An image processing method based on strip surface detection

Abstract

The present invention relates to the technical field of industrial defect detection, and discloses an image processing method based on strip surface detection, including strip surface defect intelligent detection equipment, data acquisition module and intelligent detection module. The method obtains high-definition image data and high-definition light and shadow data of all strip surfaces through the data acquisition module, and classifies and forms a data set. The intelligent detection module analyzes and generates a grayscale data group and an appearance index, determines whether there are potholes and color differences on the strip surface, and transmits the judgment result to the central control console. Defect recognition is performed based on pixel grayscale values, with high computational efficiency, capable of meeting the real-time detection requirements in industrial production, timely discovering and processing strip surface defects, and flexible adaptation detection accuracy. The intelligent detection module compares the light and shadow data of all viewing angles of the strip surface with the reference image frame by frame. If there is a difference in any group of images, it means that there are texture defects on the surface of the strip section, and the coating quality is good.

Description

Technical Field

[0001] The invention relates to the technical field of industrial defect detection, and in particular to an image processing method based on strip steel surface detection. Background Art

[0002] The intelligent detection system for strip surface defects uses advanced artificial intelligence technology and machine vision technology to perform real-time and efficient quality inspection and defect identification on the surface of strip steel. Strip surface defect detection usually adopts a variety of methods to ensure accurate control of product quality and process. The detection method based on machine vision collects high-definition images through high-resolution cameras and identifies defects through image processing technology. This method has a fast detection speed but relatively low accuracy. Since different materials and defect types show different characteristics in the spectrum, defects can be identified and classified by spectral analysis detection methods. In addition, different defect types will produce different reflections and scattering of sound waves, so defects can be detected and identified by analyzing vibration tones. The detection method based on laser scanners uses laser scanners to measure the surface of the strip in three dimensions and detect defects by analyzing changes in surface morphology. This method is suitable for detecting defects with greater depth. There are many methods for strip surface defect detection, each with its own advantages and disadvantages. With the development of technology, the development of more efficient, accurate and intelligent strip surface defect detection systems will become the focus of future research.

[0003] At present, the traditional image processing method based on strip surface detection has a single detection mode and low detection accuracy, and is unable to identify surface potholes and scratches in real time. In addition, the quality of raw materials and the temperature control, cooling speed, rolling process, etc. in the production process will cause uneven color or spots, stripes and other defects on the strip surface. The traditional strip surface defect intelligent detection system cannot identify color difference defects and texture defects of strips with color coatings. Summary of the invention

[0004] 1. Technical issues to be solved

[0005] In view of the shortcomings of the prior art, the present invention provides an image processing method based on strip surface detection, which has the advantages of flexible adaptation, high detection accuracy, and efficient identification of good coating quality. It solves the problem that the traditional strip surface defect intelligent detection system has low detection accuracy and cannot identify strip defects with color coating.

[0006] (II) Technical solution

[0007] To achieve the above-mentioned object, the present invention provides the following technical solutions: an image processing method based on strip surface detection, wherein the image processing method based on strip surface detection performs detection through a strip surface defect intelligent detection device, a data acquisition module and an intelligent detection module;

[0008] The strip surface defect intelligent detection equipment includes a central control console, a high-definition camera, a projector, a transparent placement table, a conveyor belt, a mechanical arm, a white detection chamber and a black detection chamber; the data acquisition module is composed of an image data unit and a light and shadow data unit, the image data unit is connected to the high-definition camera through a network to collect an image data set, the image data set includes high-definition image data of all strip surfaces, the light and shadow data unit is connected to the high-definition camera through a network to collect a light and shadow data set, the light and shadow data set includes high-definition light and shadow data of all strip surfaces, and the data acquisition module transmits the image data set and the light and shadow data set to the intelligent detection module through the network;

[0009] The intelligent detection module is composed of an image analysis unit and a light and shadow analysis unit. The image analysis unit is connected to the central control console via a network to obtain a unified reference image C and a reference color value CYA of all strip surfaces, and then combined with the image data set, analyzes and generates a grayscale data set Hdsj and an appearance index Wzs to determine whether there are potholes and color differences on the strip surface. If the appearance index Wzs is not equal to 1, it indicates that there is color difference on the strip surface, and the judgment result is transmitted to the central control console. The light and shadow analysis unit is connected to the central control console via a network to obtain a unified reference image YX of all strip surfaces, and then combined with the light and shadow data set to determine whether there are texture defects on the strip surface, and the judgment result is transmitted to the central control console.

