Image processing-based parameter measurement method and system for DBD-type ozone generator

Abstract

The application provides a DBD (Dielectric Barrier Discharge) type ozone generator parameter measurement method and system based on image processing. Specifically, the method comprises the following steps: image preprocessing, obtaining grouped key corner points, preliminary fitting of quadrilaterals, secondary fitting of quadrilaterals, and DBD related parameter calculation and output. The method automatically fits quadrilaterals through a program, replaces manual fitting, and realizes high-precision measurement of parameters such as the equivalent capacitance of a DBD type ozone generator, the discharge maintenance voltage, and the discharge energy per cycle. The method has strong applicability and high measurement precision, and solves the problems of long time consumption and low precision of the existing manual fitting parameter calculation method.

Description

Technical Field

[0001] This invention belongs to the field of electrical parameter measurement, specifically a method and system for measuring the parameters of DBD-type ozone generators based on image processing. This invention uses a computer program to fit Lissajous figures, replacing manual fitting methods, to improve the efficiency and accuracy of DBD-type ozone generator parameter measurement. Background Art

[0002] Ozone, a pale blue gas, possesses strong bactericidal and oxidizing properties while producing no pollution or residue during use. Therefore, it is widely used in a variety of fields, including environmental protection, medicine, and food. In industrial production, ozone is primarily generated through dielectric barrier discharge, which has become the primary method for industrialized ozone production both domestically and internationally.

[0003] Existing research on DBD ozone generators has primarily focused on power supply, load characteristics, and ozone production efficiency. However, limited research has focused on the measurement and automated determination of equivalent circuit parameters for DBD ozone generators. However, accurately determining the equivalent parameters of DBD ozone generators is crucial for optimizing discharge systems and improving ozone production efficiency.

[0004] Traditionally, methods for obtaining DBD ozone generator parameters include calculation method and Lissajous figure measurement method. The calculation method derives the equivalent capacitance based on geometric structure and physical formulas, but this method is easily affected by material properties and processing errors, its accuracy is difficult to guarantee, and it is unable to measure the discharge maintenance voltage. In contrast, the Lissajous figure measurement method can measure three important parameters at the same time with high accuracy, and has become the main measurement method. However, most existing Lissajous figure measurement methods rely on manual fitting of parallelograms. This method has problems such as low efficiency and poor accuracy, and it is difficult to meet the needs of automation and high-precision measurement.

[0005] In order to solve this problem, it is urgent to develop an automated Lissajous figure measurement method to improve the efficiency and accuracy of the equivalent parameter measurement of DBD type ozone generator. Summary of the Invention

[0006] The present invention proposes an automatic DBD ozone generator Lissajous image processing method. This method uses DBSCAN (Density-Based Spatial Clustering of Applications with Noise) clustering filtering to extract the Lissajous image data points of the DBD ozone generator, complete the preprocessing of the Lissajous image, and determine the four key corner points A, B, C, and D of the Lissajous image. Based on these four corner points, the data points are divided into four groups. Linear fitting is performed on each group of data points and the intersection points of adjacent edges are obtained to obtain a preliminary fitted quadrilateral. After further quadratic fitting, the final Lissajous image fitting quadrilateral is obtained. This method replaces manual fitting with a computer program, greatly improving the efficiency of parameter fitting while also ensuring the accuracy of fitting.

[0007] The DBD ozone generator parameter measurement method proposed in the present invention consists of the following main steps:

[0008] Preprocessing the Lissajous image of the DBD ozone generator, wherein the preprocessed image only contains a valid set of white data points;

[0009] Obtaining grouped key corner points A, B, C, and D based on the white data point set;

[0010] Divide the white data point set into 4 arrays, perform straight line fitting on the data point set of each array, and preliminarily construct a quadrilateral;

[0011] Based on the initially constructed quadrilateral, the boundary line including the new data point set is translated up and down, and a quadratic fitting is performed to construct a quadrilateral with higher accuracy;

[0012] The equivalent capacitance, discharge maintenance voltage and discharge energy per cycle of the DBD type ozone generator are calculated and output.

[0013] Furthermore, the step of “preprocessing the Lissajous image of the DBD ozone generator, wherein the preprocessed image only contains a valid set of white data points” includes the following sub-steps:

[0014] intercepting the Lissajous image generated by the oscilloscope to obtain a sub-image containing only a valid data area;

[0015] Convert the sub-image from RGB (Red, Green, Blue) to HSV (Hue, Saturation, Value) to extract color data points and generate a logical mask image;

[0016] Grayscale and binarization are performed on the logical mask image to obtain an image retaining only white data points; and a DBSCAN clustering algorithm is used to perform denoising on the data points in the image retaining the white data points to obtain a denoised binarized image.

