intensity in the area; The difference between the length of the resident and intercepted frames ( ): Measures the efficiency difference between the time pollutants reside in the filter area and the time they are captured.

[0042] This indicator reflects the interception coverage effect and response speed of the filtration system to pollutants of different levels in unit time and unit space.

[0043] Molecules Part I : Taking interception density as weight, the composite response capacity between pollution level and residence efficiency is calculated; Intermediate term : Indicates the degree of detection response lag. The smaller the value, the more timely the interception. Denominator : Inversely suppress areas that are biased by high density but low pollution values ​​to ensure balanced results.

[0044] The efficiency index value per unit area provides a spatial performance reference for system deployment or improvement; it evaluates the pollutant treatment efficiency of different filtration modules and assists in establishing regional filtration priorities.

[0045] Call the residence difference index value to normalize the pollution level number. If the original number range is 1-4, the normalization method is number / 4. For example, when the level number is 3, the corresponding normalized value is 0.75. The area interception density information is also normalized. If the maximum density value is 0.6 / m 2 , the current regional density is 0.3, then the normalized density is 0.3 / 0.6=0.5, and then the three normalized values ​​are input into the efficiency index calculation formula: ; Substitute the example values: , , , ,but: ; The calculation result is 0.0532, which represents the efficiency expression of the unit area of ​​the region taking into account the pollution level, target residence time and interception time. This value is stored and included in the corresponding block record to obtain the unit area efficiency index value.

[0046] The efficiency value generation submodule performs mapping statistics according to the unit area efficiency index value in the image space according to the projection block of the filter device, extracts the mean of the corresponding index value in the block and outputs it as the basis for regional filtering efficiency evaluation, and establishes unit filtering effect evaluation information; According to the unit area efficiency index value, all image areas are divided into rows and columns based on the interception device mapping block in the image space. For example, the image is divided into 20 areas in a 5×4 partitioning manner. The average value of all unit area efficiency index values ​​in each partition is calculated. If the index value in a certain block is [0.0532, 0.0456, 0.0671, 0.0510], the mean value is (0.0532+0.0456+0.0671+0.0510) / 4=0.0542. This mean value is used as the efficiency evaluation result of the current block. Repeat the processing of all areas in the image to form a unit area efficiency index mean distribution map. The block number, corresponding efficiency value and evaluation level label are output to establish unit filtering effect evaluation information.

[0047] It should be understood that the term "and / or" as used herein simply describes a relationship between associated objects, indicating that three possible relationships exist. For example, "A and / or B" can represent: A alone, A and B together, or B alone. A and B can be singular or plural. Furthermore, the character " / " as used herein generally indicates an "or" relationship between the associated objects, but it may also indicate an "and / or" relationship. For specific understanding, please refer to the context.

[0048] In this disclosure, "at least one" means one or more, and "plurality" means two or more. "At least one of the following" or similar expressions refers to any combination of these items, including any combination of single or plural items. For example, "at least one of a, b, or c" can mean: a, b, c, ab, ac, bc, or abc, where a, b, and c can be single or plural.

[0049] It should be understood that in various embodiments of the present invention, the size of the serial numbers of the above-mentioned processes does not mean the order of execution. The execution order of each process should be determined by its function and internal logic, and should not constitute any limitation on the implementation process of the embodiments of the present invention.

[0050] Those skilled in the art will appreciate that the units and algorithm steps of each example described in conjunction with the embodiments disclosed herein can be implemented in electronic hardware, or a combination of computer software and electronic hardware. Whether these functions are performed in hardware or software depends on the specific application and design constraints of the technical solution. Professionals and technicians can use different methods to implement the described functions for each specific application, but such implementation should not be considered beyond the scope of the present invention.

[0051] Those skilled in the art will clearly understand that, for the convenience and brevity of description, the specific working processes of the above-described equipment, devices and units can refer to the corresponding processes in the aforementioned method embodiments and will not be repeated here.

[0052] In the several embodiments provided by the present invention, it should be understood that the disclosed devices, apparatuses and methods can be implemented in other ways. For example, the device embodiments described above are merely illustrative. For example, the division of the units is merely a logical function division. In actual implementation, there may be other division methods, such as multiple units or components can be combined or integrated into another device, or some features can be ignored or not executed. Another point is that the mutual coupling or direct coupling or communication connection shown or discussed can be through some interface, indirect coupling or communication connection of the device or unit, which can be electrical, mechanical or other forms.

