A strip color difference detection system and method based on machine vision

Abstract

The present application belongs to the technical field of detection optimization, and provides a plate strip color difference detection system and method based on machine vision, which comprises the following steps: obtaining surface images of rolled hard coils and tin plates, converting through Lab images, and performing color difference defect analysis to obtain coil color difference areas and plate color difference areas, performing roughness analysis to screen out coil color difference rough areas and plate color difference rough areas, and performing defect morphology comparison to determine whether the color difference defect of the tin plate is caused by the rolled hard coil, if yes, obtaining a rolled hard coil sample between each rolling mill stand, and performing surface color difference defect and roughness analysis to determine the influencing factors of the color difference defect of the rolled hard coil, if the color difference defect is caused by the rolling mill itself, screening out the problem mill stand, detecting the distribution of the oil film to determine whether the oil film distribution is uniform, if not, adjusting the oil film contact angle to ensure that the oil film distribution is uniform, which is conducive to solving the tin plate color difference defect problem caused by uneven oil film.

Description

Technical Field

[0001] This invention belongs to the field of detection optimization technology, specifically a strip color difference detection system and method based on machine vision. Background Technology

[0002] In the current metal processing field, hard-rolled coils are the core precursor substrate for tinplate production. Defects such as surface color difference and abnormal roughness in hard-rolled coils can easily be transmitted to tinplates through subsequent rolling, resulting in gray-white stripe color difference defects on the tinplates. This seriously affects the quality of the end product and increases production costs. Therefore, it is necessary to inspect the surface quality of tinplates and hard-rolled coils, trace the root cause of tinplate color difference defects, and optimize the mill operating parameters to address the root cause problem in order to solve the tinplate color difference defect problem.

[0003] In existing technologies, there is a lack of systematic analysis of color difference defects in hard-rolled coils and tinplates. The causal relationship between color difference between tinplates and hard-rolled coils is unclear. In the process of tracing color difference defects in hard-rolled coils, the location of the causes of color difference defects is unclear. There is a lack of stand-by-stand sampling analysis and research on the synergistic changes of roughness and color difference. It is difficult to distinguish the influence of the substrate and the rolling mill. When the rolling mill causes hard-rolled coils, it is often impossible to accurately diagnose the core faults in the stand, such as uneven oil film distribution. There is no detailed analysis of uneven oil film distribution, resulting in inaccurate color difference defect processing.

[0004] Therefore, the present invention provides a strip color difference detection system and method based on machine vision. Summary of the Invention

[0005] In order to overcome the shortcomings of the prior art, at least one technical problem raised in the background art is solved.

[0006] The technical solution adopted by this invention to solve its technical problem is: a method for detecting color difference in strips based on machine vision, comprising the following steps:

[0007] Surface images of rolled coils and tin sheets are acquired, converted using Lab image processing, and color difference defect analysis is performed to obtain color difference areas of the coil and the sheet.

[0008] Roughness analysis was performed on the color difference areas of the roll and the board to screen out the rough areas of the roll and the board, and the defect morphology was compared to determine whether the color difference defect of the tin board was caused by the hard rolling.

[0009] If the color difference defect of the tinplate is caused by the hot-rolled coil, then obtain the hot-rolled coil sample between each mill stand and perform surface color difference defect and roughness analysis to determine the influencing factors of the color difference defect of the hot-rolled coil, whether it is affected by the hot-rolled coil material or by the mill itself.

[0010] If the problem is caused by the rolling mill itself, the problematic stand will be identified, and the distribution of the oil film will be tested to determine whether the oil film is even. If it is not even, the oil film contact angle will be adjusted to ensure that the oil film is evenly distributed.

[0011] Furthermore, the process involves Lab image conversion and color difference defect analysis to obtain the roll color difference region and the board color difference region, as follows:

[0012] The images of the rolled coil and the tin plate are converted into Lab images to obtain the coil Lab image and the plate Lab image, respectively. The coil Lab image and the plate Lab image are then divided into several sub-regions of equal area, which are used as the coil image sub-region and the plate image sub-region, respectively.

[0013] Obtain the overall color difference value of the roll image sub-region corresponding to the roll image sub-region, and the overall color difference value of the board image sub-region corresponding to the board image sub-region;

[0014] If the overall color difference value of a sub-region of a roll image is greater than or equal to the overall color difference threshold of the roll image, then the corresponding sub-region of the roll image is marked as a roll color difference region.

[0015] If the overall color difference value of a sub-region of the board image is greater than or equal to the overall color difference threshold of the board image, then the corresponding sub-region of the board image is marked as a board color difference region.

[0016] Furthermore, roughness analysis is performed to screen out the roll color difference roughness area and the board color difference roughness area. The process is as follows:

[0017] The roll color difference region and the board color difference region are equally divided to obtain several roll color difference sub-regions and board color difference sub-regions of equal size. The test points of the roll color difference sub-regions and board color difference sub-regions are obtained by using the grid equal division method.

[0018] The roughness of the test points in the roll color difference sub-region and the board color difference sub-region is measured. The mean values ​​of the roll color difference sub-region and the board color difference sub-region are obtained by summing and averaging. The mean values ​​of the roughness of all roll color difference sub-regions in the roll color difference region and the mean values ​​of the roughness of all board color difference sub-regions in the roll color difference region are then averaged to obtain the mean roughness of the roll color difference and the mean roughness of the board color difference.

[0019] If the roll roughness difference value is greater than or equal to the roll roughness difference threshold, then the roll color difference area is marked as the roll color difference rough area;

[0020] If the board roughness difference value is greater than or equal to the board roughness difference threshold, then the board color difference area is marked as the board color difference rough area.

[0021] Furthermore, the process of comparing defect morphologies to determine whether the color difference defect in the tinplate is caused by the hard rolling process is as follows:

[0022] Obtain the microscopic roll image and microscopic plate image corresponding to the color difference rough area of ​​the roll and the color difference rough area of ​​the plate. Perform binarization processing on the microscopic roll image and microscopic plate image respectively to obtain the microscopic roll binary image and microscopic plate binary image. Mark the corresponding rough area edges on the two-dimensional coordinate system to obtain the roll white coordinate set and the plate white coordinate set.

[0023] If the coordinate set of the rolled white area is consistent with the coordinate set of the board white area, it is initially determined that the color difference defect of the tin board is likely caused by the hard rolling process, and further judgment is made based on the texture direction characteristics.

[0024] Furthermore, the texture direction feature analysis process is as follows:

[0025] The coordinate sets of the roll white and the coordinate sets of the plate white are marked on the grayscale images of the microscopic roll and the microscopic plate, respectively, to obtain the grayscale regions of the roll and the plate to be analyzed.