[0017] Furthermore, the step of “obtaining grouped key corner points ABCD based on the white data point set” includes the following sub-steps:

[0018] Calculating the centroid of the white data points;

[0019] Based on the centroid, the data points are divided into left and right regions, and the average positions of the 10 points farthest from the centroid are selected in the left and right regions respectively to obtain key corner points A and B;

[0020] A straight line AB is constructed from the corner points A and B, and all the data points are divided into upper and lower areas through the straight line AB. The average positions of the 10 points farthest from the straight line AB are selected in the upper and lower areas respectively to obtain key corner points C and D.

[0021] Furthermore, the step of "dividing the white data point set into four arrays, performing straight line fitting on the data point set of each array, and preliminarily constructing a quadrilateral" includes the following sub-steps:

[0022] Calculate the angles of the lines connecting the key corner points A, B, C, and D and the center of mass relative to the horizontal axis of the center of mass;

[0023] Divide the white data point set into four arrays according to the angle; wherein Group 1 is a data point set within the angular range between the centroid and the key point A and the centroid and the key point C, Group 2 is a data point set within the angular range between the centroid and the key point C and the centroid and the key point B, Group 3 is a data point set within the angular range between the centroid and the key point B and the centroid and the key point D, and Group 4 is a data point set within the angular range between the centroid and the key point D and the centroid and the key point A;

[0024] Performing straight line fitting on each data point set using the least squares method to obtain four straight lines;

[0025] According to the four straight lines, the intersection points of adjacent straight lines are calculated respectively to obtain four vertices E, F, G, and H, and based on them, a preliminary fitting quadrilateral EFGH is constructed.

[0026] Furthermore, the step of “based on the initially constructed quadrilateral, translating the boundary line up and down to include a new set of data points, and performing quadratic fitting to construct a quadrilateral with higher accuracy” includes the following sub-steps:

[0027] The straight line corresponding to one side of the initially constructed quadrilateral is translated up and down to form two new translated straight lines;

[0028] Filtering out a set of data points located between the two new translation lines;

[0029] The screened data point set is again fitted using the least squares method to obtain an optimized expression for one side of the initially constructed quadrilateral;

[0030] Repeating the optimization expression obtaining process for the remaining three sides of the initially constructed quadrilateral to obtain the optimized expressions for the remaining three sides;

[0031] According to the four optimization expressions, the intersection points of adjacent edges are calculated, and finally the optimized quadrilateral A1B1C1D1 is constructed.

[0032] Furthermore, the step of “calculating and outputting the equivalent capacitance, discharge maintenance voltage, and discharge energy per cycle of the DBD type ozone generator” includes the following sub-steps:

[0033] Calculate the average slopes of the opposite sides of the quadrilateral A1B1C1D1, and deduce the equivalent capacitance, dielectric barrier capacitance, and air gap capacitance of the DBD type ozone generator based on the attenuation coefficient in the measurement circuit;

[0034] The discharge maintenance voltage of the DBD type ozone generator is calculated by taking half the absolute value of the difference between the horizontal straight line where the center of mass is located and the two intersection points of the quadrilateral A1B1C1D1 and combining it with the attenuation coefficient;

[0035] By calculating the area of ​​the quadrilateral A1B1C1D1, the discharge energy per cycle of the DBD type ozone generator is obtained.

[0036] The present invention also discloses a DBD type ozone generator parameter measurement system based on image processing, which includes the following modules:

[0037] Image preprocessing module: used for extracting valid data point areas from the Lissajous image generated by the oscilloscope and performing denoising processing;

[0038] A preliminary fitting module is used to obtain the key corner points A, B, C, and D according to the data point distribution, group the data points, and perform straight line fitting to construct the preliminary fitting quadrilateral.

[0039] Quadratic fitting and parameter output module: used to construct a quadratic fitting quadrilateral with higher accuracy based on the preliminary fitting quadrilateral, and calculate and output relevant parameters of the DBD type ozone generator.

[0040] Furthermore, the "image preprocessing module" includes the following submodules:

[0041] Graphics interaction submodule: used to save the captured area image;

[0042] Data point extraction submodule: used for color space conversion, extracting data points and binarizing images;

[0043] Noise point denoising submodule: used to denoise noise points using the DBSCAR clustering algorithm.