[0010] Preferably, the central control console is electrically connected to a high-definition camera, a projector and a conveyor belt, the projector is used to project a light beam pattern onto the surface of the steel strip, two transparent placement tables are provided, and the top surface of each transparent placement table is set as a curved surface, the conveyor belt is set around a white detection bin and a black detection bin, the inner wall of the white detection bin is set as a white background, and the black detection bin is set as a black background.

[0011] Preferably, the expression of the image data set is {T1 x 、T2 x 、T3 x , ..., Tn x}, T1 x To Tn x They are respectively the high-definition image data of the surface of the strip from the first to the nth section, x represents the high-definition image data of the surface of the strip, and the high-definition image data includes front perspective image, left perspective image, right perspective image, back perspective image, top perspective image and bottom perspective image.

[0012] Preferably, the expression of the light and shadow data set is {G1 y , G2 y , G3 y , ..., Gn y}, G1 y To Gny They are the high-definition light and shadow data of the surface of the strip from the first to the nth section respectively, y represents the high-definition light and shadow data of the surface of the strip, and the high-definition light and shadow data include front perspective light and shadow, left perspective light and shadow, right perspective light and shadow, back perspective light and shadow, top perspective light and shadow and upward perspective light and shadow.

[0013] Preferably, the reference image C includes a front view reference image C zm , left view reference image C zc , right view reference image C yc , back view reference image C bm , top-down reference image C fs and the upward reference image C ys The image analysis unit identifies C through image processing software zm , C zc , C yc , C bm , C fs and C ys The grayscale value of each pixel in the image is calculated, and the grayscale threshold is generated according to the minimum and maximum grayscale values ​​in each image. The grayscale threshold includes the front view grayscale threshold HY zm , left viewing angle grayscale threshold HY zc , right viewing angle grayscale threshold HY yc , back viewing angle grayscale threshold HY bm , Top-down viewing angle grayscale threshold HY fs and the grayscale threshold HY for the upward viewing angle ys .

[0014] Preferably, the grayscale data set Hdsj calculation process is as follows:

[0015] Extract the high-definition image data of the surface of the i-th strip in the image data set and mark it as Ti x ;

[0016] Ti x The front view image in is marked as Ti zm , Ti x The left view image in the figure is marked as Ti zc , Ti x The right view image in the figure is marked as Ti yc , Ti x The back view image in is marked as Ti bm , Ti x The top-down image in the image is marked as Ti fs , Ti x The upward-view image in is marked as Ti ys ;

[0017] Identify Ti by image processing softwarezm 、Ti zc 、Ti yc 、Ti bm 、Ti fs and Ti ys The gray value of each pixel in Ti zm The gray value of the pixel is marked as HTi zm , Ti zc The gray value of the pixel is marked as HTi zc , Ti yc The gray value of the pixel is marked as HTi yc , Ti bm The gray value of the pixel is marked as HTi bm , Ti fs The gray value of the pixel is marked as HTi fs , Ti ys The gray value of the pixel is marked as HTi ys ;

[0018]

[0019] In the formula, Hdsj represents the grayscale data group, minHY zm ≤HTi zm ≤maxHY zm Indicates that Ti zm Gray value of the middle pixel compared to the gray threshold HY of the front view zm , minHY zc ≤HTi zc ≤maxHY zc Indicates that Ti zc Gray value of the middle pixel compared to the gray threshold HY of the left viewing angle zc , minHY yc ≤HTi yc ≤maxHY yc Indicates that Ti yc Gray value of the middle pixel compared to the gray threshold HY of the right viewing angle yc , minHY bm ≤HTi bm ≤maxHY bm Indicates that Ti bm The gray value of the middle pixel is compared with the gray threshold HY of the back view bm , minHY fs ≤HTi fs ≤maxHY fs Indicates that Ti fs Gray value of the middle pixel compared to the gray threshold HY of the top-down angle fs , minHY ys ≤HTiys ≤maxHY ys Indicates that Ti ys Gray value of the middle pixel compared to the gray threshold HY of the upward viewing angle ys .

