Full-automatic intelligent grading method and system for strip steel surface corrosion and storage medium

Abstract

The invention discloses a full-automatic intelligent grading method and system for strip steel surface corrosion and a storage medium, and relates to the field of intelligent detection. Continuously acquiring a strip steel surface image which is being processed in the chemical agent pool according to a preset period, and preprocessing the strip steel surface image to reconstruct the strip steel surface image as an image to be detected; collecting surface color information of the to-be-detected image, determining a suspected corrosion connected domain in combination with the surface color information, and obtaining a gray difference value and a gray change parameter in the suspected corrosion connected domain; edge pixel points of the suspected corrosion connected domain are extracted according to the gray difference value and the gray change parameter, the edge of the corrosion connected domain is determined, and the area of the suspected corrosion connected domain is calculated; and determining the corrosion depth according to the gray difference value and the gray change parameter, and determining the surface corrosion level of the strip steel in combination with the corrosion depth and the corrosion connected domain area. The method can avoid excessive treatment on the strip steel which does not reach the corrosion degree, and saves treatment resources such as chemical agents.

Description

Technical Field

[0001] The present invention relates to the field of intelligent detection, and more specifically to a fully automatic intelligent grading method, system and storage medium for strip surface corrosion. Background Art

[0002] In the steel processing industry chain, steel coils are key intermediate products, and their transformation into finished products such as steel pipes is crucial. Before steel coils are processed into steel pipes, removing surface corrosion products is a necessary process to ensure the quality of the steel pipes, and chemical cleaning is a common method.

[0003] At present, under the given chemical agent pool concentration conditions, the control of the cleaning time of steel coils relies on empirical rules. It is generally believed that the longer the steel coil is immersed in the agent pool, the better the surface corrosion removal effect. However, this method has significant disadvantages. On the one hand, as the immersion time increases, the uncorroded surface metal of the steel coil will be excessively corroded by the chemical agent, causing damage to the steel coil matrix, which directly affects the mechanical properties and service life of the subsequent processing into steel pipes. On the other hand, manual monitoring of the cleaning status of the steel coil is unsustainable. As the chemical agent evaporates during the cleaning process, the surrounding area of ​​the agent pool is filled with smoke and the environment is harsh. If the operator is in this environment filled with irritating chemical smoke for a long time, the body will be greatly harmed, including frequent occupational health risks such as respiratory diseases and eye burns. Moreover, human eye observation is limited by subjective judgment differences, visual fatigue, and smoke blocking the line of sight, making it difficult to accurately determine whether the surface corrosion of the steel coil is cleaned. Even experienced workers cannot guarantee the accuracy and consistency of each judgment, which makes the cleaning quality of the steel coil uneven, burying quality risks for subsequent steel pipe production.

[0004] With the deepening of the concept of intelligent manufacturing and the development of automation technology, all walks of life are seeking intelligent solutions to replace traditional manual operations. In the field of steel coil cleaning, there is an urgent need for an intelligent means that is not disturbed by the smoke environment and can accurately identify the surface corrosion and cleaning conditions of the steel coil, so as to accurately control the time when the steel coil enters the reagent pool. Through intelligent detection and automatic control, it can not only ensure that the steel coil cleaning quality meets the requirements of subsequent processing, but also protect the health of operators to the greatest extent, improve the efficiency and stability of the entire steel processing process, and adapt to the stringent needs of modern industrial production. Summary of the invention

[0005] In view of this, the present invention provides a fully automatic intelligent grading method, system and storage medium for strip surface corrosion. On the one hand, accurate detection can be used to avoid excessive processing of strips that have not reached the corrosion level, saving processing resources such as chemicals; on the other hand, it can reduce the work pollution of operators.

