A method for quality assessment of coatings based on image processing

Abstract

The present application relates to the field of image processing technology, and in particular to a method for evaluating the quality of a coating based on image processing, the method comprising: scanning the coating to obtain a coating splicing image during the rotation of the roller; filtering the coating splicing image to obtain a correction value and correction confidence of each pixel; performing anomaly detection on the correction value to obtain the abnormal probability of each pixel; marking the pixel points whose abnormal probability is greater than the abnormal threshold or whose correction confidence is less than the confidence threshold as defective points, and taking the number ratio of defective points as the evaluation result. Through the technical solution of the present application, the quality of the coating can be accurately evaluated.

Description

Technical Field

[0001] The present application relates to the technical field of image processing, and in particular to a method for evaluating the quality of a coating based on image processing. Background Art

[0002] In the field of painting, pigments such as gouache and watercolor usually need to be painted on canvas to ensure that the pigments can be stably adhered to create the desired artistic effect. In order to improve the quality and durability of the painting, the canvas needs to be coated to improve the pigment adhesion of the canvas and prevent the pigment from penetrating to the back of the canvas.

[0003] At present, a patent application document with publication number CN115115646A discloses a waterproof cloth coating quality evaluation method based on image processing, including: obtaining a gradient map corresponding to a grayscale map of the waterproof cloth surface coating, obtaining the similarity between each pixel and a target pixel according to the gradient characteristics of each pixel in a texture neighborhood pixel set of each pixel in the gradient map, and then obtaining a position reference value and a grayscale reference value of each pixel relative to the target pixel according to the distance and grayscale value difference between each pixel and the target pixel, and further obtaining a reference weight of each pixel relative to the target pixel, and obtaining the degree of easy peeling of each target pixel by using the reference weight of each pixel relative to the target pixel and the similarity between each pixel and the target pixel, and calculating the quality coefficient of the waterproof cloth surface coating according to the degree of easy peeling of each target pixel and judging the quality of the waterproof cloth coating.

[0004] The above method calculates the degree of peelability of the surface coating of the waterproof cloth by analyzing the texture clarity and regularity in the grayscale image, thereby realizing the quality assessment of the surface coating. However, the above method ignores the monitoring of the grayscale image acquisition process of the surface coating and cannot guarantee the accuracy of the grayscale image of the surface coating, which makes the quality assessment of the coating inaccurate. Summary of the invention

[0005] In order to solve the technical problem of inaccurate quality assessment of applied coatings, the present application provides a quality assessment method for applied coatings based on image processing, which can accurately assess the quality of applied coatings.

[0006] In a first aspect of the present application, a method for evaluating the quality of a coating based on image processing is provided, wherein the coating is tensioned on a roller and rotates with the roller, and the roller is provided with a vibration sensor, and the evaluation method comprises: scanning the coating to obtain a coating splicing image during the rotation of the roller, wherein the coating splicing image comprises a plurality of target pixel rows, and each target pixel row corresponds to an amplitude; filtering the coating splicing image to obtain a correction value and a correction confidence of each pixel point, comprising: performing ordered sample clustering on the amplitude sequence to obtain a plurality of sequence segments; calculating the degree of abnormality of each target pixel row, wherein the degree of abnormality is negatively correlated with the length of the sequence segment to which it belongs, and the similarity between the sequence segment to which it belongs and other sequence segments; filtering each pixel point in a vertical neighborhood based on the degree of abnormality to obtain a correction value; the correction confidence is negatively correlated with the average degree of abnormality in the vertical neighborhood; performing abnormality detection on the correction value to obtain the abnormal probability of each pixel point; marking the pixel point whose abnormal probability is greater than the abnormal threshold or whose correction confidence is less than the confidence threshold as a defective point, and taking the proportion of the number of defective points as the evaluation result.