[0020] Preferably, when the grayscale value of any pixel point in the grayscale data group Hdsj exceeds the grayscale threshold HY, it indicates that there are potholes on the surface of the strip steel section.

[0021] Preferably, the calculation process of the appearance index Wzs is as follows:

[0022] By using image processing software, Ti zm and C zm 、Ti zc and C zc 、Ti yc and C yc 、Ti bm and C bm 、Ti fs and C fs 、Ti ys and C ys Perform a positive film overlay and obtain a mixed color value YA after the positive film overlay of each group of images;

[0023]

[0024] In the formula, YA represents the mixed color value after a single set of images are superimposed, T ya represents the color value of the steel strip surface in the image dataset, C ya Represents the color value in the reference image, 255 represents the constant in the color value calculation formula, Indicates the mixed color value after a single set of strip surface images and a reference image are superimposed;

[0025]

[0026] In the formula, Wzs represents the appearance index, YA zm Ti zm and C zm The mixed color value after multiplication, YA zc Ti zc and C zc The mixed color value after multiplication, YA yc Ti yc and C yc The mixed color value after multiplication, YA bm Ti bm and C bm The mixed color value after multiplication, YA fs Ti fsand C fs The mixed color value after multiplication, YA ys Ti ys and C ys The mixed color value after multiplication. The ratio of the sum of the mixed color values ​​of the six groups of images on the surface of the i-th section of the strip after superposition to the reference color value is the appearance index.

[0027] Preferably, the reference image YX includes a front view reference image YX zm , left view reference image YX zc , right view reference image YX yc , back view reference image YX bm , top-down reference image YX fs and the upward reference image YX ys .

[0028] Preferably, the light and shadow analysis unit determines whether there are texture defects on the surface of the strip according to the reference image YX and the light and shadow data set, and the judgment process is as follows:

[0029] Extract the light and shadow data of the i-th strip surface in the light and shadow data set and mark it as Gi y ;

[0030] Will Gi y The front view light and shadow in the image is marked as Gi zm , will Gi y The left perspective light and shadow in the image is marked as Gi zc , will Gi y The right perspective light and shadow in the image is marked as Gi yc , will Gi y The back view light and shadow in the image is marked as Gi bm , will Gi y The top-down perspective light and shadow in the image is marked as Gi fs , will Gi y The upward perspective light and shadow in the image is marked as Gi ys ;

[0031] Through video processing software zm and YX zm 、Gi zc and YX zc 、Gi yc and YX yc 、Gi bm and YX bm 、Gi fs and YX fs 、Gi ys and YX ysA frame-by-frame comparison is performed. If any group of images is different, it means that there is a texture defect on the surface of the i-th section of the strip.

[0032] Compared with the prior art, the present invention provides an intelligent detection system and equipment for strip surface defects, which have the following beneficial effects:

[0033] 1. The present invention obtains high-definition image data and high-definition light and shadow data of all strip surfaces by connecting the high-definition camera through the data acquisition module network, and classifies them into image data sets and light and shadow data sets. The intelligent detection module obtains the unified reference image C and reference color value CYA of all strip surfaces through the network connection of the central control console, and then combines the image data set to analyze and generate the grayscale data group Hdsj and the appearance index Wzs, and performs defect recognition based on the pixel grayscale value. It has high calculation efficiency, can meet the real-time detection needs in industrial production, timely discover and process strip surface defects, and has high flexible adaptation detection accuracy.

[0034] 2. The present invention determines whether there are potholes and color differences on the surface of the steel strip by analyzing the grayscale data group Hdsj and the appearance index Wzs. If the grayscale value of any pixel point in the grayscale data group Hdsj exceeds the grayscale threshold HY, it indicates that there are potholes on the surface of the steel strip. If the appearance index Wzs is not equal to 1, it indicates that there is color difference on the surface of the steel strip, and the judgment result is transmitted to the central control console. The closer the appearance index is to 1, the finer the color difference is, and the judgment result is transmitted to the central control console. The intelligent detection module connects to the central control console through the network to obtain a unified reference image YX of all steel strip surfaces, and then compares the light and shadow data of all viewing angles of the steel strip surface with the reference image YX frame by frame through the video processing software. If there is a difference in any group of images, it indicates that there are texture defects on the surface of the steel strip, and the judgment result is transmitted to the central control console, so as to efficiently identify the coating quality. BRIEF DESCRIPTION OF THE DRAWINGS