[0026] Extract the corresponding texture edge contours to obtain the roll edge binary map and the plate edge binary map, and obtain the corresponding roll rough area texture principal direction angle and plate rough area texture principal direction angle;

[0027] The directional energy value is calculated at 1° intervals within the range of 0°-180°. The direction with the highest energy value is taken as the local texture direction angle of the roll and the local texture direction angle of the plate. The difference is processed with the main texture direction angle of the roll rough area and the main texture direction angle of the plate rough area respectively, and the texture standard deviation is calculated.

[0028] If the texture standard deviation is less than or equal to the texture standard deviation threshold, it indicates that the texture orientation is concentrated.

[0029] Analysis combining the principal direction angle of the rough texture area of ​​the roll with the principal direction angle of the rough texture area of ​​the plate and the standard deviation of the texture;

[0030] If the main direction angle of the texture in the rough area of ​​the coil is the same as that in the rough area of ​​the board, and both show a concentrated texture direction, then the color difference defect of the tin board is caused by the hard-rolled coil.

[0031] Furthermore, the process for detecting the distribution of the oil film is as follows:

[0032] The detection area is divided into several sub-regions of equal area, which are denoted as oil film thickness sub-regions.

[0033] The oil film thickness of each sub-region is measured, and the values ​​are summed and averaged to obtain the average oil film thickness of the detection area.

[0034] If the average oil film thickness in the detection area is between the threshold for excessively thick oil film and the threshold for excessively thin oil film, it is marked as an area where the oil film thickness is suspected to meet the standard.

[0035] Furthermore, the analysis of areas where the oil film thickness is suspected to meet the standard, to determine whether the oil film distribution is uniform, is performed as follows:

[0036] The standard deviation of the oil film thickness in the sub-regions of the detection area is obtained by calculating the standard deviation of the detection area and comparing it with the standard deviation threshold.

[0037] If the standard deviation of the detection area is greater than or equal to the standard deviation threshold, it indicates that the oil film distribution in the detection area is uneven, which is a phenomenon of uneven oil film thickness.

[0038] Furthermore, the process of adjusting the oil film contact angle is as follows:

[0039] Sub-regions that are below the oil film thinning threshold and above the oil film thickening threshold are marked and denoted as abnormal oil film thickness sub-regions;

[0040] Obtain the interface between the oil film and the roll within the abnormal oil film thickness sub-region to obtain the oil film contact angle in the oil film contact image;

[0041] If the oil film contact angle exceeds the range, the rolling speed and emulsion stability should be analyzed. By changing the rolling speed and emulsion stability, the oil film contact angle can be reduced to ensure uniform oil film distribution.

[0042] Furthermore, the analysis of changes in rolling speed and emulsion stability is as follows:

[0043] Based on the existing rolling speed, the existing rolling speed is increased, and the oil film contact angle is detected after each speed increase, and the uniformity of the oil film thickness is analyzed.

[0044] If the oil film contact angle is already within the range, but the oil film thickness is still uneven, then the stability of the emulsion should be analyzed.

[0045] Let the emulsion stand at the same temperature as during operation and observe whether stratification occurs.

[0046] If stratification occurs, it is preliminarily judged that the emulsion is aging and recorded as suspected aging;

[0047] This is confirmed by measuring the oil phase separation volume. If the proportion of the oil phase separation volume is greater than the standard emulsion oil phase separation volume threshold, it indicates that the emulsion is aging.

[0048] A machine vision-based strip color difference detection system includes the following modules:

[0049] Image acquisition and preprocessing module: acquires surface images of rolled hard coils and tin plates, performs Lab image conversion to obtain coil Lab images and plate Lab images, performs color difference defect analysis on both, and extracts color difference regions in the coil Lab images and plate Lab images respectively to obtain coil color difference regions and plate color difference regions;

[0050] Color difference defect detection and roughness analysis module: Performs roughness analysis on the color difference area of ​​the roll and the color difference area of ​​the board respectively, filters out the rough areas of the roll color difference and the rough areas of the board color difference, and compares the defect morphology to determine whether the color difference defect of the tin board is caused by the hard rolling.

[0051] Defect Influence Factor Judgment Module: If the color difference defect of the tin plate is caused by the hard-rolled coil, then obtain the hard-rolled coil sample between each mill stand, and perform surface color difference defect and roughness analysis to determine the influencing factors of the color difference defect of the hard-rolled coil, whether it is affected by the hot-rolled coil material or by the mill itself.

[0052] Cause tracing and adjustment optimization module: If the problem is caused by the rolling mill itself, the problematic stand is identified. The distribution of the oil film is tested to determine whether the oil film is uniform. If it is not uniform, the contact angle between the oil film and the roll is obtained. The oil film contact angle is adjusted to ensure that the oil film is uniformly distributed.

[0053] The beneficial effects of this invention are as follows:

[0054] (1) By acquiring surface images of the rolled coil and the tin plate, Lab image conversion is performed to obtain the comprehensive color difference value and compare it with the comprehensive color difference threshold to obtain the color difference region of the coil and the color difference region of the board. Roughness analysis is performed on the color difference region of the coil and the color difference region of the board to screen out the rough area of ​​the color difference of the coil and the rough area of ​​the color difference of the board. Geometric features and texture direction features are analyzed on the rough area of ​​the rough area of ​​the color difference of the coil and the rough area of ​​the color difference of the board. By analyzing the rough area area of ​​the rough area of ​​the rough area of ​​the coil and the rough area of ​​the color difference of the board, and whether the edge coordinate points correspond, it is preliminarily judged that the color difference defect of the tin plate is caused by the rolled coil, combined with the main direction angle of the texture and whether the texture direction is concentrated. By analyzing the color difference defect, the color difference region of the coil and the color difference region of the board are obtained, which provides data support for the subsequent analysis of the color difference correlation between the rolled coil and the tin plate and is conducive to finding the influencing factors of the color difference defect.

[0055] (2) If the color difference defect of the tinplate is caused by the hard-rolled coil, obtain the hard-rolled coil sample between each mill stand, analyze whether the color difference defect of the hard-rolled coil is affected by the hot-rolled coil material or by the mill itself. If it is affected by the mill itself, screen out the problematic mill stand and conduct oil film distribution test. If the oil film distribution is uneven, adjust the oil film contact angle, adjust the rolling speed and emulsion stability to ensure uniform oil film distribution. By analyzing the oil film contact angle, determine the appropriate rolling parameters and emulsion state, thereby optimizing the uniform distribution of oil film. By ensuring the uniformity of oil film distribution, effectively solve the problem of tinplate color difference defect caused by uneven oil film, and improve product quality and production efficiency. Attached Figure Description

[0056] The invention will now be further described with reference to the accompanying drawings.

[0057] Figure 1 This is a flowchart illustrating the steps of a machine vision-based strip color difference detection method according to an embodiment of the present invention.

