A steel stamp online detection system and method based on color CCD

Abstract

The invention discloses a steel stamp online detection system and method based on color CCD, comprising an image acquisition module, a color correction module, a color space conversion module, a multiple search positioning module, a print separation module, a full-width detection module, a color difference detection module, and a BLOB analysis output module; the invention improves the accuracy of the steel stamp detection system through multiple positioning and color detection algorithms, and can be applied to steel stamps with complex patterns, textures, characters and colors; secondly, it can monitor the production status of cigarette paper in real time during the cigarette paper printing process; thirdly, it solves the problem of the need for manual detection in traditional detection methods, and greatly reduces labor costs.

Description

Technical Field

[0001] The invention relates to the technical field of cigarette paper online detection systems, and in particular to a steel print online detection system and method based on a color CCD. Background Art

[0002] Cigarette paper is an important part of the production process of cigarettes. Cigarette paper fixes the tobacco into shape and then connects the mouthpiece to form a cigarette. It controls the burning speed during the burning process of the cigarette and also affects the appearance and taste of the cigarette. Domestic and foreign cigarette manufacturers print unique marks composed of words, patterns, etc. on their cigarettes, which are called brand steel stamps. At the same time, some cigarettes are also printed with codes composed of letters and numbers, which are called coding steel stamps. The cigarette steel stamp used to indicate the brand and trademark of the cigarette is printed on the cigarette, which to some extent reflects the brand characteristics of the cigarette, which is equivalent to the nameplate of the product. Therefore, the quality of the cigarette steel stamp indicates the quality and quality of the brand cigarette to a certain extent. If the cigarette steel stamp has quality problems and flows into the market, the reputation of the brand cigarette will also be greatly affected.

[0003] At present, in cigarette making equipment, there are printing errors caused by blurred steel stamp patterns, misaligned positions, missing patterns, etc. on the cigarette rolling machine, as well as quality problems such as wrinkles, cracks, stains, etc. in the appearance characteristics of the paper itself. Traditional detection methods rely on human eye detection, but human eyes are greatly affected by subjective factors when checking printed defects, and have low efficiency and slow feedback, which cannot well meet the needs of large-scale production of modern printing companies. Therefore, using machine vision systems to detect printed products in real time, using cameras instead of human eyes, and using computer hardware and image processing software instead of human brains for image analysis and processing, defect identification, and information feedback can greatly improve the detection efficiency and quality of printing quality. In addition, machine vision systems are also the mainstream direction of quality control in printing companies.

[0004] CN201320700922.4 uses two black and white CCDs to detect the brand steel stamp and the code steel stamp respectively, and performs defect analysis on the grayscale image, which increases the complexity of the system. CN200910044366.8 acquires the grayscale information of the steel stamp through black and white CCD, and compares and analyzes it with the standard grayscale steel stamp to determine whether the steel stamp is qualified. The above patents all use black and white CCDs for image acquisition and defect analysis on grayscale images, and cannot analyze and process the color information of the steel stamp. Summary of the invention

[0005] The invention provides a steel stamp online detection system and method based on color CCD, so as to solve the problem that the prior art cannot analyze and process the color information of the steel stamp.

[0006] The technical solution adopted by the present invention is: a steel stamp online detection system based on color CCD, including an image acquisition module, a color correction module, a color space conversion module, a multiple search and positioning module, a print separation module, a full-width detection module, a color difference detection module, and a BLOB analysis output module;

[0007] The image acquisition module includes a line scan color CCD, a high brightness white LED light source and an image acquisition card; image data is captured by the line scan color CCD, and the acquisition card transmits the image data to the color correction module for analysis and processing through a callback function;

[0008] The color correction module performs color correction processing on the collected image to enhance the grayscale contrast;

[0009] The color space conversion module converts the image corrected by the color correction module into color;

[0010] The multi-search positioning module realizes positioning from coarse to fine;

[0011] The print order separation module separates different print orders and performs detection on different print orders respectively;

[0012] The full-width detection module detects the position of the brand stamp image and the entire stamp code to determine whether the plate is off, thereby achieving comprehensive quality detection;

[0013] The color difference detection module compares the color difference of the collected steel stamp image with the standard template image to achieve steel stamp defect classification;

[0014] The BLOB analysis output module extracts features and classifies the residual image to determine whether it is a defect error or a noise point.

[0015] Preferably, the color correction module adopts histogram normalization.

[0016] Preferably, the color space conversion module adopts CIELab color space conversion.