[0035] Figure 1 It is a schematic diagram of the process of the present invention. DETAILED DESCRIPTION

[0036] The following will be combined with the drawings in the embodiments of the present invention to clearly and completely describe the technical solutions in the embodiments of the present invention. Obviously, the described embodiments are only part of the embodiments of the present invention, not all of the embodiments. Based on the embodiments of the present invention, all other embodiments obtained by ordinary technicians in this field without creative work are within the scope of protection of the present invention.

[0037] Since the traditional image processing method based on strip surface detection has a single detection method and low detection accuracy, it is unable to identify surface pits and scratches in real time. In addition, the quality of raw materials and the temperature control, cooling speed, rolling process, etc. in the production process will cause uneven color or spots, stripes and other defects on the strip surface. The traditional strip surface defect intelligent detection system cannot identify color difference defects and texture defects of strips with color coatings. Therefore, an image processing method based on strip surface detection is provided. Please refer to Figure 1 , based on the image processing method of strip surface detection, detection is performed through the strip surface defect intelligent detection equipment, data acquisition module and intelligent detection module;

[0038] The intelligent detection equipment for strip steel surface defects includes a central control console, a high-definition camera, a projector, a transparent placement table, a conveyor belt, a robotic arm, a white detection chamber and a black detection chamber. The central control console is electrically connected to the high-definition camera, the projector and the conveyor belt. The projector is used to project a light beam pattern onto the surface of the strip steel. Two transparent placement tables are provided, and the top surface of each transparent placement table is set as a curved surface to prevent the strip steel from being adsorbed on the transparent placement table. The conveyor belt is set around the white detection chamber and the black detection chamber to facilitate the rapid classification of strip steel with different types of defects. The inner wall of the white detection chamber is set as a white background for collecting high-definition image data, and the black detection chamber is set as a black background for collecting high-definition light and shadow data.

[0039] The data acquisition module consists of an image data unit and a light and shadow data unit. The image data unit collects image data sets through a network connection to a high-definition camera. The image data set includes high-definition image data of all strip surfaces. The expression of the image data set is {T1 x 、T2 x 、T3 x , ..., Tn x}, T1 x To Tn x are high-definition image data of the surface of the first to nth sections of the strip steel, respectively, and x represents the high-definition image data of the surface of the strip steel. The high-definition image data includes front view image, left view image, right view image, back view image, top view image and top view image, which are convenient for subsequent rapid identification of whether there are obvious pits and color differences on the surface coating of the strip steel;

[0040] The light and shadow data unit collects light and shadow data sets through a network connection with a high-definition camera. The light and shadow data sets include high-definition light and shadow data of all strip surfaces. The expression of the light and shadow data set is {G1 y , G2 y , G3 y , ..., Gn y}, G1 y To Gn yare high-definition light and shadow data of the surface of the first to nth sections of the strip steel, respectively; y represents high-definition light and shadow data of the surface of the strip steel, and the high-definition light and shadow data include light and shadow from the front view, light and shadow from the left view, light and shadow from the right view, light and shadow from the back view, light and shadow from the top view, and light and shadow from the top view. Through the light and shadow reflected from the surface of the strip steel, it is possible to quickly identify whether there are obvious texture defects on the surface coating of the strip steel;

[0041] The data acquisition module transmits the image data set and the light and shadow data set to the intelligent detection module through the network;

[0042] The intelligent detection module consists of an image analysis unit and a light and shadow analysis unit. The image analysis unit is connected to the central control console through the network to obtain the unified reference image C and reference color value CYA of all strip surfaces, and then combines the image data set to analyze and generate the grayscale data set Hdsj and appearance index Wzs. The calculation process is as follows:

[0043] Reference image C includes front view reference image C zm , left view reference image C zc , right view reference image C yc , back view reference image C bm , top-down reference image C fs and the upward reference image C ys The image analysis unit recognizes C through image processing software zm , G zc , C yc , C bm , C fs and C ys The grayscale value of each pixel in the image is calculated, and the grayscale threshold is generated according to the minimum and maximum grayscale values ​​in each image. The grayscale threshold includes the front view grayscale threshold HY zm , left viewing angle grayscale threshold HY zc , right viewing angle grayscale threshold HY yc , back viewing angle grayscale threshold HY bm , Top-down viewing angle grayscale threshold HY fs and the grayscale threshold HY for the upward viewing angle ys ;

[0044] Extract the high-definition image data of the surface of the i-th strip in the image data set and mark it as Ti x ;

[0045] Ti x The front view image in is marked as Ti zm , Ti x The left view image in the figure is marked as Ti zc , Ti x The right view image in the figure is marked as Ti yc , Tix The back view image in is marked as Ti bm , Ti x The top-down image in the image is marked as Ti fs , Ti x The upward-view image in is marked as Ti ys ;

[0046] Identify Ti by image processing software zm 、Ti zc 、Ti yc 、Ti bm 、Ti fs and Ti ys The gray value of each pixel in Ti zm The gray value of the pixel is marked as HTi zm , Ti zc The gray value of the pixel is marked as HTi zc , Ti yc The gray value of the pixel is marked as HTi yc , Ti bm The gray value of the pixel is marked as HTi bm , Ti fs The gray value of the pixel is marked as HTi fs , Ti ys The gray value of the pixel is marked as HTi ys ;

[0047]

[0048] In the formula, Hdsj represents the grayscale data group, minHY zm ≤HTi zm ≤maxHY zm Indicates that Ti zm Gray value of the middle pixel compared to the gray threshold HY of the front view zm , minHY zc ≤HTi zc ≤maxHY zc Indicates that Ti zc Gray value of the middle pixel compared to the gray threshold HY of the left viewing angle zc , minHY yc ≤HTi yc ≤maxHY yc Indicates that Ti yc Gray value of the middle pixel compared to the gray threshold HY of the right viewing angle yc , minHY bm ≤HTi bm ≤maxHY bm Indicates that Ti bmThe gray value of the middle pixel is compared with the gray threshold HY of the back view bm , minHY fs ≤HTi fs ≤maxHY fs Indicates that Ti fs Gray value of the middle pixel compared to the gray threshold HY of the top-down angle fs , minHY ys ≤HTi ys ≤maxHY ys Indicates that Ti ys Gray value of the middle pixel compared to the gray threshold HY of the upward viewing angle ys , defect recognition based on image pixel grayscale values ​​has high computational efficiency, can meet the real-time detection needs in industrial production, timely discover and deal with strip surface defects, and has high flexible adaptation detection accuracy;

[0049] By using image processing software, Ti zm and C zm 、Ti zc and C zc 、Ti yc and C yc 、Ti bm and C bm 、Ti fs and C fs 、Ti ys and C ys Perform a positive film overlay and obtain a mixed color value YA after the positive film overlay of each group of images;

[0050]

[0051] In the formula, YA represents the mixed color value after a single set of images are superimposed, T ya represents the color value of the steel strip surface in the image dataset, C ya Represents the color value in the reference image, 255 represents the constant in the color value calculation formula, Indicates the mixed color value after the single set of strip surface images and the reference image are superimposed. The mixed color value after the superimposition of each set of images is calculated according to the above formula;

[0052]

[0053] In the formula, Wzs represents the appearance index, YA zm Ti zm and C zm The mixed color value after multiplication, YA zc Ti zc and C zc The mixed color value after multiplication, YA ycTi yc and C yc The mixed color value after multiplication, YA bm Ti bm and C bm The mixed color value after multiplication, YA fs Ti fs and C fs The mixed color value after multiplication, YA ys Ti ys and C ys The mixed color value after multiplication. The ratio of the sum of the mixed color values ​​of the six groups of images on the surface of the i-th strip after the positive film is superimposed to the reference color value is the appearance index. The closer the appearance index is to 1, the more subtle the color difference is.

[0054] The image analysis unit determines whether there are pits and color differences on the surface of the steel strip according to the grayscale data group Hdsj and the appearance index Wzs. If the grayscale value of any pixel point in the grayscale data group Hdsj exceeds the grayscale threshold HY, it indicates that there are pits on the surface of the steel strip. If the appearance index Wzs is not equal to 1, it indicates that there is color difference on the surface of the steel strip. The judgment result is transmitted to the central control console.