[0058] Figure 2 This is a logic analysis diagram of a machine vision-based strip color difference detection method according to Embodiment 1 of the present invention;

[0059] Figure 3 This is a logic analysis diagram of a machine vision-based strip color difference detection method according to Embodiment 2 of the present invention;

[0060] Figure 4 This is a flowchart of a machine vision-based strip color difference detection system according to an embodiment of the present invention.

[0061] Figure 5 This is the color difference image of the No. 1 frame sample in the machine vision-based strip color difference detection method described in this embodiment of the invention;

[0062] Figure 6 This is the color difference image of the #2 frame sample in the machine vision-based strip color difference detection method described in this embodiment of the invention;

[0063] Figure 7 This is the color difference image of the No. 3 frame sample in the machine vision-based strip color difference detection method described in this embodiment of the invention;

[0064] Figure 8 This is the color difference image of the No. 4 frame sample in the machine vision-based strip color difference detection method described in this embodiment of the invention;

[0065] Figure 9 This is the color difference image of the No. 5 frame sample in the machine vision-based strip color difference detection method described in this embodiment of the invention. Detailed Implementation

[0066] To make the technical means, creative features, objectives and effects of this invention easier to understand, the invention will be further described below in conjunction with specific embodiments.

[0067] Example 1: Please refer to Figure 1 - Figure 2 As shown in the embodiment of the present invention, a method for detecting color difference in strips based on machine vision includes:

[0068] Step 1: Obtain surface images of the rolled coil and tin plate, perform Lab image conversion to obtain the coil Lab image and the plate Lab image, perform color difference defect analysis on both, and extract the color difference areas in the coil Lab image and the plate Lab image respectively to obtain the coil color difference area and the plate color difference area.

[0069] The surface images of the rolled coil and the tin plate are acquired using an image acquisition device. The initial images of the rolled coil and the tin plate are preprocessed to obtain the images of the rolled coil and the tin plate.

[0070] It should be noted that the image acquisition device is an industrial camera (RGB color area scan camera), combined with a low-angle ring light source (main light source) and a coaxial light source (auxiliary light source) and an industrial lens. The preprocessing includes: reflection point correction and noise filtering (using Gaussian filtering to eliminate interference caused by dust and uneven lighting).

[0071] The image of the rolled coil is converted into a Lab image to obtain a roll Lab image. The roll Lab image is then divided into several sub-regions of equal area, which are used as sub-regions of the roll image.

[0072] Using a sub-region of the roll image as the object of color difference defect analysis, the specific process is as follows:

[0073] Obtain the Lab values ​​(L, a, b) corresponding to each pixel within a sub-region of the roll image, sum them, and take the average to obtain the Lab mean of the roll image sub-region. Then, perform difference processing between the Lab mean of the roll image sub-region and the standard Lab value of the roll image to obtain the brightness difference of the roll image sub-region. Red-green difference in sub-regions of the roll image And the yellow-blue difference in sub-regions of the roll image L j Indicates brightness value, a j Indicates red-green value, b j Indicates the yellow-blue value;

[0074] The overall color difference value of the roll image is calculated based on the brightness difference, red-green difference, and yellow-blue difference of the roll image sub-regions. The calculation formula is: ;

[0075] If the color difference value of the sub-region of the roll image is... If the color difference is greater than or equal to the overall color difference threshold of the roll image, the corresponding roll image sub-region is marked as a roll color difference region;

[0076] If the color difference value of the sub-region of the roll image is... If the difference is less than the overall color difference threshold of the roll image, the corresponding roll image sub-region is marked as a non-roll color difference region.

[0077] The tin plate image is converted into a Lab image to obtain a board Lab image. The board Lab image is then divided into several sub-regions of equal area, which are used as board image sub-regions.

[0078] Using a sub-region of the plate image as the object of color difference defect analysis, the specific process is as follows:

[0079] Obtain the Lab values ​​(L, a, b) corresponding to each pixel within a sub-region of the board image, sum them, and take the average to obtain the mean Lab value of the sub-region of the board image. Then, perform difference processing between the mean Lab value of the sub-region of the board image and the standard Lab value of the board to obtain the brightness difference of the sub-region of the board image. Red-green difference in sub-regions of the image And the yellow-blue difference in the sub-region of the image board L x Indicates brightness value, a x Indicates red-green value, b x Indicates the yellow-blue value;

[0080] The overall color difference value of the board is calculated based on the brightness difference, red-green difference, and yellow-blue difference of the board image sub-regions. The calculation formula is: ;

[0081] If the overall color difference value of the sub-region of the board image is... If the color difference is greater than or equal to the overall color difference threshold of the board image, then the corresponding sub-region of the board image is marked as the board color difference region;

[0082] If the overall color difference value of the sub-region of the board image is... If the color difference is less than the overall color difference threshold of the board image, the corresponding sub-region of the board image is marked as a non-board color difference region.

[0083] It should be noted that the division method is the same when dividing the board Lab diagram and the roll Lab diagram;

[0084] The process for obtaining the standard Lab value of a roll is as follows: acquire surface images of multiple defect-free hard-rolled rolls, process them in the same way as described above to obtain the corresponding Lab images, obtain the Lab values ​​(L0, a0, b0) corresponding to pixels in each sub-region, and sum and average them to obtain the average Lab value (L0, a0, b0) for each sub-region. 0ja 0j b 0j The Lab mean (L) of the same region in surface images of multiple defect-free rolled coils was used to calculate the average value of the surface area. 0j a 0j b 0j Sum and average the values ​​to obtain the corresponding volume standard Lab value. Similarly, obtain the plate standard Lab value in the same way.

[0085] It should also be noted that the setting of the plate image comprehensive color difference threshold and the roll image comprehensive color difference threshold needs to be determined based on actual production needs, product specifications and standards, and experimental calibration data. Specifically, this includes referring to the industry color difference tolerance range, user-defined quality parameters, or setting a reasonable threshold by statistically analyzing the comprehensive color difference value ΔE distribution of defect-free samples to ensure the accuracy and applicability of the test results.

[0086] The purpose of obtaining the color difference areas of the coil and the board is to accurately trace and locate the color difference defects of the tinplate. By obtaining the color difference areas of the coil and the board, data support is provided for subsequent analysis of the color difference correlation between the rolled coil and the tinplate. In addition, by combining the specific characteristics of the coil color difference area, the cause of the color difference of the rolled coil can be further located, and it can be distinguished whether it is a problem with the raw material of the rolled coil or an improper rolling mill operating parameter. For example, during the rolling process of the rolled coil itself, if the oil film contact angle is too large, the oil film thickness will be uneven. This step provides key data support for subsequent optimization of the rolling mill operating parameters to address the root cause of the problem.

[0087] Step 2: Perform roughness analysis on the roll color difference area and the board color difference area respectively, screen out the roll color difference rough area and the board color difference rough area, and compare the defect morphology to determine whether the color difference defect of the tin board is caused by the hard roll.