[0017] Preferably, the print separation module divides different detection areas according to different regional positioning cores in accordance with the printing order, uses respective print models for detection, and sets a threshold for whether or not the steel print is present to represent the correlation coefficient between the detection area image and the positioning core image. When the contents of the detection area image and the positioning core image are similar or consistent, the greater the correlation coefficient.

[0018] Preferably, the full-format detection module includes an adaptive template matching module and an OCR module. The adaptive template matching module establishes a color model according to different printing orders, divides it into small areas based on multiple positioning, automatically counts the comprehensive features of the image in the small area, and divides the different printing areas of the brand name steel stamp image through an adaptive template threshold, and compares and analyzes it with the template image of each print; the OCR module performs detection according to the expected steel stamp code, detects each coded character separately and calculates the most serious defect type according to priority, and at the same time performs position detection on the entire steel stamp code to determine whether the plate is off, thereby achieving comprehensive quality detection.

[0019] Preferably, the color difference detection module compares the collected steel stamp image with the standard template image by color difference, and uses single-layer and comprehensive color difference analysis methods to classify steel stamp defects according to preset color difference parameters;

[0020] First, the color difference between the collected steel stamp image and the standard template image is compared, and the Lab space of the standard template image is transformed into L1 * 、a1 * 、b1 * , collect the stamp image and transform it into L2 space * 、a2 * 、b2 * ;but:

[0021] ΔL=L1 * -L2 *

[0022] Δa=a2 * -a1 *

[0023] Δb=b2 * -b1 *

[0024] Define the total color difference as: ΔC 2 =(ΔL * ) 2 +(Δa * ) 2 +(Δb * ) 2

[0025] In the CIELab color space, the a* and b* components reflect the chromaticity information, while the L component reflects the brightness information. Therefore, the comprehensive color difference calculation formula is defined as: ΔC 2 =(Δa * ) 2 +(Δb * ) 2 .

[0026] Preferably, the BLOB analysis output module performs feature extraction and classification on the residual image, that is, classifies the residual points that exceed the expected value of the template, and determines whether it is a defect error or a noise point through the residual point threshold and graphic features; at the same time, performs graphic feature analysis on the residual Blob image, quickly converts the simple pattern information into the shape information of the pattern, thereby distinguishing the real defects from the false defects according to the different graphic features.

[0027] A method for online detection of steel stamps based on color CCD, characterized by comprising the following steps:

[0028] S101: collecting images of two cigarettes on cigarette paper;

[0029] S102: color correction and color space transformation of the double cigarette images;

[0030] S103: locating the edge of the cigarette paper and locating different patterns in the paper;

[0031] S104: Calculate the similarity between the positioning area of ​​print 1 and the template;

[0032] S105: Determine whether the similarity is less than the threshold value for the presence or absence of a steel stamp; if the similarity is greater than the threshold value for the presence or absence of a steel stamp, proceed to step S106; if the similarity is less than the threshold value for the presence or absence of a steel stamp, proceed to step S107;

[0033] S106: Separation of different printing areas;

[0034] Use the template threshold to segment the printing 1 and printing 2 areas, compare with the steel stamp template, and calculate the brand steel stamp residual points;

[0035] Color difference detection, calculation of brand stamp residual points;

[0036] Calculate the residual points of the coding stamp through OCR;

[0037] Then proceed to step S108;

[0038] S107: Calculate the similarity between the positioning area of ​​print 2 and the template;

[0039] And judge whether the similarity of print 2 is greater than that of print 1; if the similarity of print 2 is greater than that of print 1, report a joint defect; if the similarity of print 2 is less than that of print 1, report a defect;

[0040] S108: Determine whether the residual point ratio is greater than the threshold value; if the residual point ratio is greater than the threshold value, report a defect; if the residual point ratio is less than the threshold value, it is a normal steel stamp.

[0041] Preferably, in S102, the image color correction is performed by histogram normalization to enhance the contrast of grayscale values, and the specific steps include:

[0042] S1021: performing grayscale equalization on the histogram of the collected image;

[0043] S1022: Select a histogram and perform a histogram equalization transformation;

[0044] S1023: Invert the transformation obtained in step S1021, and map the original histogram to the selected histogram.

[0045] Preferably, in S102, the color space transformation adopts CIELab color space transformation.