[0007] The coating is tensioned on the roller and rotated with the roller. First, the coating in the production process is scanned by an image acquisition device, and the scanned multiple target pixel rows are spliced ​​into a coating splicing image, and each target pixel row corresponds to an amplitude, which can reflect the shaking state of the coating when the target pixel row is collected; the abnormality degree of each target pixel row is determined according to the amplitude sequence, and the greater the abnormality degree, the abnormal shaking state when the corresponding target pixel row is collected, and the coating characteristics in the target pixel row are inaccurate. Therefore, the pixel values ​​are weighted and summed in the vertical neighborhood according to the abnormality degree to obtain the correction value and correction confidence of each pixel point. The correction value can represent the accurate coating characteristics, and the correction confidence is used to represent the degree of credibility of the correction value of a pixel point to accurately represent the coating characteristics; further, the correction value is detected for abnormality to obtain the abnormal probability of each pixel point; the pixel points whose abnormal probability is greater than the abnormal threshold or the correction confidence is less than the confidence threshold are marked as defective points, and the proportion of the number of defective points is used as the evaluation result to accurately evaluate the quality of the coating.

[0008] Preferably, scanning the coating to obtain a coating spliced ​​image includes: for any coating row in the coating, obtaining all local images containing the coating row and the amplitude of each local image; taking the pixel row of the coating row in the local image with the minimum amplitude as the target pixel row, and taking the minimum amplitude as the amplitude of the target pixel row; arranging all target pixel rows in order from top to bottom according to the coating rows to obtain a coating spliced ​​image.

[0009] Use image acquisition equipment to scan the coating in the production process, collect the target pixel rows corresponding to each row on the coating, and the amplitude of the target pixel rows is minimized to ensure the collection accuracy of the target pixel rows. After splicing the target pixel rows, the coating splicing image can be obtained. The coating splicing image can truly reflect the coating characteristics of the coating surface.

[0010] Preferably, the target pixel row The degree of abnormality for:

[0011] , The target pixel row The length of the sequence segment to which it belongs, is the total number of sequence segments, The target pixel row Sequence segment and sequence segment The similarity of For sequence segment Length, is the length of the amplitude sequence, is the average length of the sequence segment.

[0012] Target pixel row The larger the length of the sequence segment, the larger the target pixel row. The longer the amplitude lasts, the greater the probability that the amplitude is normal. The smaller the abnormality, the greater the similarity between the sequence segment and other sequence segments, indicating that the target pixel row The greater the frequency of the amplitude in the entire amplitude sequence, the greater the probability that the amplitude is normal. The smaller the abnormality, the more accurate the quantification of the abnormality can be achieved by combining the duration and the frequency in the entire amplitude sequence.

[0013] Preferably, the vertical neighborhood includes a preset number of adjacent pixels on the upper and lower sides.

[0014] Preferably, the target pixel row Medium pixel Correction value for:

[0015] , The row of target pixels in the vertical neighborhood The set of target pixel rows other than The target pixel row The degree of abnormality, For collection The sum of the abnormality levels of each target pixel row in The target pixel row The target pixel row The distance For collection The sum of the distances of each target pixel row in The target pixel row Medium pixel The pixel value of .

[0016] For any pixel point, it is corrected according to the pixel values ​​of the adjacent pixels on the upper and lower sides. A larger weight is assigned to the pixel point with a smaller abnormality, and a larger weight is assigned to the pixel point with a closer distance. The corrected value of the pixel point is obtained so that the corrected value can represent the accurate coating characteristics and avoid the influence of amplitude abnormality on the coating characteristics.

[0017] Preferably, the modified confidence is equal to 1 minus the average degree of abnormality in the vertical neighborhood.

[0018] If the average abnormality in the vertical neighborhood is greater, it means that the vertical neighborhood contains fewer accurate coating features, and therefore, the correction confidence is lower, thereby achieving accurate quantification of the correction confidence.

[0019] Preferably, performing anomaly detection on the correction value to obtain the anomaly probability of each pixel point includes: using an anomaly detection algorithm to detect the correction values ​​of all pixels in the coated spliced ​​image to obtain the anomaly probability of each pixel point.

[0020] Preferably, performing anomaly detection on the correction value to obtain the anomaly probability of each pixel point includes: using an anomaly detection algorithm to detect the correction value of the target pixel row where any pixel point is located to obtain the anomaly probability of the pixel point.

[0021] For pixels in the same target pixel row, their amplitudes are exactly the same. Anomaly detection is performed only in the target pixel row, which can not only accurately obtain the abnormal probability of the pixel, but also reduce the amount of calculation for anomaly detection.

[0022] Preferably, the anomaly detection algorithm is a LOF algorithm or an isolation forest algorithm.