[0088] The color difference regions of rolls and plates are analyzed using a roughness tester and an optical microscope. The roughness tester includes, but is not limited to, a contact roughness tester, and the optical microscope includes, but is not limited to, a metallographic microscope.

[0089] Using the color difference area of ​​the roll as the analysis object, the rough area of ​​the roll color difference is obtained. The specific process is as follows:

[0090] The roll color difference area is divided equally to obtain several roll color difference sub-regions of equal size. The test points of the roll color difference sub-regions are obtained by using the grid division method. For example, the roll color difference sub-region is divided into 4 test points by using the grid division method, which are denoted as A, B, C and D.

[0091] The roughness of the test point is measured by a contact roughness meter, and the mean value is obtained by summing and averaging the values. The mean roughness of all the sub-regions of color difference within the color difference region is summed and averaged to obtain the average roughness of the color difference region, which is denoted as the mean roughness of color difference.

[0092] The difference between the roll color difference roughness mean and the corresponding roll normal area roughness mean is processed to obtain the roll roughness difference value, and the roll roughness difference value is compared with the roll roughness difference threshold.

[0093] If the roll roughness difference value is greater than or equal to the roll roughness difference threshold, then the roll color difference area is marked as the roll color difference rough area;

[0094] If the roll roughness difference value is less than the roll roughness difference threshold, then the roll color difference area is marked as a non-roll color difference roughness area;

[0095] The color difference area of ​​the board is used as the analysis object to obtain the rough area of ​​the color difference. The specific process is as follows:

[0096] The color difference area of ​​the board is divided equally to obtain several color difference sub-regions of equal size. The test points of the color difference area are obtained by using the grid division method. For example, the color difference sub-region of the board is divided into 4 test points by using the grid division method, which are denoted as a, b, c, and d.

[0097] The roughness of the test points is measured using a contact roughness meter, and the mean value is obtained by summing and averaging the values. The mean roughness of the color difference sub-regions of the board is obtained by summing and averaging the mean roughness of all color difference sub-regions within the color difference region of the board, and is denoted as the mean roughness of the board color difference region.

[0098] The obtained average roughness value of the board color difference is processed by the difference between the average roughness value of the corresponding normal area of ​​the board to obtain the board roughness difference value, and the board roughness difference value is compared with the board roughness difference threshold.

[0099] If the board roughness difference value is greater than or equal to the board roughness difference threshold, then the board color difference area is marked as the board color difference rough area;

[0100] If the board roughness difference value is less than the board roughness difference threshold, then the board color difference area is marked as a non-board color difference roughness area;

[0101] It should be noted that the roughness difference threshold can be selected from 0.06μm to 0.09μm according to the actual situation;

[0102] For example, roughness measurements were performed on the white stripe area (color difference area) and non-white stripe area of ​​the rolled coil and the tin plate respectively, and the roughness values ​​of the corresponding non-white stripe area (normal area) of the rolled coil and the tin plate were obtained and compared, as shown in Table 1 below.

[0103] Table 1: Reflects the roughness of non-white strip areas of rolled hard coils and tin sheets;

[0104]

[0105] Table 1

[0106] As shown in Table 1, whether it is a hard-rolled coil or a tinplate, the roughness of the white stripe area is 0.06~0.09μm greater than that of the normal non-white stripe area;

[0107] It should also be noted that the division method of the roll color difference area and the board color difference area is the same. The acquisition method of the average roughness of the roll normal area and the average roughness of the board normal area is the same as that of the average roughness of the roll color difference and the average roughness of the board color difference. The only difference is that the analysis target is different. The analysis target of the average roughness of the roll normal area and the average roughness of the board normal area is the corresponding area of ​​the known qualified hard-rolled roll and tin plate.

[0108] Based on the rough areas of color difference between the coil and the board, the corresponding defect morphology is analyzed from the geometric features and texture direction features to determine whether the color difference defect of the tin board is caused by the hard-rolled coil.

[0109] Geometric feature analysis:

[0110] The rough area of ​​color difference in the roll is taken as the analysis object, and the corresponding geometric features are analyzed. The specific process is as follows:

[0111] Microscopic images of the rough areas of color difference in the roll were obtained using a metallographic microscope and preprocessed to obtain microscopic roll images;

[0112] A two-dimensional coordinate system is established with the lower left corner of the rough area of ​​the roll color difference as the origin, the length direction of the rough area of ​​the roll color difference as the x-axis, and the width direction of the rough area of ​​the roll color difference as the y-axis.

[0113] The microscopic roll image is binarized (with rough areas represented by white and non-rough areas by black) to obtain a microscopic roll binary image. The area of ​​the rough area of ​​the roll, i.e. the area of ​​the white area, is calculated. The edges of the white areas in the microscopic roll binary image are marked on a two-dimensional coordinate system to obtain the set of coordinates of the white area edges of the roll, denoted as the roll white coordinate set.

[0114] The rough area of ​​the color difference on the board is taken as the analysis object, and the corresponding geometric features are analyzed. The specific process is as follows:

[0115] Microscopic images of the color difference and roughness areas of the plate were obtained using a metallographic microscope and preprocessed to obtain microscopic plate images;

[0116] A two-dimensional coordinate system is established with the lower left corner of the rough area of ​​the color difference as the origin, the length direction of the rough area of ​​the color difference as the x-axis, and the width direction of the rough area of ​​the color difference as the y-axis.

[0117] The image of the microplate is binarized (the rough area is white and the non-rough area is black) to obtain the binary image of the microplate. The area of ​​the rough area of ​​the plate, i.e. the area of ​​the white area, is calculated. The edges of the white areas in the binary image of the microplate are marked on a two-dimensional coordinate system to obtain the set of coordinates of the edges of the white areas of the plate, denoted as the plate white coordinate set.

[0118] Compare the size of the white region within the binary images of the micro-roll and the micro-plate;

[0119] If the area of ​​the white region in the binary image of the microscopic roll and the binary image of the microscopic plate is the same, then analyze whether the coordinate set of the roll white region is the same as the coordinate set of the plate white region.

[0120] If the coordinate set of the roll white is consistent with the coordinate set of the board white, it is preliminarily determined that the color difference defect of the tin board is likely caused by the hard rolling process.

[0121] If the coordinate set of the roll white is inconsistent with the coordinate set of the board white, then the color difference defect of the tin board is determined not to be caused by the hard roll.

[0122] Texture orientation feature analysis:

[0123] Taking the rough area of ​​color difference in the roll as the analysis object, the corresponding texture direction features are analyzed. The specific process is as follows:

[0124] The microscopic roll image is converted to grayscale to obtain a grayscale image of the microscopic roll.

[0125] A two-dimensional coordinate system is established with the lower left corner of the micro-roll grayscale image as the origin, the length direction of the micro-roll grayscale image as the x-axis, and the width direction of the micro-roll grayscale image as the y-axis. The coordinates of the roll are marked on the micro-roll grayscale image to obtain the grayscale region of the roll to be analyzed.