[0044] Furthermore, the "preliminary fitting module" includes the following submodules:

[0045] Centroid calculation and region segmentation submodule: used to calculate the centroid of the white data point and segment the data point region;

[0046] Key point A, B selection submodule: used to determine the point A and the point B;

[0047] Key point C, D selection submodule: used to determine the point C and the point D;

[0048] Angle calculation and grouping submodule: used to calculate the angles between the vectors from the centroid to the key corner points A, B, C, D and the horizontal axis, and divide the data points into four groups based on these angles;

[0049] Linear fitting submodule: used for performing least squares linear fitting on each set of data points;

[0050] Quadrilateral vertex calculation submodule: used to calculate the intersection of every two adjacent straight lines, obtain the vertices of the quadrilateral, and complete the preliminary fitting.

[0051] Furthermore, the "quadratic fitting and parameter output module" includes the following submodules:

[0052] Linear translation and data screening submodule: used to translate each initially fitted straight line up and down, and obtain a new set of data points by screening key data points that meet the translation range;

[0053] Quadratic linear fitting submodule: used to perform least squares linear fitting on the filtered data point set to calculate a new slope and intercept;

[0054] Optimized vertex calculation submodule: used for calculating the intersection points of adjacent straight lines based on the straight lines after quadratic fitting, obtaining the vertices of the quadratic fitting quadrilateral, and obtaining the quadratic fitting quadrilateral;

[0055] Equivalent capacitance calculation submodule: used to calculate the equivalent capacitance of the DBD type ozone generator, including dielectric barrier capacitance and air gap capacitance;

[0056] Discharge maintenance voltage calculation submodule: used to calculate the discharge maintenance voltage of the DBD type ozone generator;

[0057] Discharge energy calculation submodule: used to calculate the discharge energy per cycle.

[0058] Beneficial effects:

[0059] This invention provides a method and system for measuring parameters of a DBD (Dielectric Barrier Discharge) ozone generator based on image processing. Specifically, the method includes the following steps: image preprocessing, determining key corner points for grouping, preliminary quadrilateral fitting, secondary quadrilateral fitting, and calculation and output of DBD-related parameters. This method uses a program to automatically fit quadrilaterals, replacing manual fitting. This method achieves high-precision measurement of parameters such as the equivalent capacitance, discharge holding voltage, and discharge energy per cycle of a DBD ozone generator. This method offers strong applicability and high measurement accuracy, resolving the time-consuming and low-precision issues of existing manual parameter fitting methods. BRIEF DESCRIPTION OF THE DRAWINGS

[0060] Figure 1 This is a flow chart of the measurement method;

[0061] Figure 2 To manually capture the interactive interface diagram;

[0062] Figure 3 To complete the preprocessing data point map;

[0063] Figure 4 is a schematic diagram of the center of mass position;

[0064] Figure 5 Schematic diagram of the left and right area segmentation of the data point;

[0065] Figure 6 Schematic diagram of the upper and lower area segmentation of data points;

[0066] Figure 7 Schematic diagram of the positions of key corner points A, B, C, and D;

[0067] Figure 8 Schematic diagram for grouping data points;

[0068] Figure 9Schematic diagram of the preliminary fitted quadrilateral EFGH;

[0069] Figure 10 It is a schematic diagram of the data points included in the straight line translation up and down;

[0070] Figure 11 Schematic diagram of the final fitted quadrilateral A1B1C1D1. DETAILED DESCRIPTION

[0071] In the following description, many technical details are provided to help readers better understand this application. However, those skilled in the art will understand that even without these technical details and the various changes and modifications based on the following embodiments, the technical solutions claimed in the claims of this application can be implemented. The measurement method proposed in this invention can be Figure 1 As shown, the following will be discussed in detail.

[0072] Image preprocessing

[0073] This part corresponds to Figure 1 The S0 step in .

[0074] Capture image recognition area

[0075] The image cropping step aims to address the problem of the original image containing a large amount of irrelevant information. By allowing users to manually or automatically select specific regions of the image for fitting (i.e., regions of interest, ROIs), subsequent processing and analysis can be focused on the important parts. When performing operations such as centroid calculation and line fitting, selecting the appropriate cropping region can significantly improve computational efficiency and reduce data noise caused by irrelevant areas.

[0076] The following takes manual capture as an example to illustrate the specific operation process of image capture. The operation steps are as follows:

[0077] (1) Image loading: The program first loads the original image data provided by the user.

[0078] (2) Image display: The image is displayed to the user through a graphical interface tool (such as matplotlib) so that the user can view the entire image.

[0079] (3) User manual selection of region: The program allows the user to draw a rectangular box on the image with the mouse to define the region (ROI) they want to process.

[0080] (4) Determine and extract the selected area: After the user confirms the selected area, the program saves the coordinates of the area and extracts the corresponding part from the image. If the user does not select a valid area, the program will prompt an error and ask the user to re-capture the valid area. The valid area will be saved as a new image file and used in subsequent processing steps.