[0046] The beneficial effects of the present invention are: compared with the prior art, firstly, the present invention improves the accuracy of the steel stamp detection system through multiple positioning and color detection algorithms, and can be applied to steel stamps with complex patterns, textures, characters and colors; secondly, it can monitor the production status of cigarette paper in real time during the cigarette paper printing process; thirdly, it solves the problem of manual detection required in traditional detection methods, greatly reducing labor costs. BRIEF DESCRIPTION OF THE DRAWINGS

[0047] Figure 1 This is a structural diagram of a steel stamp online detection system based on color CCD disclosed by the present invention;

[0048] Figure 2 It is a flow chart of the algorithm of the color difference detection module in the steel stamp online detection system based on color CCD disclosed by the present invention;

[0049] Figure 3 The present invention discloses a flow chart of a method for online detection of steel stamps based on color CCD. DETAILED DESCRIPTION

[0050] In order to make the purpose, technical solutions and advantages of the present invention more clear, the present invention will be further described in detail below with reference to the accompanying drawings, but the embodiments of the present invention are not limited thereto.

[0051] Embodiment 1:

[0052] See also Figure 1-2 A steel stamp online detection system based on color CCD includes an image acquisition module 1, a color correction module 2, a color space transformation module 3, a multiple search and positioning module 4, a print separation module 5, a full-width detection module 6, a color difference detection module 7, and a BLOB analysis output module 8.

[0053] The image acquisition module 1 includes a line scanning color CCD, a high brightness white LED light source and an image acquisition card. The line scanning color CCD is used to acquire images. The image acquisition card is used to upload the acquired images and transmit them to the color correction module 2 for analysis and processing through a callback function.

[0054] During the imaging process of the system, different machines and different cameras lead to different light environments. Each machine needs a separate template to adapt to different light environments. For this reason, the color correction module 2 adopts histogram regulation.

[0055] Due to the complexity of the color of the stamped pattern, especially the rich colors of the stamped pattern, the small color difference in the mixed color area, the light color of the pattern, etc., the color space transformation module 3 adopts the CIELab standard colorimetry system. In the CIELab color space, the Euclidean distance between two colors is closer to the color difference perceived by people, and it is easier to detect fine color differences, instead of producing uneven deviations like the RGB color model.

[0056] The multi-search positioning module 4 is to ensure the full-width detection accuracy, allowing a certain degree of expansion and contraction deformation of cigarette paper and a certain degree of overprint tolerance in different printing processes, so it is necessary to perform fine multi-level search and positioning on the image area. First, coarse positioning is performed through the large paper edge positioning area, and then precise positioning of different printing layers is performed through the regional positioning core. Therefore, the multi-search positioning module 4 can achieve coarse to fine positioning of different printing layers.

[0057] Different steel stamps print different patterns on paper, such as Chinese characters are a pattern printed by a steel stamp, English letters are a pattern printed by a steel stamp, and digital codes are a pattern printed by a steel stamp. Since the spacing between Chinese characters, English letters, and digital codes on paper is not fixed, different printing times need to be separated for detection. In this embodiment, Chinese characters are printed at printing time 1 and English letters are printed at printing time 2.

[0058] The print separation module 5 is divided into different detection areas according to the printing order based on the different regional positioning cores, and uses their respective print models for detection. The threshold value of the steel stamp is set to represent the correlation coefficient between the detection area image and the positioning core image. When the content of the detection area image and the positioning core image is similar or consistent, the greater the correlation coefficient.

[0059] First, calculate the similarity between the positioning area of ​​print 1 and the template, and then determine whether the similarity is less than the threshold for steel stamping. If the similarity is greater than the threshold for steel stamping, it means it is normal; if the similarity is less than the threshold for steel stamping, calculate the similarity between the positioning area of ​​print 2 and the template. If the similarity of print 2 is greater than that of print 1, a joint defect is reported; if the similarity of print 2 is less than that of print 1, a defect is reported; the so-called joint is the joint of different rolls of paper. When the detection system determines that the similarities of print 1 and print 2 are both poor, it means it is a joint defect. The threshold for steel stamping indicates the correlation coefficient between the detection area image and the positioning core image.

[0060] The full-format detection module 6 includes an adaptive template matching module and an OCR module. The adaptive template matching module establishes a color model according to different printing orders, divides it into small areas based on multiple positioning, and automatically counts the comprehensive features of the image in the small area, such as average brightness, etc. The different printing areas of the brand steel stamp image are segmented by an adaptive template threshold, and compared and analyzed with the template image of each printing; the OCR module detects according to the expected steel stamp code, and each coded character is detected separately and the most serious defect type is calculated according to priority, including ink dots, light flowers, color cross-linking, missing prints, wrong numbers, etc. At the same time, the entire steel stamp code is detected to determine whether the plate is off, so as to achieve comprehensive quality detection. Since each coded character is different, it cannot be directly compared with the template like printing 1 and printing 2, so the steel stamp code is detected here through the OCR module.