[0126] The texture direction features of the grayscale region to be analyzed are extracted using a direction detection algorithm. The specific process is as follows:

[0127] The Canny operator is used to obtain the texture edge contour within the grayscale region of the roll to be analyzed, resulting in a roll edge binary image.

[0128] Based on the convolution edge binary map, a Hough transform is performed to map the convolution edge pixels to polar coordinate space, obtaining the convolution Hough parameter pairs (ρ). z ,θ z The resulting convolutional Hough parameter pairs (ρ) z ,θ z Sort the sequence in descending order to obtain the Convolution Hough parameter sequence;

[0129] Obtain the first five orientation angles θ from the Convolution Hough parameter sequence. za(This corresponds to the most obvious texture and can be changed based on actual measurements), where 'a' represents the sequence number in the Convolution Hough parameter sequence, denoted as θ. z1 θ z2 θ z3 θ z4 θ z5 Summing and averaging yields the principal direction angle θ of the rough texture region. z0 ;

[0130] The gray-level co-occurrence matrix (GLCM) is used to perform directional concentration analysis on the gray-level region of the volume to be analyzed. The directional energy value is calculated at 1° intervals within the range of 0°-180°, where the directional energy value represents the strength of the regular change of gray-level values ​​in each direction.

[0131] The direction with the highest energy value is taken as the local texture direction angle θ. zf and the principal direction angle θ of the rough texture area z0 Perform interpolation to obtain the roll texture deviation, take the absolute value of the roll texture deviation to obtain the texture absolute deviation, and determine whether the texture direction is concentrated by calculating the standard deviation, where the smaller the standard deviation, the more concentrated the texture is;

[0132] Taking the rough areas of the color difference on the board as the analysis object, the corresponding texture direction features are analyzed. The specific process is as follows:

[0133] The image of the microplate is converted to grayscale to obtain a grayscale image of the microplate.

[0134] A two-dimensional coordinate system is established with the lower left corner of the grayscale image of the microplate as the origin, the length direction of the grayscale image of the microplate as the x-axis, and the width direction of the grayscale image of the microplate as the y-axis. The coordinates of the plate are marked on the grayscale image of the microplate to obtain the grayscale region of the plate to be analyzed.

[0135] The texture direction features of the grayscale region of the board to be analyzed are extracted using a direction detection algorithm. The specific process is as follows:

[0136] The Canny operator is used to obtain the texture edge contour within the grayscale region of the board to be analyzed, and a binary image of the board edge is obtained.

[0137] Based on the binary image of the plate edge, a Hough transform is performed to map the plate edge pixels to polar coordinate space, obtaining the plate Hough parameter pairs (ρ). y ,θ y The resulting Platehough parameter pairs (ρ) y ,θ y Sort the parameters in descending order to obtain the Bankhov parameter sequence;

[0138] Obtain the first five orientation angles θ from the plate Hough parameter sequence. ya(This corresponds to the most obvious texture and can be changed based on actual measurements), where 'a' represents the sequence number in the Plate Hough parameter sequence, denoted as θ. y1 θ y2 θ y3 θ y4 θ y5 The summation and averaging are performed to obtain the principal direction angle θ of the rough texture region of the plate. y0 ;

[0139] The gray-level co-occurrence matrix (GLCM) is used to perform directional concentration analysis on the gray-level region of the plate to be analyzed. The directional energy value is calculated at 1° intervals within the range of 0°-180°, where the directional energy value represents the strength of the regularity of gray-level value changes in each direction.

[0140] The direction of the highest energy value is taken as the local texture direction angle θ of the plate. yf and the principal direction angle θ of the rough texture area of ​​the plate. y0 Perform interpolation to obtain board texture deviation, take the absolute value of board texture deviation to obtain texture absolute deviation, calculate the standard deviation to obtain texture standard deviation, and compare the texture standard deviation with the texture standard deviation threshold.

[0141] If the texture standard deviation is less than or equal to the texture standard deviation threshold, it indicates that the texture orientation is concentrated.

[0142] If the texture standard deviation is greater than the texture standard deviation threshold, it indicates that the texture orientation is not concentrated.

[0143] It should be noted that the specifications of the two-dimensional coordinate system mentioned above are all consistent, and the orientation detection algorithms include: Canny operator, Hough transform, gray-level co-occurrence matrix (GLCM), etc.

[0144] The standard deviation threshold for texture is set in advance by those skilled in the art based on the texture distribution characteristics;

[0145] It should also be noted that the Canny operator includes: Gaussian filtering for noise reduction, Sobel operator to calculate the gradient of each pixel, non-maximum suppression, and double threshold detection to determine the final edge, where θ is the angle between the line and the x-axis (0°≤θ≤180°) and ρ is the distance from the line to the origin.

[0146] The main texture direction angle and direction concentration are used to determine whether the defect morphology of the tin plate corresponds one-to-one with the defect morphology of the hard-rolled coil.

[0147] If the main direction angle of the rough texture area is θ z0 The angle θ between the main direction of the texture in the rough area of ​​the plate y0 If the color difference defects of the tinplate are consistent and all show a concentrated texture direction, it indicates that the color difference defects of the tinplate are caused by the hard-rolled coil.

[0148] If the main direction angle of the rough texture area is θ z0 The angle θ between the main direction of the texture in the rough area of ​​the plate y0 If the inconsistency is manifested as a lack of concentration in the texture direction, it indicates that the color difference defect of the tin plate is not caused by the hard rolling process.

[0149] Example 2: Please refer to Figure 1 - Figure 3 As shown in the embodiment of the present invention, a method for detecting color difference in strips based on machine vision includes:

[0150] Step 3: If the color difference defect of the tin plate is caused by the hot-rolled coil, obtain the hot-rolled coil sample between each mill stand and perform surface color difference defect and roughness analysis to determine the influencing factors of the color difference defect of the hot-rolled coil, whether it is affected by the hot-rolled coil material or by the mill itself.

[0151] Obtain hard-rolled coil samples between each mill stand. Using the color difference defect detection method in step one and the roughness analysis method in step two, perform color difference defect and roughness analysis on the hard-rolled coil samples to obtain the color difference defect and roughness conditions of the hard-rolled coils between each mill stand.

[0152] For example, such as Figures 5-9 The hard-rolled coil samples between each stand were obtained and color difference defects and roughness analysis were performed, as shown in Table 2.

[0153]

[0154] Table 2

[0155] Analysis of the table shows that the landscape painting defect on stand #1 has been eliminated by the rolling friction on stand #2. Stands #2-#4 do not have obvious color difference defects. Stand #5 shows obvious white stripe color difference defects. Therefore, the color difference defects are not related to the hot-rolled coil and are caused by the rolling mill itself. The cause needs to be found in stand #5.