[0081] Since the images exported by oscilloscopes of different models and brands used in the experiment vary, manual capture can accurately capture the target area and eliminate irrelevant information, thereby ensuring the computational efficiency and accuracy of subsequent processing. Figure 2 A graphical interactive interface is presented in which users can freely select the region of interest.

[0082] Data point extraction

[0083] In this section, we'll use the orange Lissajous data point as an example to illustrate the data point extraction process. The goal is to extract the orange data point from an RGB image through a series of image processing steps and generate a binary image for subsequent data analysis. The specific steps are as follows:

[0084] Step 1: RGB to HSV color space conversion

[0085] Convert the RGB image to HSV color space to make it easier to extract the orange area. In the RGB color space, color is represented by three channels: red, green, and blue. In the HSV color space, hue represents the color type, saturation represents the color purity, and value represents the brightness. The conversion formula from RGB to HSV is:

[0086] HSV=f(RGB)# (1)

[0087] Function f converts the RGB color space to the HSV color space. The hue range of orange is set to 10 to 25 degrees, the saturation range is 100 to 255, and the brightness range is 100 to 255. By conditionally filtering the three HSV channels, the orange pixels are successfully extracted and a logical mask image (orange_mask) is generated.

[0088] Step 2: Generate grayscale image

[0089] After extracting the orange area, the resulting logical image has pixel values ​​of True (i.e., 1) in the orange area and False (i.e., 0) in all other areas. Multiplying these logical values ​​by 255 generates a grayscale image containing only 0 and 255, with white (255) representing the orange data points and black (0) representing the background.

[0090] Step 3: Binarization

[0091] Perform binarization on the grayscale image, set the threshold to retain only pixels with a grayscale value of 255 as the foreground, and treat all other pixels as the background. The specific expression is as follows:

[0092]

[0093] This step ensures that the orange data points are presented in a clear binary form.

[0094] Through the above steps, the orange data point area is successfully extracted and converted into a binary image, laying the foundation for subsequent denoising analysis.

[0095] Clustering algorithm DBSCAN denoising

[0096] In this part, the extracted white data points need to be filtered to further complete the image preprocessing operation. There are many filtering methods, the most common of which are median filtering and mean filtering. Both algorithms are based on local statistical operations of fixed windows. Among them, mean filtering achieves image smoothing by calculating the mean of pixels in the neighborhood, but it is easy to cause image blur; median filtering can better preserve edges, but its effect is relatively limited when dealing with complex shapes and non-uniform noise. In contrast, the DBSCAN (Density-Based Spatial Clustering of Applications with Noise) algorithm is a density-based clustering method that can automatically group data points according to their density. For complex structures such as Lissajous figures, DBSCAN can effectively distinguish between dense valid data points and discrete noise points, and automatically adapt to the local density distribution of the data, thereby removing noise while maintaining the structural information of the figure. Therefore, the present invention uses the DBSCAN clustering filtering algorithm.

[0097] DBSCAN, a density-based clustering algorithm, is highly efficient at handling noisy and irregularly shaped data points. Its core concept is to define clusters by density, without pre-specifying the number of clusters. The basic principles of the algorithm include the following: a core point is defined as one whose number of neighbors within a radius of ε (eps) is at least min_samples; a boundary point is a point that has a core point in its neighborhood but does not meet the core point criteria; and a noise point is a point that does not belong to any cluster. The algorithm starts from any data point, finds a core point that meets the criteria, and expands the cluster until no further expansion is possible. The remaining unclassified points are marked as noise points, and the noise point label is generally set to -1.

[0098] During image processing, the coordinates of all white data points in the binary image are first extracted and clustered using the DBSCAN algorithm. Experimental tuning revealed that the optimal settings for eps = 5 and min_samples = 6 produce the best results. Next, noise points labeled -1 are filtered out, while non-noise points are retained. A denoised binary image is generated and visualized to ensure that interfering points in the image are effectively removed.

[0099] After DBSCAN clustering and denoising, purer white data points are obtained, laying the foundation for subsequent centroid calculation and key point selection. At this point, the image preprocessing is completed, and the image exported by the oscilloscope also becomes a white data point diagram with a black background, as shown in the figure below. Figure 2 As shown, the white data points are the target points for the subsequent fitting analysis.

[0100] Determine group key corner points

[0101] This part corresponds to Figure 1 Step S1 in .