[0061] The color difference detection module 7 compares the color difference of the collected steel stamp image with the standard template image, and uses single-layer and comprehensive color difference analysis methods to classify steel stamp defects according to the preset color difference parameters. The a* and b* components of the L*a*b* color space reflect the chromaticity information, while the L component reflects the brightness information. The color detection process is as follows Figure 2 shown.

[0062] First, the color difference between the collected steel stamp image and the standard template image is compared. For example, the Lab space of the standard template image is transformed into L1 * 、a1 * 、b1 * , collect the stamp image and transform it into L2 space * 、a2 * 、b2 * .but:

[0063] ΔL=L1 * -L2 *

[0064] Δa=a2 * -a1 *

[0065] Δb=b2 * -b1 *

[0066] Define the total color difference as: ΔC 2 =(ΔL * ) 2 +(Δa * ) 2 +(Δb * ) 2

[0067] In the CIELab color space, the a* and b* components reflect the chromaticity information, while the L component reflects the brightness information. Therefore, the comprehensive color difference calculation formula is defined as: ΔC 2 =(Δa * ) 2 +(Δb * ) 2

[0068] Comprehensive consideration of ΔL, Δa, Δb, ΔC 2 , single-layer and comprehensive color difference analysis methods are used to realize steel stamp color difference detection.

[0069] The BLOB analysis output module extracts and classifies features of the residual image, that is, it classifies the residual points that exceed the expected value of the template, and determines whether it is a defect error or a noise point through the residual point threshold and graphic features. Specifically, the BLOB analysis output module sets a threshold value, and when the residual area exceeds the threshold value, the steel stamp defect information is reported. The threshold value means that if the residual point exceeds the set value, it is considered a defect; at the same time, by performing graphic feature analysis on the residual Blob image, the simple pattern information can be quickly converted into the shape information of the pattern, such as the image centroid, image area, image perimeter, image circumscribed minimum rectangle and other graphic information, so as to distinguish the real defects from the false defects according to the different graphic features.

[0070] Embodiment 2:

[0071] See also Figure 1-3 , a steel stamp online detection method based on color CCD, the detection method comprises the following steps:

[0072] Step S101, collecting a double-cigarette image of cigarette paper; specifically, establishing a standard image model, selecting a double-cigarette area in the collected image, dividing it into small areas, setting the level of each printing area, and applying it to the full format to obtain a template image and parameters;

[0073] Step 102, color correction and LAB space transformation of the double cigarette images; specifically, color correction processing is performed on the collected images to selectively enhance the contrast within a certain gray value range, the main idea of ​​which is histogram regulation; the color correction image is converted into Lab space to facilitate quantitative analysis of color difference;

[0074] The specific steps of histogram specification are as follows (here M and N are the number of gray levels in the original image and the specified image, respectively, and only the case of N ≤ M is considered):

[0075] Step 1021, grayscale equalization is performed on the histogram of the collected image;

[0076] Step 1022, specifying a required histogram and calculating a transformation that can equalize the specified histogram;

[0077] Step 1023, invert the transformation obtained in step 1021, and map the original histogram to the specified histogram.

[0078] Step 103, locating the cigarette paper edge and locating different patterns in the paper.

[0079] Step 104, calculate the similarity between the positioning area of ​​print 1 and the template; specifically, divide the image into different areas according to the printing order for processing, and each detection area is further finely positioned on the basis of paper edge positioning; based on the paper edge positioning and the positioning of each area, each print area is detected using the model of its own print.

[0080] Step 105, determine whether the similarity is less than the threshold for the presence or absence of a steel stamp; if the similarity is greater than the threshold for the presence or absence of a steel stamp, proceed to step 106. If the similarity is less than the threshold for the presence or absence of a steel stamp, proceed to step 107. Specifically, the threshold for the presence or absence of a steel stamp is set to represent the correlation coefficient between the detection area image and the positioning core image; the position of the steel stamp pattern in the print 1 area is located based on the paper edge positioning; the correlation coefficient between the detection area image and the positioning core steel stamp image is calculated, and compared with the threshold for the presence or absence of a steel stamp; the so-called positioning core is to mark an area on the template image, and search for the area with the largest correlation coefficient with this area on the target image. The image of the area marked on the template is the positioning core image;

[0081] Step 106, separating different printing areas.