[0006] In order to achieve the above object, the present invention adopts the following technical solution:

[0007] An automatic intelligent grading method for strip steel surface corrosion, comprising the following steps:

[0008] Continuously obtain the surface image of the strip steel being processed in the chemical reagent pool according to a preset period, and preprocess the strip steel surface image according to the illumination correction algorithm and the image correction algorithm to reconstruct the strip steel surface image as the image to be detected;

[0009] Use the color aggregation vector to collect the surface color information of the image to be detected, combine the surface color information to determine the suspected corrosion connected domain, and obtain the gray-scale difference value and the gray-scale change parameter in the suspected corrosion connected domain;

[0010] Extract the edge pixel points of the suspected corrosion connected domain according to the gray-scale difference value and the gray-scale change parameter, connect the edge pixel points to determine the edge of the corrosion connected domain, and calculate the area of the suspected corrosion connected domain;

[0011] Input the gray-scale difference value, the gray-scale change parameter, and the area of the suspected corrosion connected domain into the automatic intelligent grading model, determine the corrosion depth according to the gray-scale difference value and the gray-scale change parameter, and combine the corrosion depth and the area of the corrosion connected domain to determine the strip steel surface corrosion grade.

[0012] Optionally, preprocess the strip steel surface image according to the illumination correction algorithm and the image correction algorithm to reconstruct the strip steel surface image, specifically:

[0013] Extract and compare the strip steel surface image before preprocessing and the strip steel surface image after preprocessing, determine whether there is overprocessing according to the comparison result, and compensate the overprocessing result with the strip steel surface image before preprocessing to obtain the first image;

[0014] Extract the spot feature of the first image and input the first image into the reconstructed image model, estimate the spot feature distribution, obtain the distribution information of the spot, and align the spot distribution information with the first image. The aligned area is the target area, and determine again whether the spot distribution information in the target area is the chemical reagent projection spot or the suspected corrosion connected domain;

[0015] If it is the chemical reagent projection spot, remove the projection spot from the target area and reconstruct the target area image; if it is the suspected corrosion connected domain, enhance the image feature of the target area and strengthen the pixel point presentation of the target area to obtain the reconstructed target area image.

[0016] Optionally, use an industrial CCD camera to continuously obtain the surface image of the strip steel being processed in the chemical reagent pool according to a preset period, specifically:

[0017] Regularly collect image data of a predetermined area through an industrial-grade CCD camera, and map monitoring data of different modalities to a unified spatio-temporal coordinate system in combination with the position and sampling timestamp of the industrial-grade CCD camera to form a standard set of monitoring data.

[0018] Optionally, the specific method for determining suspected connected regions is as follows:

[0019] Segment the surface color information and grayscale it to obtain a grayscale image;

[0020] Perform threshold segmentation on the grayscale image to obtain a binary image, and successively perform morphological processing on the binary image using opening operation and closing operation to obtain a target image, and perform connected region monitoring on the target image to obtain several suspected connected regions.

[0021] Optionally, it further includes using the potential difference scanning method to judge the potential difference of the suspected corrosion connected regions, and determining whether the suspected corrosion connected regions are corrosion connected regions according to the potential difference.

[0022] Optionally, in the fully automatic intelligent grading model, the random forest algorithm is used to classify the corrosion connected regions according to the corrosion area and corrosion depth.

[0023] A fully automatic intelligent grading system for strip steel surface corrosion, including:

[0024] A to-be-detected image reconstruction module: used to continuously obtain the strip steel surface image being processed in the chemical reagent pool at a preset period, and preprocess and reconstruct the strip steel surface image according to the illumination correction algorithm and image correction algorithm as the to-be-detected image;

[0025] A grayscale determination module: used to collect the surface color information of the to-be-detected image using the color aggregation vector, determine the suspected corrosion connected regions in combination with the surface color information, and obtain the grayscale difference value and grayscale change parameter in the suspected corrosion connected regions;

[0026] An area calculation module: used to extract the edge pixel points of the suspected corrosion connected regions according to the grayscale difference value and grayscale change parameter, connect the edge pixel points to determine the edge of the suspected corrosion connected regions, and calculate the area of the suspected corrosion connected regions;

[0027] A grading model corrosion level classification module: input the grayscale difference value, grayscale change parameter, and corrosion connected region area into the fully automatic intelligent grading model, determine the corrosion depth according to the grayscale difference value and grayscale change parameter, and determine the strip steel surface corrosion level in combination with the corrosion depth and corrosion connected region area.

[0028] Optionally, in the to-be-detected image reconstruction module, an industrial CCD camera is used to extract the strip surface image, and the industrial CCD camera is used to continuously obtain the strip surface image being processed in the chemical reagent pool according to a preset period. Specifically: the image data of a predetermined area is regularly collected by the industrial CCD camera, and combined with the position and sampling timestamp of the industrial CCD camera, the monitoring data of different modalities is mapped to a unified spatio-temporal coordinate system to form a standard monitoring data set.