[0102] Centroid calculation and key point selection (points A and B)

[0103] Step 1: Calculate the center of mass. First, extract all white data points from the preprocessed image and use these white data points as the basis for the center of mass calculation. The center of mass is determined by the average of the coordinates of all data points and represents the geometric center of the entire data point set. The coordinates of the center of mass, C, are calculated using the following formula:

[0104]

[0105] Where: (x i ,y i ) represents the coordinates of the i-th white data point; N is the total number of data points. The calculated centroid position will serve as the basis for grouping and key point selection in subsequent steps. The centroid solution is visualized as follows: Figure 4 shown.

[0106] Step 2: Left and right region segmentation. Based on the centroid C, the white data points are divided into two regions, the left region contains all the points whose horizontal coordinates are smaller than the centroid C. x The right area contains points with a horizontal coordinate greater than or equal to C x The mathematical expression is as follows:

[0107] P left ={(x i ,y i )|x i <C x}# (4)

[0108] P right ={(x i ,y i )|x i ≥C x}# (5)

[0109] Among them, P left and P rightRepresents the data point set on the left and right respectively. This step of segmentation lays the foundation for the subsequent selection of key points A and B. The left and right area segmentation visualization is as follows Figure 5 shown.

[0110] Step 3: Select key points A and B. In the left and right regions, select the 10 data points farthest from the centroid and calculate the average position of these points to determine key points A and B. First, use the Euclidean distance formula to calculate the distance from each data point to the centroid:

[0111]

[0112] Among them, d i Indicates the distance from the i-th point to the center of mass, (x i ,y i ) is the coordinate of the data point, C x and C y The 10 data points farthest from the center of mass in the left and right regions are selected respectively, and the average coordinates of these points are calculated to obtain the positions of key points A and B:

[0113]

[0114]

[0115] Among them: A represents the average position of the 10 points on the left that are farthest from the center of mass; B represents the average position of the 10 points on the right that are farthest from the center of mass. This step determines the key points A and B, providing a basis for subsequent linear fitting and quadrilateral construction.

[0116] Region segmentation and key point selection (points C and D)

[0117] 1. Calculate the equation of line AB:

[0118] Determine the equation of line AB through points A and B. First, calculate the standard form equation of the line using the coordinates of points A and B:

[0119] Ax+By=C#(9)

[0120] Where: A is the slope part, which means A y -B y ; B is another slope part, indicating B x -A x ; C is the intercept part, through the formula C = A·A x +B·A y Calculate. This equation defines the geometric expression of line AB and lays the foundation for the subsequent division of the upper and lower areas.

[0121] 2. Upper and lower area segmentation:

[0122] Based on the calculated AB line equation, all data points are divided into two areas, upper and lower. Using the equation A·A x +B·A y -C value is used to determine the position of the point. If the value is greater than 0, the point is located in the upper area, otherwise, the point is located in the lower area. The formula is:

[0123] P upper ={(x i ,y i )|A·x i +B·y i ―C>0}# (10)

[0124] P lower ={(x i ,y i )|A·x i +B·y i ―C≤0}# (11)

[0125] This segmentation provides a basis for further extracting the key points C and D of the upper and lower parts. Figure 6 A schematic diagram of the upper and lower region segmentation is given.

[0126] 3. Selection of key points C and D:

[0127] After completing the region segmentation, calculate the 10 data points farthest from the AB line in the upper and lower regions. Use the Euclidean distance formula to calculate the distance between the data point and the AB line:

[0128]

[0129] Among them, d i represents the distance from the i-th data point to the AB line, and A, B, and C are the coefficients of the equation of the AB line. Calculate the average coordinates of the 10 points farthest from the AB line in the upper and lower halves respectively:

[0130]

[0131]

[0132] Where: C represents the average position of the 10 farthest points in the upper half; D represents the average position of the 10 farthest points in the lower half; (x i,upper ,y i,upper ) is the coordinate of the upper half of the data point; (x i,lower ,y i,lower ) are the coordinates of the data points in the lower half.

[0133] Figure 7A schematic diagram of the positions of grouped key corner points A, B, C, and D is given.

[0134] Preliminary quadrilateral fitting

[0135] This part corresponds to Figure 1 The S2 step in .

[0136] 1. Calculate the angle of the key point relative to the center of mass:

[0137] After determining the center of mass and key points A, B, C, and D, we need to calculate the angle of each key point relative to the center of mass. This angle represents the angle between the vector from the center of mass to each key point and the x-axis. The formula is as follows:

[0138] θ=arctan 2(y―y c ,x―x c )#(15)

[0139] Where: (x, y) is the coordinate of the point; (x c ,y c ) are the coordinates of the center of mass; arctan2 is a function that calculates the angle between the vector from the point to the center of mass and the x-axis.