[0082] Use the template threshold to segment the print 1 and 2 areas, compare with the steel stamp template, and calculate the brand steel stamp residual points; specifically, the template threshold is the threshold of the established standard image in the RGB space; it is divided into small areas based on multiple positioning, and the comprehensive features of the image in the small area, such as average brightness, are automatically counted, and the template threshold is adaptively adjusted according to the template image to perform template matching.

[0083] Color difference detection calculates the residual points of brand steel stamp; specifically, color difference detection is to compare the color difference of the collected steel stamp image with the standard template image, and according to the preset color difference parameters, single-layer and comprehensive color difference analysis methods are used to classify steel stamp defects.

[0084] The residual dots of the coded steel stamp are calculated through OCR. Specifically, OCR is detected according to the expected steel stamp code. Each coded character is detected separately and the most serious defect type is calculated according to priority, including ink dots, light flowers, color mixing, missing prints, wrong numbers, etc. At the same time, the position of the entire steel stamp code is detected to determine whether it is off the plate, so as to achieve comprehensive quality detection.

[0085] Step 107, calculate the similarity between the print 2 positioning area and the template; specifically, locate the position of the stamp pattern in the print 2 area based on the paper edge positioning; calculate the correlation coefficient between the detection area image and the positioning core stamp image, and compare it with the threshold of whether there is a stamp.

[0086] And determine whether the similarity of Print 2 is greater than that of Print 1; if the similarity of Print 2 is greater than that of Print 1, a joint defect is reported; if the similarity of Print 2 is less than that of Print 1, a defect is reported; the joint here refers to the joint between each roll of paper.

[0087] Step 108, determine whether the residual point ratio is greater than the threshold value; specifically, perform feature extraction and classification on the residual image, that is, classify the residual points that exceed the expected value of the template, and determine whether it is a defect error or a noise point through the residual point threshold value and graphic features. The threshold value means that the residual points exceeding the specified value are considered to be defects.

[0088] Compared with the existing technology, firstly, the present embodiment provides an online steel stamp detection method based on color CCD, which improves the accuracy of the steel stamp detection system through multiple positioning and color detection algorithms, and can be applied to steel stamps with complex patterns, textures, characters and colors; secondly, it can monitor the production status of cigarette paper in real time during the cigarette paper printing process; thirdly, it solves the problem of manual detection required in traditional detection methods, greatly reducing labor costs.

[0089] The above embodiments are only used to illustrate the technical solutions of the present invention, rather than to limit it. Although the invention has been described in detail with reference to the aforementioned embodiments, those skilled in the art should understand that the technical solutions described in the aforementioned embodiments may still be modified, or some of the technical features thereof may be replaced by equivalents. However, these modifications or replacements do not deviate the essence of the corresponding technical solutions from the spirit and scope of the technical solutions of the embodiments of the present invention.