[0029] A computer storage medium, on which a computer program is stored, and when the computer program is executed by a processor, the steps of any one of the strip surface corrosion full-automatic intelligent grading methods are implemented.

[0030] It can be seen from the above technical solutions that, compared with the prior art, the present invention provides a strip surface corrosion full-automatic intelligent grading method, system and storage medium, mainly in two aspects: detection accuracy and efficiency automation. Specifically as follows:

[0031] 1. Continuously obtain the strip surface image according to a preset period, and use the illumination correction algorithm and the image correction algorithm for preprocessing to reconstruct the image, which can effectively remove the interference caused by factors such as uneven illumination and image distortion, so that subsequent various analyses based on the image are all based on relatively accurate and reliable image data, thereby improving the accuracy of the entire corrosion detection and grading. Comprehensively use the color aggregation vector to collect color information to determine the suspected corrosion connected domain, further obtain the gray-scale difference value, gray-scale change parameter and calculate multi-dimensional parameters such as the area of the suspected corrosion connected domain, and then determine the corrosion depth and the final corrosion grade division based on these parameters, avoiding the limitation of single-index judgment. The mutual verification and comprehensive consideration of multiple parameters can more accurately reflect the actual corrosion situation of the strip surface and improve the accuracy of the grading result.

[0032] 2. Automatically and continuously obtain the strip surface image according to a preset period. The whole process does not require frequent manual operation intervention, and can continuously monitor the strip being processed in the chemical reagent pool, greatly improving the detection efficiency, and can timely master the change of the corrosion state of the strip during the processing, which helps to make corresponding adjustments in time. Finally, the relevant parameters are input into the full-automatic intelligent grading model to obtain the corrosion grade, realizing the intelligent grading operation, reducing the workload of manual complex judgment and grading, reducing the error that may be caused by the difference in manual subjective judgment while improving the efficiency, and ensuring the consistency and objectivity of the grading result. Description of the Drawings

[0033] To more clearly illustrate the technical solutions in the embodiments of the present invention or the prior art, the following will briefly introduce the accompanying drawings required for the description of the embodiments or the prior art. Obviously, the accompanying drawings in the following description are only the embodiments of the present invention. For those of ordinary skill in the art, without creative efforts, other accompanying drawings can also be obtained according to the provided accompanying drawings.

[0034] Figure 1 Schematic diagram of the method flow of the present invention;

[0035] Figure 2 Schematic diagram of the system structure of the present invention. Detailed implementation manners

[0036] The following will clearly and completely describe the technical solutions in the embodiments of the present invention with reference to the accompanying drawings in the embodiments of the present invention. Obviously, the described embodiments are only a part of the embodiments of the present invention, rather than all of the embodiments. Based on the embodiments of the present invention, all other embodiments obtained by those of ordinary skill in the art without creative efforts belong to the scope of protection of the present invention.

[0037] The embodiments of the present invention disclose a fully automatic intelligent grading method for strip steel surface corrosion, as Figure 1 shown, including the following steps:

[0038] Step 1: Continuously acquire the strip steel surface image being processed in the chemical reagent pool according to a preset period, and preprocess the strip steel surface image according to the illumination correction algorithm and the image correction algorithm to reconstruct the strip steel surface image as the image to be detected;

[0039] Step 2: Use the color aggregation vector to collect the surface color information of the image to be detected, determine the suspected corrosion connected domain in combination with the surface color information, and obtain the gray-scale difference value and the gray-scale change parameter in the suspected corrosion connected domain;

[0040] Step 3: Extract the edge pixel points of the suspected corrosion connected domain according to the gray-scale difference value and the gray-scale change parameter, connect the edge pixel points to determine the edge of the corrosion connected domain, and calculate the area of the suspected corrosion connected domain;

[0041] Step 4: Input the gray-scale difference value, the gray-scale change parameter, and the area of the suspected corrosion connected domain into the fully automatic intelligent grading model, determine the corrosion depth according to the gray-scale difference value and the gray-scale change parameter, and determine the strip steel surface corrosion grade in combination with the corrosion depth and the area of the corrosion connected domain.