[0156] Step 4: If the problem is caused by the mill itself, the problematic stand is identified. The distribution of the oil film is tested to determine whether the oil film is uniform. If it is not uniform, the contact angle between the oil film and the roll is obtained. The oil film contact angle is adjusted to ensure that the oil film is uniformly distributed.

[0157] Oil film distribution was tested on frame #5 to determine whether the oil film was uniform.

[0158] Obtain the contact interface between the rolling mill roll and the workpiece to get the detection area. Divide the detection area into several sub-regions of equal area, which are denoted as oil film thickness sub-regions.

[0159] The oil film thickness of each sub-region is measured using infrared spectroscopy, and the average value is obtained by summing the values. The average oil film thickness of the detection area is then compared with the oil film thickness threshold to determine whether the oil film is too thick. The oil film thickness threshold includes an excessively thick oil film threshold and an excessively thin oil film threshold.

[0160] If the average oil film thickness in the detection area is between the oil film too thick threshold and the oil film too thin threshold, it is marked as an area where the oil film thickness is suspected to meet the standard.

[0161] If the average oil film thickness in the detection area is less than or equal to the oil film thinning threshold, or if the average oil film thickness in the detection area is greater than or equal to the oil film thinning threshold, then it is marked as an area where the oil film thickness does not meet the standard.

[0162] To help understand this, the significance of obtaining areas where the oil film thickness is suspected to meet the standard is as follows:

[0163] Further oil film uniformity analysis is conducted on areas where the oil film thickness is suspected to meet the standard in order to accurately determine whether the oil film truly meets the standard and prevent interference from special conditions. For example, when the oil film thickness is tested in the detection area, there are two situations: the oil film is too thick and the oil film is too thin. Moreover, the areas occupied by the oil film that is too thick and the oil film that is too thin are similar in size. As a result, when calculating the average value, the average oil film thickness of the entire detection area is within the oil film thickness threshold, but in reality, the oil film thickness is not uniform.

[0164] It should be noted that the oil film overthickness threshold and oil film underthickness threshold are obtained by measuring an oil film sample (with the same composition as the rolling oil) with a known standard overthickness oil film thickness using infrared spectroscopy. Similarly, the oil film underthickness threshold is obtained by measuring an oil film sample with a known standard underthickness oil film thickness using infrared spectroscopy.

[0165] Based on the suspected qualified oil film thickness area, the standard deviation of the oil film thickness in the sub-region of the detection area is calculated to obtain the standard deviation of the detection area, and compared with the standard deviation threshold to determine whether the oil film distribution in the detection area is uniform.

[0166] If the standard deviation of the detection area is greater than or equal to the standard deviation threshold, it indicates that the oil film distribution in the detection area is uneven, which is a phenomenon of uneven oil film thickness.

[0167] If the standard deviation of the detection area is less than the standard deviation threshold, it indicates that the oil film distribution in the detection area is uniform, which is a phenomenon of uniform oil film thickness.

[0168] It should be noted that the standard deviation threshold setting needs to be determined based on historical experimental data and industry standards to ensure the accuracy and applicability of oil film uniformity testing and to ensure that the oil film uniformity is within an acceptable range.

[0169] If there is uneven oil film thickness, compare the oil film thickness in the sub-region with the oil film thickness threshold and the oil film thickness threshold, and mark the sub-regions that are below the oil film thickness threshold and above the oil film thickness threshold, and record them as abnormal oil film thickness sub-regions.

[0170] For example, the spatial distribution of the abnormal oil film thickness sub-region is analyzed. If the abnormal oil film thickness sub-region is located in the middle of the detection area and the oil film is too thick, the oil film contact angle is analyzed.

[0171] At the observation window of frame #5, a high-speed camera is used to capture the interface between the oil film and the roll in the sub-region of abnormal oil film thickness, and the oil film contact image is obtained through preprocessing.

[0172] The oil film contact angle in the oil film contact image is obtained by using the dynamic tangent method and compared with the oil film contact angle range threshold to determine whether the oil film in the central detection area is too thick due to the excessively large oil film contact angle.

[0173] Among them, the dynamic tangent method is to identify the edge pixels of the oil film in the oil film contact image and draw tangents along these edge pixels. The inclination angle of these tangents is the oil film contact angle.

[0174] It should be noted that the setting of the oil film contact angle range threshold needs to be determined based on historical experimental data, industry standards and actual production needs. Specifically, this includes referring to the industry standard range of oil film contact angle and setting a reasonable range value by statistically analyzing the oil film contact angle distribution of qualified samples to ensure the accuracy and applicability of oil film uniformity adjustment.

[0175] If the oil film contact angle exceeds the range, the oil film contact angle can be reduced by changing the rolling speed and the stability of the emulsion.

[0176] For example, in terms of rolling speed, the existing rolling speed is increased based on the existing rolling speed, for example, by 5% to 10% each time. The oil film contact angle is detected after each speed increase, and the uniformity of the oil film thickness is analyzed.

[0177] If the oil film contact angle is already within the range, but the oil film thickness is still uneven, then the stability of the emulsion should be analyzed.

[0178] For example, an emulsion is obtained and left to stand in an environment with the same temperature as during operation. By observing whether stratification occurs, it can be preliminarily determined whether the emulsion has aged.

[0179] If stratification occurs, it is preliminarily judged that the emulsion is aging and recorded as suspected aging;

[0180] If no stratification occurs, no further action is taken;

[0181] Based on suspected aging, confirmation was made through centrifugal stability tests. For example, 20 mL of emulsion was obtained, centrifuged at 3000 r / min for 30 minutes, and the oil phase separation volume was measured.

[0182] If the oil phase separation volume ratio is greater than the standard emulsion oil phase separation volume threshold, it indicates that the emulsifier interface film strength has decreased, oil droplets are prone to coalescence, and the emulsion is aging. The emulsion needs to be replaced to ensure uniform oil film distribution.

[0183] It should be noted that the standard emulsion oil phase separation volume threshold needs to be obtained based on the type of emulsion and multiple experimental verifications to ensure that the emulsion can stably perform its lubrication and cooling functions during its service life and avoid product quality problems caused by oil phase separation.