[0053] The units described as separate components may or may not be physically separate, and the components shown as units may or may not be physical units, that is, they may be located in one place or distributed across multiple network units. Some or all of these units may be selected to achieve the purpose of this embodiment according to actual needs.

[0054] In addition, each functional unit in each embodiment of the present invention may be integrated into one processing unit, or each unit may exist physically separately, or two or more units may be integrated into one unit.

[0055] If the functions are implemented as software functional units and sold or used as independent products, they can be stored in a computer-readable storage medium. Based on this understanding, the technical solution of the present invention, or the portion that contributes to the prior art, or the portion of the technical solution, can be embodied in the form of a software product. This computer software product is stored in a storage medium and includes several instructions for enabling a computer device (which can be a personal computer, server, or network device, etc.) to execute all or part of the steps of the methods described in various embodiments of the present invention. The aforementioned storage media include various media that can store program code, such as USB flash drives, mobile hard drives, read-only memories (ROM), random access memories (RAM), magnetic disks, or optical disks.

[0056] The above description is merely a specific embodiment of the present invention, but the scope of protection of the present invention is not limited thereto. Any modifications or substitutions that can be easily conceived by a person skilled in the art within the technical scope disclosed in the present invention should be included in the scope of protection of the present invention. Therefore, the scope of protection of the present invention should be based on the scope of protection of the claims.

Claims

1. Water surface floating object filtering data analysis system, characterized by: The system includes: The disturbance frequency extraction module obtains pixel points within a continuous time window in the water surface image sequence frame, performs attribution analysis on the direction change rate and amplitude value of each group of pixel vectors, excludes pixel groups within the image disturbance frequency threshold range, and generates disturbance frequency exclusion area data; The target contour separation module excludes the regional data based on the disturbance frequency, marks pixels whose gradient direction change rate is greater than the direction difference angle setting value and whose optical flow amplitude standard deviation is less than the amplitude fluctuation reference value, retains the regions with closed edge attributes, and generates a stable contour structure set of floating objects; The pollution level determination module calls the stable contour structure set of floating objects, obtains the average brightness of pixels in the corresponding area, and combines the apparent area and estimated mass value of the floating objects to calculate the pollution level of the floating objects and generate pollution level information; The spatial interception assessment module obtains the stable contour structure set of the floating objects, calculates the number of targets per unit area, extracts the circumscribed rectangle of the target's image contour, uses the diagonal length value as the scale feature, and performs correction using the scale value as the weighting factor to generate area interception density information.

2. The water surface floating object filtering data analysis system according to claim 1, characterized in that: The disturbance frequency exclusion area data includes a non-periodic disturbance vector cluster, a frequency domain exclusion pixel index group and a frequency boundary outside judgment label; the floating object stable contour structure set is specifically a direction alienated edge set, a closed contour identification mapping and a continuous boundary graphic index; the pollution level information includes a light reflection intensity level index and an apparent density ratio group; the area interception density information is specifically a unit area target density value, a scale weighting coefficient and a standardized interception ratio.

3. The water surface floating object filtering data analysis system according to claim 2, characterized in that: The disturbance frequency extraction module includes: The vector direction extraction submodule obtains pixel points within a continuous time window in the water surface image sequence frame, detects the grayscale value change of the corresponding position of the pixel point in the continuous frame, calculates the two-dimensional coordinate offset value of the pixel between adjacent frames, and performs differential processing on the offset value in the time dimension to extract the motion direction vector, calls the pixel motion direction vector result, and generates a pixel vector direction sequence value; The disturbance frequency screening submodule obtains the direction change amplitude and the number of change cycles of each group of pixel vectors in consecutive frames based on the pixel vector direction sequence value, determines whether the number of change cycles is between the lower limit threshold of the image disturbance frequency and the upper limit threshold of the image disturbance frequency, screens and marks the pixel labels that meet the judgment conditions, and generates a frequency interference pixel label set; The regional pixel exclusion submodule excludes the pixel positions corresponding to the label index in the image frame according to the frequency interference pixel label set, retains the available area of ​​the current frame image, and establishes disturbance frequency exclusion area data.