[0042] Further, in Step 1, preprocessing the strip steel surface image according to the illumination correction algorithm and the image correction algorithm to reconstruct the strip steel surface image is specifically:

[0043] Step 1.1: Extract and compare the strip surface images before and after preprocessing. Determine whether there is overprocessing based on the comparison results, and compensate for the overprocessing results using the strip surface image before preprocessing to obtain the first image.

[0044] Step 1.2: Extract the light spot features of the first image and input the first image into the reconstructed image model. Estimate the distribution of the light spot features to obtain the distribution information of the light spots, and align the light spot distribution information with the first image. The aligned area is the target area. Determine again whether the light spot distribution information in the target area is the chemical agent projection light spot or the suspected corrosion connected domain.

[0045] Step 1.3: If it is the chemical agent projection light spot, remove the projection light spot from the target area and reconstruct the target area image; if it is the suspected corrosion connected domain, enhance the image features of the target area and strengthen the presentation of the pixel points in the target area to obtain the reconstructed target area image.

[0046] Furthermore, in Step 1, an industrial-grade CCD camera is used to continuously acquire the strip surface images being processed in the chemical agent pool at a preset period. Specifically: Regularly collect the image data of a predetermined area through the industrial-grade CCD camera, and combine the position and sampling timestamp of the industrial-grade CCD camera to map the monitoring data of different modalities to a unified spatio-temporal coordinate system to form a standard monitoring data set.

[0047] Furthermore, in Step 2, the specific method for determining the suspected connected domain is as follows:

[0048] Step 2.1: Segment the surface color information and grayscale it to obtain a grayscale image.

[0049] Step 2.2: Perform threshold segmentation on the grayscale image to obtain a binary image. Successively perform morphological processing on the binary image using opening operation and closing operation to obtain the target image, and perform connected domain monitoring on the target image to obtain several suspected connected domains.

[0050] Even further, in this embodiment, it also includes using the potential difference scanning method to judge the potential difference of the suspected corrosion connected domain, and determining whether the suspected corrosion connected domain is a corrosion connected domain according to the potential difference. Specifically as follows:

[0051] After collecting the potential difference data, it needs to be processed. First, noise interference needs to be removed. Filtering techniques such as low-pass filtering can be used. If there are high-frequency fluctuations in the scanned potential difference data, by setting a low-pass filter with an appropriate cut-off frequency (such as 100Hz), these high-frequency noises can be filtered out to make the data smoother.

[0052] Calculate statistical parameters such as the average value and standard deviation of the potential difference within the suspected corrosion connected region. Generally speaking, if the average value deviates from the normal strip surface potential difference range (this range can be determined in advance by measuring the potential difference of the non-corroded strip surface. For example, the normal potential difference range is between -0.2V and +0.2V), and the standard deviation is large (such as greater than 0.1V), then this suspected region is very likely to be a corrosion connected region.

[0053] It can also be judged by the gradient analysis of the potential difference. At the edge of the corrosion connected region, there is usually a large gradient change in the potential difference. For example, if at the edge position of the suspected region, the potential difference changes by more than 0.2V within a short distance (such as 1 - 2 millimeters), this can also be used as one of the bases for judging that this region is a corrosion connected region.

[0054] According to the results of the potential difference data processing and analysis, set appropriate thresholds to determine whether the suspected corrosion connected region is a corrosion connected region. For example, if the average value of the potential difference exceeds the normal range and the standard deviation or gradient change meets the above corrosion determination criteria, it can be determined that this suspected region is a corrosion connected region.

[0055] At the same time, comprehensively compare and verify the potential difference data with other parameters (such as the gray level difference value, connected region area, etc.). If the potential difference is judged as a corrosion connected region and other parameters also show situations consistent with the corrosion characteristics, then the credibility of this judgment result is higher.

[0056] Furthermore, in the fully automatic intelligent grading model, the random forest algorithm is used to classify the corrosion connected regions according to the corrosion area and corrosion depth. Input the characteristic data of the suspected corrosion connected regions to be classified into the trained random forest model. Each decision tree in the model will classify and predict this sample, and then obtain the final prediction result through voting (for classification problems) or averaging (for regression problems). For example, if 70 out of 100 decision trees predict a suspected corrosion connected region as moderately corroded (grade 2) and 30 predict it as severely corroded (grade 3), then finally this region is classified as moderately corroded (grade 2).