Claims

1. A steel stamp online detection system based on color CCD, characterized by: It includes image acquisition module, color correction module, color space transformation module, multiple search and positioning module, print separation module, full-width detection module, color difference detection module, and BLOB analysis and output module; The image acquisition module includes a line scan color CCD, a high brightness white LED light source and an image acquisition card; image data is captured by the line scan color CCD, and the acquisition card transmits the image data to the color correction module for analysis and processing through a callback function; The color correction module performs color correction processing on the collected image to enhance the grayscale contrast; The color space conversion module converts the image corrected by the color correction module into color; The multi-search positioning module realizes positioning from coarse to fine; The print order separation module separates different print orders and performs detection on different print orders respectively; The full-format detection module performs position detection on the image of the brand steel stamp and the entire steel stamp code to determine whether there is a misalignment, thereby achieving comprehensive quality detection; the full-format detection module includes an adaptive template matching module and an OCR module. The adaptive template matching module establishes a color model according to different printing orders, divides it into small areas based on multiple positioning, automatically counts the comprehensive features of the image in the small area, and divides the different printing areas of the brand steel stamp image through an adaptive template threshold, and compares and analyzes it with the template image of each printing; the OCR module performs detection according to the expected steel stamp code, detects each coded character separately and calculates the most serious defect type according to priority, and at the same time performs position detection on the entire steel stamp code to determine whether there is a misalignment, thereby achieving comprehensive quality detection; The color difference detection module compares the color difference of the collected steel stamp image with the standard template image to realize the classification of steel stamp defects; the color difference detection module compares the color difference of the collected steel stamp image with the standard template image, and uses single-layer and comprehensive color difference analysis methods to realize the classification of steel stamp defects according to preset color difference parameters; First, the color difference between the collected steel stamp image and the standard template image is compared, and the Lab space of the standard template image is transformed into L1 * 、a1 * 、b1 * , collect the stamp image and transform it into L2 space * 、a2 * 、b2 * ;but: ΔL=L1 * -L2 * <h2 style=";text-align:left;direction:ltr">Δa = a2<h2 style=";text-align:left;direction:ltr"> * <h2 style=";text-align:left;direction:ltr"> -a1<h2 style=";text-align:left;direction:ltr"> * Δb=b2 * -b1 * Define the total color difference as: ΔC 2 =(ΔL * ) 2 +(Δa * ) 2 +(Δb * ) 2 In the CIELab color space, the a* and b* components reflect the chromaticity information, while the L component reflects the brightness information. Therefore, the comprehensive color difference calculation formula is defined as: ΔC 2 =(Δa * ) 2 +(Δb * ) 2 ; The BLOB analysis output module extracts features and classifies the residual image to determine whether it is a defect error or a noise point.

2. The color CCD-based steel stamp online detection system according to claim 1, characterized in that: The color correction module uses histogram normalization.

3. The color CCD-based steel stamp online detection system according to claim 1, characterized in that: The color space conversion module adopts CIELab color space conversion.

4. The color CCD-based steel stamp online detection system according to claim 1, characterized in that: The print separation module divides different detection areas according to different regional positioning cores in accordance with the printing order, uses respective print models for detection, and sets a threshold value for the presence or absence of steel stamps to represent the correlation coefficient between the detection area image and the positioning core image. When the contents of the detection area image and the positioning core image are similar or consistent, the greater the correlation coefficient.

5. The color CCD-based steel stamp online detection system according to claim 1, characterized in that: The BLOB analysis output module performs feature extraction and classification on the residual image, that is, classifies the residual points that exceed the expected value of the template, and determines whether they are defect errors or noise points through the residual point threshold and graphic features; at the same time, performs graphic feature analysis on the residual Blob image, quickly converts the simple pattern information into the shape information of the pattern, so as to distinguish the real defects from the false defects according to the different graphic features.

6. A method for online detection of steel stamps based on color CCD, characterized in that: Using the steel stamp online detection system as described in any one of claims 1 to 5 comprises the following steps: S101: collecting images of two cigarettes on cigarette paper; S102: color correction and color space transformation of the double cigarette images; S103: locating the edge of the cigarette paper and locating different patterns in the paper; S104: Calculate the similarity between the positioning area of ​​print 1 and the template; S105: Determine whether the similarity is less than the threshold value for the presence or absence of a steel stamp; if the similarity is greater than the threshold value for the presence or absence of a steel stamp, proceed to step S106; if the similarity is less than the threshold value for the presence or absence of a steel stamp, proceed to step S107; S106: Separation of different printing areas; Use the template threshold to segment the printing 1 and printing 2 areas, compare with the steel stamp template, and calculate the brand steel stamp residual points; Color difference detection, calculation of brand steel stamp residual points; Calculate the residual points of the coding stamp through OCR; Then proceed to step S108; S107: Calculate the similarity between the positioning area of ​​print 2 and the template; And judge whether the similarity of print 2 is greater than that of print 1; if the similarity of print 2 is greater than that of print 1, report a joint defect; if the similarity of print 2 is less than that of print 1, report a defect; S108: Determine whether the residual point ratio is greater than a threshold value; If the residual point ratio is greater than the threshold, a defect is reported; if the residual point ratio is less than the threshold, it is a normal stamp.

7. The method for online steel stamp detection based on color CCD as claimed in claim 6, characterized in that: In S102, the image color correction is performed by histogram regulation to enhance the contrast of grayscale values, and the specific steps include: S1021: performing grayscale equalization on the histogram of the collected image; S1022: Select a histogram and perform a histogram equalization transformation; S1023: Invert the transformation obtained in step S1021, and map the original histogram to the selected histogram.

8. The method for online steel stamp detection based on color CCD as claimed in claim 6, characterized in that: In S102, the color space conversion adopts CIELab color space conversion.

Images