[0023] The technical solution of this application has the following beneficial technical effects:

[0024] The coating is tensioned on the roller and rotated with the roller. First, the coating in the production process is scanned by an image acquisition device, and the scanned multiple target pixel rows are spliced ​​into a coating splicing image, and each target pixel row corresponds to an amplitude, which can reflect the shaking state of the coating when the target pixel row is collected; the abnormality degree of each target pixel row is determined according to the amplitude sequence, and the greater the abnormality degree, the abnormal shaking state when the corresponding target pixel row is collected, and the coating characteristics in the target pixel row are inaccurate. Therefore, the pixel values ​​are weighted and summed in the vertical neighborhood according to the abnormality degree to obtain the correction value and correction confidence of each pixel point. The correction value can represent the accurate coating characteristics, and the correction confidence is used to represent the degree of credibility of the correction value of a pixel point to accurately represent the coating characteristics; further, the correction value is detected for abnormality to obtain the abnormal probability of each pixel point; the pixel points whose abnormal probability is greater than the abnormal threshold or the correction confidence is less than the confidence threshold are marked as defective points, and the proportion of the number of defective points is used as the evaluation result to accurately evaluate the quality of the coating. BRIEF DESCRIPTION OF THE DRAWINGS

[0025] By reading the detailed description below with reference to the accompanying drawings, the above and other purposes, features and advantages of the exemplary embodiments of the present application will become easy to understand. In the accompanying drawings, several embodiments of the present application are shown in an exemplary and non-limiting manner, and the same or corresponding reference numerals represent the same or corresponding parts, wherein:

[0026] Figure 1 is a schematic diagram of scanning a coating according to an embodiment of the present application;

[0027] Figure 2 is a flow chart of a method for evaluating the quality of a coating based on image processing according to an embodiment of the present application;

[0028] Figure 3 is a schematic diagram of a scanned image according to an embodiment of the present application. DETAILED DESCRIPTION

[0029] The following will be combined with the drawings in the embodiments of the present application to clearly and completely describe the technical solutions in the embodiments of the present application. Obviously, the described embodiments are part of the embodiments of the present application, not all of the embodiments. Based on the embodiments in the present application, all other embodiments obtained by those skilled in the art without creative work are within the scope of protection of the present application.

[0030] It should be understood that when the terms "first", "second", etc. are used in the claims, specification and drawings of the present application, they are only used to distinguish different objects, rather than to describe a specific order. The terms "include" and "comprise" used in the specification and claims of the present application indicate the presence of the described features, wholes, steps, operations, elements and / or components, but do not exclude the presence or addition of one or more other features, wholes, steps, operations, elements, components and / or their collections.

[0031] According to the first aspect of the present application, the present application provides a coating quality assessment method based on image processing, which is used to determine whether the coating on the coated surface meets the requirements, wherein the coating is a canvas or any other cloth with a surface coating. Figure 1 , is a schematic diagram of scanning a coating according to an embodiment of the present application, wherein the coating 200 is stretched on a roller 100, and the roller 100 rotates at a preset speed, thereby driving the coating 200 to move together, and an image acquisition device 300 is placed on one side of any roller 100 and on the same horizontal plane as the roller 100, so that the field of view of the image acquisition device 300 can cover the coating 200 on the roller 100.

[0032] The image acquisition device 300 is an industrial camera, a mobile phone or other equipment with an image acquisition function. The roller 100 is also provided with a vibration sensor for acquiring the amplitude of the roller 100 .

[0033] Figure 2 FIG. 1 is a flow chart of a method for evaluating the quality of a coating layer based on image processing according to an embodiment of the present application. Figure 2 As shown, the coating quality assessment method based on image processing includes steps S101 to S104, which are described in detail below.

[0034] S101, during the rotation of the roller, scanning the coating to obtain a coating splicing image, wherein the coating splicing image includes a plurality of target pixel rows, and each target pixel row corresponds to an amplitude.

[0035] In one embodiment, during the rotation of the roller at a preset speed, the image acquisition device acquires the coating on the roller at a fixed posture. Since the image acquisition device is fixed and the roller rotates at a preset speed, the image acquisition device continuously acquires multiple scanned images, and there is an intersection area between adjacent scanned images. The size of the intersection area is related to the preset speed. For example, the acquisition frequency of the image acquisition device is adjusted, and a scanned image is acquired when the coating moves a distance of one pixel in the scanned image. See Figure 3 , is a schematic diagram of a scanned image according to an embodiment of the present application, at time The collected local image and time The collected local image There is an intersection area between them, and the local image Relative to the local image For example, the distance of moving one pixel is the same as that of moving one pixel. The scanned images at each moment are stitched together to obtain the coated stitched image.