4. The water surface floating object filtering data analysis system according to claim 3, characterized in that: The target contour separation module includes: The direction change extraction submodule excludes the regional data based on the disturbance frequency, obtains the grayscale value change and coordinate position offset of the remaining pixel area in continuous frames, calculates the gradient direction angle difference and the mean square error of the optical flow amplitude of the regional pixels, combines the gradient direction angle difference and the mean square error of the optical flow amplitude, and classifies them to establish the direction amplitude distribution interval value; The edge pixel marking submodule calls the direction amplitude distribution interval value, determines the pixel points whose gradient direction angle difference value is greater than the direction difference angle setting value and whose optical flow amplitude mean square error value is less than the amplitude fluctuation reference value, and records the position of the pixels that meet the dual conditions, and counts the spatial clustering distribution of the pixels in the image frame to obtain the edge feature aggregation value; The closed area screening submodule screens pixel clusters with closed boundary characteristics in the continuous area according to the edge feature aggregation value, calculates the connectivity index and edge closure degree of the boundary contour, and integrates the areas that meet the closed feature judgment criteria into a boundary structure group to generate a stable contour structure set of floating objects.

5. The water surface floating object filtering data analysis system according to claim 4, characterized in that: The pollution level determination module includes: The reflectivity recognition submodule calls the stable contour structure set of floating objects, extracts the grayscale maximum value and grayscale average value per unit area of ​​each contour target area based on the pixel distribution range, determines whether the grayscale maximum per unit area is within a valid interval within the light reflectivity grade segment, and records the segment position as a grade location index to establish a reflectivity grade location value; The density value calculation submodule obtains the apparent area and mass estimation value of the corresponding target according to the reflectivity level positioning value, calculates the average brightness value per unit area as the brightness index, and obtains the pollution level mapping value by calculation; The level classification submodule divides the value range into pollution level distribution segments based on the pollution level mapping value, identifies the level number according to the pollution intensity type corresponding to each segment, counts the level number and the number of contour targets, and establishes pollution level information.

6. The water surface floating object filtering data analysis system according to claim 5, characterized in that: The formula for calculating the pollution level mapping value is as follows: ； in, Represents the pollution level mapping value, represents the normalized value of the estimated quality, represents the normalized value of the apparent area, Indicates the brightness value per unit area, represents the average brightness value of grayscale, Indicates the maximum grayscale value.

7. The water surface floating object filtering data analysis system according to claim 6, characterized in that: The space interception assessment module includes: The contour area determination submodule obtains the stable contour structure set of the floating object, extracts the pixel boundary of each contour based on the image position of the target within the projection area of ​​the interception device, calculates the corresponding image projection area, partitions the actual projection area in the image and the image area within the interception area, counts the number of contours and records the blocks they are located in, and generates regional target statistics; The target density extraction submodule calculates the number of target contours within the unit image area based on the regional target statistics, extracts the image contour circumscribed rectangle corresponding to the contour, calculates the diagonal length, uses the length value as the image scale indicator of the target, and generates a scale distribution feature value; The weighted density correction submodule calls the scale distribution characteristic value, calculates the ratio of the number of targets in each unit area to the total number of targets, multiplies the product with the diagonal length as a weighting factor, calculates the interception adjustment density value per unit area, and generates area interception density information.

8. The water surface floating object filtering data analysis system according to claim 7, characterized in that: The system further comprises: The efficiency index construction module calculates the difference between the average residence frame length and the average interception frame length of the target based on the area interception density information and the pollution level information, combined with the frame sequence residence length of each target in the filtering area, and calculates the unit area efficiency index value based on the unit area interception density to generate unit filtering effect evaluation information; The unit filtration effect evaluation information specifically refers to the average treatment response difference, the area interception correction factor and the time-effect harmonic parameter.

9. The water surface floating object filtering data analysis system according to claim 8, characterized in that: The efficiency index building block includes: The resident difference determination submodule extracts the frame number of the same target in each frame based on the area interception density information and the pollution level information, calculates the difference between the minimum and maximum numbers as the resident frame length, obtains the minimum value of the corresponding frame number of the same target being intercepted, calculates the target resident frame length and the interception frame length, and generates a resident difference index value; The density value superposition submodule calls the residence difference index value, the level number corresponding to each pollution level, takes the normalized value of the level number value as the pollution factor weight, and jointly calculates the efficiency expression value of floating objects per unit area by combining the three parameters to obtain the unit area efficiency index value; The efficiency value generation submodule performs mapping statistics according to the unit area efficiency index value in the image space area according to the projection block of the filtering device, extracts the mean of the corresponding index value in the block and outputs it as the basis for regional filtering efficiency evaluation, and establishes unit filtering effect evaluation information.

10. The water surface floating object filtering data analysis system according to claim 9, characterized in that: The formula for calculating the unit area efficiency index value is as follows: ； in, Represents the unit area efficiency index value, represents the normalized value of the area interception density information, Indicates the normalized value of the pollution level number, represents the average length of the resident frame, Indicates the average intercept frame length.

Images