[0140] 2. Grouping of data points:

[0141] All data points are divided into four groups based on the angle between each key point and the centroid. The angle interval is divided according to the four key points, specifically:

[0142] Group 1: Angles between the centroid and point A and between the centroid and point C

[0143] Group 2: Angles between the centroid and point C and between the centroid and point B

[0144] Group 3: Angles between the centroid and point B and between the centroid and point D

[0145] Group 4: Angles between the centroid and point D and between the centroid and point A

[0146] After grouping, the data points in each group represent different areas, which are ready for line fitting and vertex calculation. Figure 8 shown.

[0147] 3. Linear fitting and quadrilateral vertex calculation:

[0148] For each set of data points, the least squares method is used to perform linear fitting to fit the best straight line for each set. The equation of the fitted line is:

[0149] y=kx+b#(16)

[0150] Where: k is the slope of the line; b is the intercept of the line.

[0151] The calculation formula of the least squares method is:

[0152]

[0153] Where: N is the number of points; ∑xy is the sum of the products of the x-values ​​and y-values ​​of all points; ∑x and ∑y are the sum of the x-values ​​and y-values ​​respectively; ∑(x 2 ) is the sum of the squares of the x values.

[0154] After completing the linear fitting of each set of data, calculate the intersection of adjacent straight lines and determine the vertices EFGH of the quadrilateral. The calculation formula for the intersection is:

[0155]

[0156] Where: k1 and k2 are the slopes of the two straight lines; b1 and b2 are the intercepts of the two straight lines.

[0157] Through the intersection of adjacent straight lines, the vertices E, F, G, and H of the quadrilateral are finally determined. At this point, the preliminary fitting quadrilateral is completed, and the result is visualized as follows: Figure 9 shown.

[0158] Quadratic fitting

[0159] This part corresponds to Figure 1 Step S3 in .

[0160] 1. Linear translation and filtering data points:

[0161] After the four sides of the quadrilateral are obtained through preliminary fitting, these sides need to be translated to improve the fitting accuracy. Assume that the equation of the initial straight line is: y = kx + b#(19)

[0162] Here, k is the slope of the line and b is the intercept of the line.

[0163] To improve the fitting accuracy, the line is translated perpendicularly to the line, and the translation distance is d. Since the unit vector perpendicular to the line is (-k, 1), the new line intercept can be calculated using the following formula:

[0164] Where b1 and b2 represent the intercepts of the two straight lines after upward and translation, respectively, and d is the translation distance. The value used in this study is d = 7. After translation, the key data points between the two translated lines are selected. These points will be used for subsequent fitting. The data points are selected based on the following conditions:

[0165] b2≤y i ―kx i ≤b1#(21)

[0166] The points that meet this condition are the set of points located between the translated lines.

[0167] 2. Linear fitting again:

[0168] For the filtered data points {(x i ,y i )}, re-fit the line and use the least squares method to calculate the new slope k new and intercept b new , the formula is as follows:

[0169] Where: N is the number of filtered data points; x i and y i They are i The horizontal and vertical coordinates of the point; k new is the slope of the new line; b new is the new intercept of the line.

[0170] By refitting the new data points, the optimized straight line equation can be obtained. Figure 10 A schematic diagram of the optimization fitting of A1B1 is given. This process is repeated for each side of the quadrilateral to obtain the optimized four sides.

[0171] 3. Calculation of optimized vertices:

[0172] After completing the optimized straight line fitting, it is necessary to calculate the intersection of each two adjacent straight lines to determine the optimized quadrilateral vertices. Assume that the equations of the two straight lines are:

[0173] y=k1x+b1#(23)

[0174] y=k2x+b2#(24)

[0175] Intersection (x0,y 0) It can be obtained by solving the following system of equations:

[0176]

[0177] Among them: k1 and k2 are the slopes of the two straight lines; b1 and b2 are the intercepts of the two straight lines; x0 and y0 are the horizontal and vertical coordinates of the intersection point respectively.

[0178] By optimizing the fitting and intersection calculation of the four sides, the optimized quadrilateral vertices A1B1C1D1 are obtained, thereby achieving a more accurate fitting of the quadrilateral. Figure 11 A schematic diagram of the fitted quadrilateral and the intercept PQ between the horizontal line where the centroid is located and the quadrilateral is given.

[0179] Calculation of equivalent parameters of DBD ozone generator

[0180] This part corresponds to Figure 1 Step S4 in .

[0181] The equivalent parameters of DBD ozone generator mainly include: air gap equivalent capacitance, barrier dielectric capacitance, discharge maintenance voltage and discharge energy per cycle. The specific calculation formula is as follows:

[0182] (1) Calculation of equivalent capacitance

[0183] To ensure accurate calculation of the equivalent capacitance, the average of the relative side slopes is used as the approximate slope of the Lissajous parallelogram. Assume that the four vertices of quadrilateral A1B1C1D1 are A1(x1, y1), B1(x2, y2), C1(x3, y3), and D1(x4, y4). Equations (16) and (17) give the expressions for the average relative side slopes.