[0036] At the same time, according to the vibration sensor, the amplitude corresponding to each moment can be obtained, that is, one local image corresponds to one amplitude, which can reflect the vibration condition of the roller when the local image is collected.

[0037] Specifically, the method of scanning the coating to obtain a coating splicing image includes: for any coating row in the coating, obtaining all local images containing the coating row and the amplitude of each local image; taking the pixel row of the coating row in the local image with the minimum amplitude as the target pixel row, and taking the minimum amplitude as the amplitude of the target pixel row; arranging all target pixel rows in order from top to bottom according to the coating rows to obtain a coating splicing image.

[0038] It can be understood that one coating row corresponds to one row on the coating, and all coating rows can cover the entire coating.

[0039] Exemplarily, the height of the local image is 4, that is, a local image contains 4 rows of pixels, and a scanned image is collected every time a pixel is moved; for the first coating row in the coating, the first to fourth local images collected all contain the first coating row, and the local image corresponding to the minimum amplitude in the first to fourth local images is obtained, and the pixel row corresponding to the first coating row in the local image is used as the target pixel row of the first coating row, and the amplitude corresponding to the target pixel row is the minimum amplitude in the first to fourth local images; the target pixel rows of all coating rows are arranged from top to bottom to obtain a coating splicing image.

[0040] In this way, during the production process, the coating is tensioned on the roller and rotates with the roller. The image acquisition equipment is used to scan the coating during the production process, and the target pixel rows corresponding to each row on the coating are collected, and the amplitude of the target pixel rows is minimized to ensure the collection accuracy of the target pixel rows. After splicing the target pixel rows, the coating splicing image can be obtained. The coating splicing image can truly reflect the coating characteristics of the coating surface.

[0041] S102, filtering the coated spliced ​​image to obtain a correction value and correction confidence of each pixel.

[0042] In one embodiment, the coating stitching image includes multiple target pixel rows, and each target pixel row corresponds to an amplitude, which can reflect the shaking state of the coating when the target pixel row is collected. The abnormal shaking state will cause inaccurate coating characteristics in the target pixel row. Therefore, in order to obtain accurate quality assessment results, the coating stitching image needs to be filtered so that each target pixel row can represent accurate coating characteristics.

[0043] Specifically, the coated spliced ​​image is filtered to obtain the correction value and correction confidence of each pixel point, including: orderly sample clustering of the amplitude sequence to obtain multiple sequence segments, the amplitude sequence includes the amplitude of each target pixel row; calculating the abnormality of each target pixel row, the abnormality is negatively correlated with the length of the sequence segment to which it belongs, and the similarity between the sequence segment to which it belongs and other sequence segments; filtering each pixel point in the vertical neighborhood based on the abnormality to obtain a correction value, the correction confidence is negatively correlated with the average abnormality in the vertical neighborhood, the vertical neighborhood includes a preset number of adjacent pixels on the upper and lower sides. Wherein, the vertical neighborhood includes 3 adjacent pixels on both sides in the vertical direction, that is, the vertical neighborhood includes 6 pixels.

[0044] The target pixel row The degree of abnormality for: , The target pixel row The length of the sequence segment to which it belongs, is the total number of sequence segments, The target pixel row Sequence segment and sequence segment The similarity of For sequence segment Length, is the length of the amplitude sequence, is the average length of the sequence segment. Each sequence segment corresponds to a mean value of abnormality, similarity It is negatively correlated with the absolute value of the difference in the mean abnormality between the two sequence segments.

[0045] Understandably, the target pixel row The larger the length of the sequence segment, the larger the target pixel row. The longer the amplitude lasts, the greater the probability that the amplitude is normal. The smaller the abnormality, the greater the similarity between the sequence segment and other sequence segments, indicating that the target pixel row The greater the frequency of the amplitude in the entire amplitude sequence, the greater the probability that the amplitude is normal. In other embodiments, the target pixel row The degree of abnormality You can also The meaning of each symbol in this formula is the same as the above formula and will not be repeated here.