[0057] And Figure 1 Corresponding to the method shown above, the present invention also discloses a fully automatic intelligent grading system for strip surface corrosion for Figure 1 the implementation of the method. The specific structure is as Figure 2 shown, including:

[0058] Image reconstruction module for the strip to be detected: used to continuously acquire the strip surface images being processed in the chemical reagent pool at a preset period, preprocess and reconstruct the strip surface images according to the illumination correction algorithm and image correction algorithm to obtain the strip surface images to be detected;

[0059] Gray-scale determination module: It is used to collect the surface color information of the image to be detected by using the color aggregation vector, determine the suspected corrosion connected domain in combination with the surface color information, and obtain the gray-scale difference value and gray-scale change parameter in the suspected corrosion connected domain;

[0060] Area calculation module: It is used to extract the edge pixel points of the suspected corrosion connected domain according to the gray-scale difference value and gray-scale change parameter, connect the edge pixel points to determine the edge of the suspected corrosion connected domain, and calculate the area of the suspected corrosion connected domain;

[0061] Classification module for corrosion level of grading model: Input the gray-scale difference value, gray-scale change parameter and the area of the corrosion connected domain into the full-automatic intelligent grading model, determine the corrosion depth according to the gray-scale difference value and gray-scale change parameter, and determine the corrosion level on the strip surface in combination with the corrosion depth and the area of the corrosion connected domain.

[0062] Furthermore, in the image reconstruction module to be detected, an industrial CCD camera is used to extract the strip surface image, and the industrial CCD camera is used to continuously obtain the strip surface image being processed in the chemical reagent pool at a preset period. Specifically: The image data of a predetermined area is regularly collected by the industrial CCD camera, and combined with the position and sampling timestamp of the industrial CCD camera, the monitoring data of different modalities are mapped into a unified spatio-temporal coordinate system to form a standard monitoring data set.

[0063] Finally, this embodiment discloses a computer storage medium, on which a computer program is stored. When the computer program is executed by a processor, the steps of any one of the full-automatic intelligent grading methods for strip surface corrosion are realized.

[0064] In this specification, each embodiment is described in a progressive manner. The key point of each embodiment is to illustrate the differences from other embodiments. The same or similar parts among the embodiments can be referred to each other. For the device disclosed in the embodiment, since it corresponds to the method disclosed in the embodiment, the description is relatively simple, and the relevant parts can be referred to the description of the method part.

[0065] The above description of the disclosed embodiments enables those skilled in the art to implement or use the present invention. Various modifications to these embodiments will be obvious to those skilled in the art, and the general principles defined herein can be implemented in other embodiments without departing from the spirit or scope of the present invention. Therefore, the present invention will not be limited to these embodiments shown herein, but will be accorded the widest scope consistent with the principles and novel features disclosed herein.

Claims

1. A fully automatic intelligent grading method for strip surface corrosion, characterized in that: The following steps are involved: Continuously acquiring the surface image of the steel strip being processed in the chemical agent tank according to a preset cycle, and pre-processing the surface image of the steel strip according to the illumination correction algorithm and the image correction algorithm to reconstruct the surface image of the steel strip as the image to be detected; The surface color information of the image to be detected is collected using the color aggregation vector, and the suspected corrosion connected domain is determined in combination with the surface color information, and the grayscale difference value and grayscale change parameter in the suspected corrosion connected domain are obtained; According to the grayscale difference value and grayscale change parameter, edge pixel points of the suspected corrosion connected domain are extracted, the edge pixel points are connected, the edge of the corrosion connected domain is determined, and the area of ​​the suspected corrosion connected domain is calculated; The grayscale difference value, grayscale change parameter and suspected corrosion connected domain area are input into the fully automatic intelligent grading model. The corrosion depth is determined according to the grayscale difference value and grayscale change parameter. The corrosion grade of the strip surface is determined based on the corrosion depth and corrosion connected domain area.