[0184]

[0185]

[0186] in represents the average slope of edge A1B1 and edge C1D1, represents the average slope of the edge B1C1 and the edge D1A1, It represents the slope of side A1B1. It represents the slope of the edge C1D1. It represents the slope of side B1C1. It represents the slope of side D1A1.

[0187] Once the average slope of the relative side is obtained, the equivalent capacitance, dielectric barrier capacitance, and air gap capacitance of the DBD type ozone generator can be calculated based on the attenuation coefficient in the measurement circuit.

[0188] (2) Calculation of discharge maintenance voltage

[0189] Center of mass C(C x ,C y ) is located on the horizontal line y=C y The intercept with the side of the quadrilateral can be used to calculate the discharge maintenance voltage of the dielectric barrier discharge (DBD). Let the intersection of the line where the center of mass lies and the quadrilateral be P(x P ,C y ) and Q(x Q ,C y ), then the difference in their horizontal coordinates |x P -x Q Half of ∣ is the discharge maintenance voltage U of the DBD type ozone generator Z .

[0190] (3) Discharge energy per cycle

[0191] The discharge energy per cycle of a DBD ozone generator can be obtained by calculating the area of ​​the quadrilateral A1B1C1D1. Assuming that the vertices of the quadrilateral are arranged in counterclockwise order, its area S can be calculated using formula (18):

[0192]

[0193] Among them, (x1,y1), (x2,y2), (x3,y3), and (x4,y4) are the coordinates of the vertices of the quadrilateral respectively.

Claims

1. A DBD ozone generator parameter measurement method based on image processing, It is characterized by the following steps: Preprocessing the Lissajous image of the DBD ozone generator, wherein the preprocessed image only contains a valid set of white data points; Obtaining grouped key corner points A, B, C, and D based on the white data point set; Divide the white data point set into 4 arrays, perform straight line fitting on the data point set of each array, and preliminarily construct a quadrilateral; Based on the initially constructed quadrilateral, the boundary line including the new data point set is translated up and down, and a quadratic fitting is performed to construct a quadrilateral with higher accuracy; Calculate and output the equivalent capacitance, discharge holding voltage, and discharge energy per cycle of the DBD type ozone generator; The step of "based on the initially constructed quadrilateral, translating the boundary line up and down to include a new set of data points, and performing quadratic fitting to construct a quadrilateral with higher accuracy" includes the following sub-steps: The straight line corresponding to one side of the initially constructed quadrilateral is translated up and down to form two new translated straight lines; Filtering out a set of data points located between the two new translation lines; The screened data point set is again fitted using the least squares method to obtain an optimized expression for one side of the initially constructed quadrilateral; Repeating the optimization expression obtaining process for the remaining three sides of the initially constructed quadrilateral to obtain the optimized expressions for the remaining three sides; According to the four optimization expressions, the intersection points of adjacent edges are calculated, and finally the optimized quadrilateral A1B1C1D1 is constructed.

2. The method for measuring parameters of a DBD type ozone generator based on image processing according to claim 1, characterized in that: The step of "preprocessing the Lissajous image of the DBD type ozone generator, wherein the preprocessed image only contains a valid set of white data points" includes the following sub-steps: intercepting the Lissajous image generated by the oscilloscope to obtain a sub-image containing only a valid data area; Convert the sub-image from RGB to HSV to extract color data points and generate a logical mask image; Gray-scaling and binarization processing is performed on the logical mask image to obtain an image retaining only white data points; The image with the white data points retained is subjected to denoising processing using the DBSCAN clustering algorithm to obtain a denoised binary image.

3. The method for measuring parameters of a DBD type ozone generator based on image processing according to claim 1, characterized in that: The step of "obtaining grouped key corner points A, B, C, and D based on the white data point set" includes the following sub-steps: Calculate the centroid of the white data points; Based on the centroid, the data points are divided into left and right regions, and the average positions of the 10 points farthest from the centroid are selected in the left and right regions respectively to obtain key corner points A and B; A straight line AB is constructed from the corner points A and B, and all data points are divided into upper and lower areas through the straight line AB. The average positions of the 10 points farthest from the straight line AB are selected in the upper and lower areas respectively to obtain key corner points C and D.