[0046] In one embodiment, the target pixel row Medium pixel Correction value for:

[0047] , The row of target pixels in the vertical neighborhood The set of target pixel rows other than The target pixel row The degree of abnormality, For collection The sum of the abnormality levels of each target pixel row in The target pixel row The target pixel row The distance For collection The sum of the distances of each target pixel row in The target pixel row Medium pixel The pixel value of .

[0048] It should be noted that the correction confidence is used to characterize the degree of credibility of the correction value of a pixel point to accurately characterize the coating feature; if the average abnormality in the vertical neighborhood is greater, it means that the vertical neighborhood contains fewer accurate coating features, therefore, the correction confidence is lower. When the correction confidence is less than the confidence threshold, it means that the coating feature of the corresponding pixel point cannot be accurately collected. To avoid missed detection, the pixel point is directly marked as a defective point. Specifically, the correction confidence is equal to 1 minus the average abnormality in the vertical neighborhood.

[0049] In this way, the degree of abnormality of each target pixel row is calculated according to the abnormal situation of the amplitude corresponding to each target pixel row. When the amplitude is in an abnormal situation, the corresponding target pixel row is less able to obtain accurate coating characteristics, and therefore, the degree of abnormality of the target pixel row is greater; further, for any pixel point, it is corrected according to the pixel values ​​of the adjacent pixel points on the upper and lower sides, and a larger weight is assigned to the pixel point with a smaller degree of abnormality, and a larger weight is assigned to the pixel point with a closer distance, so as to obtain the corrected value of the pixel point, so that the corrected value can represent the accurate coating characteristics and avoid the influence of the amplitude abnormality on the coating characteristics.

[0050] S103, performing anomaly detection on the correction value to obtain the abnormality probability of each pixel point.

[0051] In one embodiment, after the correction value of each pixel is obtained, the correction value can accurately characterize the coating characteristics; the correction value of each pixel is detected using an abnormality detection algorithm to obtain the abnormality probability of each pixel in the coating splicing image.

[0052] The anomaly detection algorithm is a LOF algorithm or an isolation forest algorithm.

[0053] In another embodiment, since the correction value is calculated based on the pixel value of each pixel in the vertical neighborhood of the pixel, and the amplitudes of the pixels in the same target pixel row are exactly the same, anomaly detection can be performed only in the target pixel row to determine the abnormal probability of each pixel in the target pixel row. Specifically, performing anomaly detection on the correction value to obtain the abnormal probability of each pixel includes: using an anomaly detection algorithm to detect the correction value of the target pixel row where any pixel is located to obtain the abnormal probability of the pixel.

[0054] In this way, the abnormal probability of the pixel point can be accurately obtained and the calculation amount of anomaly detection can be reduced.

[0055] S104, marking the pixel points whose abnormal probability is greater than the abnormal threshold or whose corrected confidence is less than the confidence threshold as defect points, and taking the number ratio of the defect points as the evaluation result.

[0056] In one embodiment, the value of the abnormality threshold is 0.5, and the value of the confidence threshold is 0.5. After marking the pixels whose abnormal probability is greater than the abnormality threshold or whose corrected confidence is less than the confidence threshold as defective points, the number of defective points in the coating can be obtained, and the proportion of the number of defective points is used as the evaluation result, which accurately reflects the quality of the coating.

[0057] The technical principle and implementation details of a coating quality assessment method based on image processing of the present application are introduced above through specific embodiments. The coating is tensioned on the roller and rotates with the roller. First, the coating in the production process is scanned by an image acquisition device, and the scanned multiple target pixel rows are spliced ​​into a coating splicing image, and each target pixel row corresponds to an amplitude, which can reflect the shaking state of the coating when the target pixel row is collected; the abnormality degree of each target pixel row is determined according to the amplitude sequence, and the greater the abnormality degree, the abnormal shaking state when the corresponding target pixel row is collected, and the coating characteristics in the target pixel row are inaccurate. Therefore, the pixel values ​​are weighted and summed in the vertical neighborhood according to the abnormality degree to obtain the correction value and correction confidence of each pixel point. The correction value can characterize the accurate coating characteristics, and the correction confidence is used to characterize the degree of credibility of the correction value of a pixel point to accurately characterize the coating characteristics; further, the correction value is detected for abnormality to obtain the abnormal probability of each pixel point; the pixel points whose abnormal probability is greater than the abnormal threshold or the correction confidence is less than the confidence threshold are marked as defective points, and the proportion of the number of defective points is used as the evaluation result to accurately evaluate the quality of the coating.