[0184] Working principle of the invention:

[0185] By acquiring surface images of the rolled coil and the tinplate, Lab image conversion was performed to obtain a comprehensive color difference value, which was then compared with a comprehensive color difference threshold to identify the coil color difference region and the board color difference region. Roughness analysis was then performed on these regions to identify the coil and board rough color difference areas. Geometric and texture direction feature analysis was conducted on these rough areas. By analyzing the area of ​​the rough regions and whether the edge coordinates corresponded, a preliminary judgment was made. Combined with the main texture direction angle and the concentration of texture directions, it was confirmed that the color difference defect in the tinplate was caused by the rolled coil. This color difference defect analysis identified the coil and board color difference regions, providing a basis for subsequent analysis of the color difference relationship between the rolled coil and the tinplate. The data provided by the company helps identify the influencing factors of color difference defects. If the color difference defect of the tinplate is caused by the rolling of hard coils, samples of the hard coils between each mill stand are obtained to analyze whether the color difference defect of the hard coils is affected by the hot-rolled coil material or by the mill itself. If it is affected by the mill itself, the problematic mill stand is screened out, and the oil film distribution is tested. If the oil film distribution is uneven, the oil film contact angle is adjusted, and adjustments are made in terms of rolling speed and emulsion stability to ensure uniform oil film distribution. By analyzing the oil film contact angle, appropriate rolling parameters and emulsion state are determined, thereby optimizing the uniform distribution of the oil film. By ensuring the uniformity of oil film distribution, the problem of tinplate color difference defects caused by uneven oil film distribution is effectively solved, improving product quality and production efficiency.

[0186] Example 3: Please refer to Figure 4 As shown in the embodiment of the present invention, a strip color difference detection system based on machine vision includes:

[0187] Image acquisition and preprocessing module: acquires surface images of rolled hard coils and tin plates, performs Lab image conversion to obtain coil Lab images and plate Lab images, performs color difference defect analysis on both, and extracts color difference regions in the coil Lab images and plate Lab images respectively to obtain coil color difference regions and plate color difference regions;

[0188] Color difference defect detection and roughness analysis module: Performs roughness analysis on the color difference area of ​​the roll and the color difference area of ​​the board respectively, filters out the rough areas of the roll color difference and the rough areas of the board color difference, and compares the defect morphology to determine whether the color difference defect of the tin board is caused by the hard rolling.

[0189] Defect Influence Factor Judgment Module: If the color difference defect of the tin plate is caused by the hard-rolled coil, then obtain the hard-rolled coil sample between each mill stand, and perform surface color difference defect and roughness analysis to determine the influencing factors of the color difference defect of the hard-rolled coil, whether it is affected by the hot-rolled coil material or by the mill itself.

[0190] If the cause tracing and adjustment optimization module is affected by the rolling mill itself, the problematic stand is screened out. The distribution of the oil film is detected to determine whether the oil film is uniform. If it is not uniform, the contact angle between the oil film and the roll is obtained. By adjusting the contact angle, the oil film distribution is ensured to be uniform.

[0191] The foregoing has shown and described the basic principles, main features, and advantages of the present invention. Those skilled in the art should understand that the present invention is not limited to the above embodiments. The embodiments and descriptions in the specification are merely illustrative of the principles of the invention. Various changes and modifications can be made to the invention without departing from its spirit and scope, and all such changes and modifications fall within the scope of the present invention as claimed. The scope of protection of the present invention is defined by the appended claims and their equivalents.

Claims

1. A machine vision-based method for detecting color difference in strips, characterized in that: Includes the following steps: Surface images of rolled coils and tin sheets are acquired, converted using Lab image processing, and color difference defect analysis is performed to obtain color difference areas of the coil and the sheet. Roughness analysis was performed on the color difference areas of the roll and the board to screen out the rough areas of the roll and the board, and the defect morphology was compared to determine whether the color difference defect of the tin board was caused by the hard rolling. The roll color difference region and the board color difference region are equally divided to obtain several roll color difference sub-regions and board color difference sub-regions of equal size. The test points of the roll color difference sub-regions and board color difference sub-regions are obtained by using the grid equal division method. The roughness of the test points in the roll color difference sub-region and the board color difference sub-region is measured. The mean values ​​of the roll color difference sub-region and the board color difference sub-region are obtained by summing and averaging. The mean values ​​of the roughness of all roll color difference sub-regions in the roll color difference region and the mean values ​​of the roughness of all board color difference sub-regions in the roll color difference region are then averaged to obtain the mean roughness of the roll color difference and the mean roughness of the board color difference. If the roll roughness difference value is greater than or equal to the roll roughness difference threshold, then the roll color difference area is marked as the roll color difference rough area; If the board roughness difference value is greater than or equal to the board roughness difference threshold, then the board color difference area is marked as the board color difference rough area; Obtain the microscopic roll image and microscopic plate image corresponding to the color difference rough area of ​​the roll and the color difference rough area of ​​the plate. Perform binarization processing on the microscopic roll image and microscopic plate image respectively to obtain the microscopic roll binary image and microscopic plate binary image. Mark the corresponding rough area edges on the two-dimensional coordinate system to obtain the roll white coordinate set and the plate white coordinate set. If the coordinate set of the rolled white area is consistent with the coordinate set of the board white area, it is initially determined that the color difference defect of the tin board is likely caused by the hard rolling process, and further judgment is made based on the texture direction characteristics. The coordinate sets of the roll white and the coordinate sets of the plate white are marked on the grayscale images of the microscopic roll and the microscopic plate, respectively, to obtain the grayscale regions of the roll and the plate to be analyzed. Extract the corresponding texture edge contours to obtain the roll edge binary map and the plate edge binary map, and obtain the corresponding roll rough area texture principal direction angle and plate rough area texture principal direction angle; The directional energy value is calculated at 1° intervals within the range of 0°-180°. The direction with the highest energy value is taken as the local texture direction angle of the roll and the local texture direction angle of the plate. The difference is processed with the main texture direction angle of the roll rough area and the main texture direction angle of the plate rough area respectively, and the texture standard deviation is calculated. If the texture standard deviation is less than or equal to the texture standard deviation threshold, it indicates that the texture orientation is concentrated. Analysis combining the principal direction angle of the rough texture area of ​​the roll with the principal direction angle of the rough texture area of ​​the plate and the standard deviation of the texture; If the main direction angle of the texture in the rough area of ​​the coil is the same as that in the rough area of ​​the board and both show a concentrated texture direction, then the color difference defect of the tin board is caused by the hard-rolled coil. If the color difference defect of the tinplate is caused by the hot-rolled coil, then obtain the hot-rolled coil sample between each mill stand and perform surface color difference defect and roughness analysis to determine the influencing factors of the color difference defect of the hot-rolled coil, whether it is affected by the hot-rolled coil material or by the mill itself. If the problem is caused by the rolling mill itself, the problematic stand will be identified, and the distribution of the oil film will be tested to determine whether the oil film is even. If it is not even, the oil film contact angle will be adjusted to ensure that the oil film is evenly distributed.

2. The method for detecting color difference in strips based on machine vision according to claim 1, characterized in that: The process of converting images using LabVIEW and performing color difference defect analysis to obtain the roll color difference area and the board color difference area is as follows: The images of the rolled coil and the tin plate are converted into Lab images to obtain the coil Lab image and the plate Lab image, respectively. The coil Lab image and the plate Lab image are then divided into several sub-regions of equal area, which are used as the coil image sub-region and the plate image sub-region, respectively. Obtain the overall color difference value of the roll image sub-region corresponding to the roll image sub-region, and the overall color difference value of the board image sub-region corresponding to the board image sub-region; If the overall color difference value of a sub-region of a roll image is greater than or equal to the overall color difference threshold of the roll image, then the corresponding sub-region of the roll image is marked as a roll color difference region. If the overall color difference value of a sub-region of the board image is greater than or equal to the overall color difference threshold of the board image, then the corresponding sub-region of the board image is marked as a board color difference region.