4. The method for measuring parameters of a DBD type ozone generator based on image processing according to claim 1, characterized in that: The step of "dividing the white data point set into four arrays, performing straight line fitting on the data point set of each array, and preliminarily constructing a quadrilateral" includes the following sub-steps: Calculate the angles of the lines connecting the key corner points A, B, C, D and the center of mass relative to the horizontal axis of the center of mass; Divide the white data point set into four arrays according to the angle; wherein Group 1 is a data point set within the angular range between the centroid and the key corner point A and the centroid and the key corner point C, Group 2 is a data point set within the angular range between the centroid and the key corner point C and the centroid and the key corner point B, Group 3 is a data point set within the angular range between the centroid and the key corner point B and the centroid and the key corner point D, and Group 4 is a data point set within the angular range between the centroid and the key corner point D and the centroid and the key corner point A; The least square method is used to fit each data point set to obtain four straight lines; According to the four straight lines, the intersection points of adjacent straight lines are calculated respectively to obtain four vertices E, F, G, and H, and based on them, a preliminary fitting quadrilateral EFGH is constructed.

5. The method for measuring parameters of a DBD type ozone generator based on image processing according to claim 4, characterized in that: The step of "calculating and outputting the equivalent capacitance, discharge maintenance voltage, and discharge energy per cycle of the DBD type ozone generator" includes the following sub-steps: Calculate the average slopes of the opposite sides of the quadrilateral A1B1C1D1, and deduce the equivalent capacitance, dielectric barrier capacitance, and air gap capacitance of the DBD type ozone generator based on the attenuation coefficient in the measurement circuit; The discharge maintenance voltage of the DBD type ozone generator is calculated by taking half the absolute value of the difference between the horizontal straight line where the center of mass is located and the two intersection points of the quadrilateral A1B1C1D1 and combining it with the attenuation coefficient; By calculating the area of ​​the quadrilateral A1B1C1D1, the discharge energy per cycle of the DBD type ozone generator is obtained.

6. A DBD type ozone generator parameter measurement system based on image processing, characterized in that: The method for measuring parameters of a DBD ozone generator based on image processing as claimed in claim 1 comprises the following modules: Image preprocessing module: used for extracting valid data point areas from the Lissajous image generated by the oscilloscope and performing denoising processing; Preliminary fitting module: used to obtain the key corner points A, B, C, and D according to the distribution of data points, group the data points and perform straight line fitting to construct a preliminary fitting quadrilateral; Quadratic fitting and parameter output module: used to construct a quadratic fitting quadrilateral with higher accuracy based on the preliminary fitting quadrilateral, and calculate and output relevant parameters of the DBD type ozone generator.

7. The DBD type ozone generator parameter measurement system based on image processing according to claim 6, characterized in that: The "image preprocessing module" includes the following submodules: Graphics interaction submodule: used to save the captured area image; Data point extraction submodule: used for color space conversion, extracting data points and binarizing images; Noise point denoising submodule: used to denoise noise points using the DBSCAR clustering algorithm.

8. The DBD type ozone generator parameter measurement system based on image processing according to claim 7 is characterized in that: The "preliminary fitting module" includes the following submodules: Centroid calculation and region segmentation submodule: used to calculate the centroid of white data points and segment the data point area; Key point A, B selection submodule: used to determine the point A and the point B; Key point C, D selection submodule: used to determine the point C and the point D; Angle calculation and grouping submodule: used to calculate the angles between the vectors from the centroid to the key corner points A, B, C, and D and the horizontal axis, and divide the white data points into four groups based on these angles; Linear fitting submodule: used to perform least squares linear fitting on each set of data points; Quadrilateral vertex calculation submodule: used to calculate the intersection of every two adjacent straight lines, obtain the vertices of the quadrilateral, and complete the preliminary fitting.

9. A DBD type ozone generator parameter measurement system based on image processing according to any one of claims 6 to 8, characterized in that: The "quadratic fitting and parameter output module" includes the following submodules: Linear translation and data screening submodule: used to translate each initially fitted straight line up and down, and obtain a new set of data points by screening key data points that meet the translation range; Quadratic linear fitting submodule: used to perform least squares linear fitting on the filtered data point set and calculate the new slope and intercept; Optimized vertex calculation submodule: used for calculating the intersection points of adjacent straight lines based on the straight lines after quadratic fitting, obtaining the vertices of the quadratic fitting quadrilateral, and obtaining the quadratic fitting quadrilateral; Equivalent capacitance calculation submodule: used to calculate the equivalent capacitance of the DBD type ozone generator, including dielectric barrier capacitance and air gap capacitance; Discharge maintenance voltage calculation submodule: used to calculate the discharge maintenance voltage of the DBD type ozone generator; Discharge energy calculation submodule: used to calculate the discharge energy per cycle.

Images