2. The method for fully automatic intelligent grading of strip surface corrosion according to claim 1 is characterized in that: The strip surface image is preprocessed and reconstructed according to the illumination correction algorithm and the image correction algorithm, specifically: Extracting and comparing the surface image of the steel strip before pretreatment and the surface image of the steel strip after pretreatment, determining whether overtreatment exists according to the comparison result, and compensating the overtreatment result by using the surface image of the steel strip before pretreatment to obtain a first image; Extract the spot features of the first image and input them into the reconstructed image model together with the first image, estimate the distribution of the spot features, obtain the distribution information of the spot, align the spot distribution information with the first image, and align the aligned area with the target area, and determine again whether the spot distribution information of the target area is a chemical agent projection spot or a suspected corrosion connected area; If it is a chemical agent projection spot, the projection spot is removed from the target area and the target area image is reconstructed; if it is a suspected corrosion connected domain, the image features of the target area are enhanced, the pixel presentation of the target area is strengthened, and the reconstructed target area image is obtained.

3. The method for fully automatic intelligent grading of strip surface corrosion according to claim 1 is characterized in that: An industrial-grade CCD camera is used to continuously acquire the surface image of the strip being processed in the chemical tank according to a preset cycle, specifically: The image data of the predetermined area is collected regularly by an industrial-grade CCD camera. Combined with the position and sampling timestamp of the industrial-grade CCD camera, the monitoring data of different modes are mapped to a unified spatiotemporal coordinate system to form a standard monitoring data set.

4. The method for fully automatic intelligent grading of strip surface corrosion according to claim 1 is characterized in that: The specific method for determining the suspected connected domain is: Segment the surface color information and grayscale it to obtain a grayscale image; The grayscale image is segmented by threshold to obtain a binary image, and the binary image is morphologically processed by opening and closing operations to obtain a target image. The target image is monitored for connected domains to obtain several suspected connected domains.

5. The method for fully automatic intelligent grading of strip steel surface corrosion according to claim 1, characterized in that: The method also includes using a potential difference scanning method to judge the potential difference of the suspected corrosion connected domain, and determining whether the suspected corrosion connected domain is a corrosion connected domain according to the potential difference.

6. The method for fully automatic intelligent grading of strip steel surface corrosion according to claim 1, characterized in that: In the fully automatic intelligent grading model, the random forest algorithm is used to grade the suspected corrosion connected domains according to the corrosion area and corrosion depth.

7. A fully automatic intelligent grading system for strip surface corrosion, characterized in that: include: The image reconstruction module to be detected is used to continuously obtain the surface image of the steel strip being processed in the chemical agent tank according to a preset period, and to pre-process the surface image of the steel strip according to the illumination correction algorithm and the image correction algorithm to reconstruct the surface image of the steel strip as the image to be detected; Grayscale determination module: used to collect surface color information of the image to be detected by using color aggregation vectors, determine the suspected corrosion connected domain in combination with the surface color information, and obtain the grayscale difference value and grayscale change parameter in the suspected corrosion connected domain; Area calculation module: used to extract edge pixel points of the suspected corrosion connected domain according to the grayscale difference value and grayscale change parameter, connect the edge pixel points, determine the edge of the suspected corrosion connected domain, and calculate the area of ​​the suspected corrosion connected domain; Grading model corrosion grade classification module: the grayscale difference value, grayscale change parameter and corrosion connected domain area are input into the fully automatic intelligent grading model, the corrosion depth is determined according to the grayscale difference value and grayscale change parameter, and the corrosion grade of the strip surface is determined in combination with the corrosion depth and corrosion connected domain area.

8. The fully automatic intelligent grading system for strip surface corrosion according to claim 7, characterized in that: In the image reconstruction module to be detected, an industrial-grade CCD camera is used to extract the surface image of the strip, and the industrial-grade CCD camera is used to continuously obtain the surface image of the strip being processed in the chemical agent tank according to a preset period. Specifically, the image data of the predetermined area is regularly collected by the industrial-grade CCD camera, and the monitoring data of different modes are mapped to a unified space-time coordinate system in combination with the position and sampling timestamp of the industrial-grade CCD camera to form a standard monitoring data set.

9. A computer storage medium, characterized in that The computer storage medium stores a computer program, which, when executed by a processor, implements the steps of a method for fully automatic intelligent grading of strip surface corrosion as described in any one of claims 1 to 6.

Images