[0058] The technical features of the above-described embodiments may be arbitrarily combined. To make the description concise, not all possible combinations of the technical features in the above-described embodiments are described. However, as long as there is no contradiction in the combination of these technical features, they should be considered to be within the scope of this specification.

[0059] The above-described embodiments only express several implementation methods of the present application, and the descriptions thereof are relatively specific and detailed, but they cannot be construed as limiting the scope of the patent application. It should be pointed out that, for a person of ordinary skill in the art, several variations and improvements can be made without departing from the concept of the present application, and these all belong to the protection scope of the present application. Therefore, the protection scope of the patent application shall be subject to the attached claims.

Claims

1. A method for evaluating the quality of a coating based on image processing, characterized in that: The coating is tensioned on the roller and rotates with the roller, and the roller is provided with a vibration sensor. The evaluation method includes: During the rotation of the roller, the coating is scanned to obtain a coating splicing image, wherein the coating splicing image includes a plurality of target pixel rows, and each target pixel row corresponds to an amplitude; The coated spliced ​​image is filtered to obtain a correction value and a correction confidence of each pixel point, including: performing ordered sample clustering on the amplitude sequence to obtain multiple sequence segments; calculating the abnormality degree of each target pixel row, wherein the abnormality degree is negatively correlated with the length of the sequence segment to which it belongs, and the similarity between the sequence segment to which it belongs and other sequence segments; filtering each pixel point in a vertical neighborhood based on the abnormality degree to obtain a correction value; the correction confidence degree is negatively correlated with the average abnormality degree in the vertical neighborhood; Perform anomaly detection on the correction value to obtain the anomaly probability of each pixel; Pixels whose abnormal probability is greater than the abnormal threshold or whose corrected confidence is less than the confidence threshold are marked as defective points, and the number and proportion of defective points are used as the evaluation result; Target pixel row Medium pixel Correction value for: , The row of target pixels in the vertical neighborhood The set of target pixel rows other than The target pixel row The degree of abnormality, For collection The sum of the abnormality levels of each target pixel row in The target pixel row The target pixel row The distance For collection The sum of the distances of each target pixel row in The target pixel row Medium pixel The pixel value of .

2. The method for evaluating the quality of a coating based on image processing according to claim 1, characterized in that: The scanning of the coating to obtain the coating splicing image comprises: For any coating line in coating, acquiring all partial images including the coating line and the amplitude of each partial image; Taking the pixel row of the coating row in the local image of the minimum amplitude value as the target pixel row, and taking the minimum amplitude value as the amplitude of the target pixel row; Arrange all target pixel rows in order from top to bottom according to the coating rows to obtain the coating stitching image.

3. The method for evaluating the quality of a coating based on image processing according to claim 1, characterized in that: Target pixel row The degree of abnormality for: , The target pixel row The length of the sequence segment to which it belongs, is the total number of sequence segments, The target pixel row Sequence segment and sequence segment The similarity of For sequence segment Length, is the length of the amplitude sequence, is the average length of the sequence segments.

4. The method for evaluating the quality of a coating based on image processing according to claim 1, characterized in that: The vertical neighborhood includes a preset number of adjacent pixels on the upper and lower sides.

5. The method for evaluating the quality of a coating based on image processing according to claim 1, characterized in that: The modified confidence is equal to 1 minus the average degree of abnormality in the vertical neighborhood.

6. The method for evaluating the quality of a coating based on image processing according to claim 1, characterized in that: The abnormality detection of the correction value to obtain the abnormality probability of each pixel point includes: The anomaly detection algorithm is used to detect the correction values ​​of all pixels in the coated splicing image to obtain the anomaly probability of each pixel.

7. The method for evaluating the quality of a coating based on image processing according to claim 1, characterized in that: The abnormality detection of the correction value to obtain the abnormality probability of each pixel point includes: An anomaly detection algorithm is used to detect the correction value of the target pixel row where any pixel point is located to obtain the abnormal probability of the pixel point.

8. A coating quality assessment method based on image processing according to claim 6 or 7, characterized in that: The anomaly detection algorithm is a LOF algorithm or an isolation forest algorithm.

Images