3. The method for detecting color difference in strips based on machine vision according to claim 1, characterized in that: The distribution of the oil film was tested, and the process is as follows: The detection area is divided into several sub-regions of equal area, which are denoted as oil film thickness sub-regions. The oil film thickness of each sub-region is measured, and the values ​​are summed and averaged to obtain the average oil film thickness of the detection area. If the average oil film thickness in the detection area is between the threshold for excessively thick oil film and the threshold for excessively thin oil film, it is marked as an area where the oil film thickness is suspected to meet the standard.

4. The method for detecting color difference in strips based on machine vision according to claim 3, characterized in that: The following steps are taken to analyze areas where the oil film thickness is suspected to meet the standard to determine whether the oil film distribution is uniform: The standard deviation of the oil film thickness in the sub-regions of the detection area is obtained by calculating the standard deviation of the detection area and comparing it with the standard deviation threshold. If the standard deviation of the detection area is greater than or equal to the standard deviation threshold, it indicates that the oil film distribution in the detection area is uneven, which is a phenomenon of uneven oil film thickness.

5. The method for detecting color difference in strips based on machine vision according to claim 1, characterized in that: The process for adjusting the oil film contact angle is as follows: Sub-regions that are below the oil film thinning threshold and above the oil film thickening threshold are marked and denoted as abnormal oil film thickness sub-regions; Obtain the interface between the oil film and the roll within the abnormal oil film thickness sub-region to obtain the oil film contact angle in the oil film contact image; If the oil film contact angle exceeds the range, the rolling speed and emulsion stability should be analyzed. By changing the rolling speed and emulsion stability, the oil film contact angle can be reduced to ensure uniform oil film distribution.

6. The method for detecting color difference in strips based on machine vision according to claim 5, characterized in that: The analysis of changes in rolling speed and emulsion stability is as follows: Based on the existing rolling speed, the existing rolling speed is increased, and the oil film contact angle is detected after each speed increase, and the uniformity of the oil film thickness is analyzed. If the oil film contact angle is already within the range, but the oil film thickness is still uneven, then the stability of the emulsion should be analyzed. Let the emulsion stand at the same temperature as during operation and observe whether stratification occurs. If stratification occurs, it is preliminarily judged that the emulsion is aging and recorded as suspected aging; This is confirmed by measuring the oil phase separation volume. If the proportion of the oil phase separation volume is greater than the standard emulsion oil phase separation volume threshold, it indicates that the emulsion is aging.

7. A machine vision-based strip color difference detection system, characterized in that: Includes the following modules: Image acquisition and preprocessing module: acquires surface images of rolled hard coils and tin plates, performs Lab image conversion to obtain coil Lab images and plate Lab images, performs color difference defect analysis on both, and extracts color difference regions in the coil Lab images and plate Lab images respectively to obtain coil color difference regions and plate color difference regions; Color difference defect detection and roughness analysis module: Performs roughness analysis on the color difference area of ​​the roll and the color difference area of ​​the board respectively, filters out the rough areas of the roll color difference and the rough areas of the board color difference, and compares the defect morphology to determine whether the color difference defect of the tin board is caused by the hard rolling. The roll color difference region and the board color difference region are equally divided to obtain several roll color difference sub-regions and board color difference sub-regions of equal size. The test points of the roll color difference sub-regions and board color difference sub-regions are obtained by using the grid equal division method. The roughness of the test points in the roll color difference sub-region and the board color difference sub-region is measured. The mean values ​​of the roll color difference sub-region and the board color difference sub-region are obtained by summing and averaging. The mean values ​​of the roughness of all roll color difference sub-regions in the roll color difference region and the mean values ​​of the roughness of all board color difference sub-regions in the roll color difference region are then averaged to obtain the mean roughness of the roll color difference and the mean roughness of the board color difference. If the roll roughness difference value is greater than or equal to the roll roughness difference threshold, then the roll color difference area is marked as the roll color difference rough area; If the board roughness difference value is greater than or equal to the board roughness difference threshold, then the board color difference area is marked as the board color difference rough area; Obtain the microscopic roll image and microscopic plate image corresponding to the color difference rough area of ​​the roll and the color difference rough area of ​​the plate. Perform binarization processing on the microscopic roll image and microscopic plate image respectively to obtain the microscopic roll binary image and microscopic plate binary image. Mark the corresponding rough area edges on the two-dimensional coordinate system to obtain the roll white coordinate set and the plate white coordinate set. If the coordinate set of the rolled white area is consistent with the coordinate set of the board white area, it is initially determined that the color difference defect of the tin board is likely caused by the hard rolling process, and further judgment is made based on the texture direction characteristics. The coordinate sets of the roll white and the coordinate sets of the plate white are marked on the grayscale images of the microscopic roll and the microscopic plate, respectively, to obtain the grayscale regions of the roll and the plate to be analyzed. Extract the corresponding texture edge contours to obtain the roll edge binary map and the plate edge binary map, and obtain the corresponding roll rough area texture principal direction angle and plate rough area texture principal direction angle; The directional energy value is calculated at 1° intervals within the range of 0°-180°. The direction with the highest energy value is taken as the local texture direction angle of the roll and the local texture direction angle of the plate. The difference is processed with the main texture direction angle of the roll rough area and the main texture direction angle of the plate rough area respectively, and the texture standard deviation is calculated. If the texture standard deviation is less than or equal to the texture standard deviation threshold, it indicates that the texture orientation is concentrated. Analysis combining the principal direction angle of the rough texture area of ​​the roll with the principal direction angle of the rough texture area of ​​the plate and the standard deviation of the texture; If the main direction angle of the texture in the rough area of ​​the coil is the same as that in the rough area of ​​the board and both show a concentrated texture direction, then the color difference defect of the tin board is caused by the hard-rolled coil. Defect Influence Factor Judgment Module: If the color difference defect of the tin plate is caused by the hard-rolled coil, then obtain the hard-rolled coil sample between each mill stand, and perform surface color difference defect and roughness analysis to determine the influencing factors of the color difference defect of the hard-rolled coil, whether it is affected by the hot-rolled coil material or by the mill itself. Cause tracing and adjustment optimization module: If the problem is caused by the rolling mill itself, the problematic stand is identified. The distribution of the oil film is tested to determine whether the oil film is uniform. If it is not uniform, the contact angle between the oil film and the roll is obtained. The oil film contact angle is adjusted to ensure that the oil film is uniformly distributed.

Images