[0055] The light and shadow analysis unit connects to the central control console through the network to obtain a unified reference image YX of all strip surfaces, and then combines the light and shadow data set to determine whether there are texture defects on the strip surface, and transmits the judgment result to the central control console. The judgment process is as follows:

[0056] Reference image YX includes front view reference image YX zm , left view reference image YX zc , right view reference image YX yc , back view reference image YX bm , top-down reference image YX fs and the upward reference image YX ys ;

[0057] Extract the light and shadow data of the i-th strip surface in the light and shadow data set and mark it as Gi y ;

[0058] Will Gi y The front view light and shadow in the image is marked as Gi zm , will Gi y The left perspective light and shadow in the image is marked as Gi zc , will Gi y The right perspective light and shadow in the image is marked as Gi yc , will Gi y The back view light and shadow in the image is marked as Gi bm , will Giy The top-down perspective light and shadow in the image is marked as Gi fs , will Gi y The upward perspective light and shadow in the image is marked as Gi ys ;

[0059] Through video processing software zm and YX zm 、Gi ze and YX zc 、Gi yc and YX yc 、Gi bm and YX bm 、Gi fs and YX fs 、Gi ys and YX ys A frame-by-frame comparison is performed. If there is a difference in any group of images, it means that there is a texture defect on the surface of the i-th section of the strip, and the coating quality is good.

[0060] Example 1: In this experiment, a steel strip with grayscale values ​​of 115 for six plane pixels is selected as the experimental object. According to statistics, the grayscale threshold of the front viewing angle of this batch of steel strip is 100-120, the grayscale threshold of the left viewing angle is 100-120, the grayscale threshold of the right viewing angle is 100-120, the grayscale threshold of the back viewing angle is 100-120, the grayscale threshold of the top viewing angle is 120-140, and the grayscale threshold of the upward viewing angle is 120-140. The calculation formula of the grayscale data group Hdsj of this section of steel strip is as follows:

[0061]

[0062] In the formula, Hdsj represents the grayscale data group, 100≤115≤120 means that the grayscale value of the pixel in the front view image of the strip is included in the front view grayscale threshold 100-120, 100≤115≤120 means that the grayscale value of the pixel in the left view image of the strip is included in the left view grayscale threshold 100-120, 100≤115≤120 means that the grayscale value of the pixel in the right view image of the strip is included in the right view grayscale threshold 100-12 0, 100≤115≤120 means that the grayscale value of the pixel in the back view image of the strip is included in the back view grayscale threshold 100-120, 120≤115≤140 means that the grayscale value of the pixel in the top view image of the strip is not included in the top view grayscale threshold 120-140, 120≤115≤140 means that the grayscale value of the pixel in the top view image of the strip is not included in the upward view grayscale threshold 120-140. It is judged that there are potholes on the surface of this section of strip.

[0063] Example 2: In this experiment, a steel strip with six viewing angle images having color values ​​of 150 is selected as the experimental object. According to statistics, the color values ​​of the six viewing angle images in the reference image are all 160, and the reference color value is 6.9. The calculation formula of the appearance index Wzs of this section of steel strip is as follows:

[0064]

[0065] In the formula, Wzs represents the appearance index, It represents the ratio of the sum of the mixed color values ​​of the six groups of images on the surface of the strip after superimposition to the reference color value. 1.0 is the appearance index. It is judged that there is no color difference on the surface of the strip.

[0066] Although embodiments of the present invention have been shown and described, it will be appreciated by those skilled in the art that various changes, modifications, substitutions and variations may be made to the embodiments without departing from the principles and spirit of the present invention, and that the scope of the present invention is defined by the appended claims and their equivalents.

Claims

1. An image processing method based on strip surface detection, characterized in that: The image processing method based on strip surface detection is detected by using strip surface defect intelligent detection equipment, data acquisition module and intelligent detection module; The intelligent detection equipment for strip surface defects includes a central control console, a high-definition camera, a projector, a transparent placement table, a conveyor belt, a robotic arm, a white detection chamber and a black detection chamber; The data acquisition module is composed of an image data unit and a light and shadow data unit. The image data unit is connected to a high-definition camera through a network to collect an image data set, and the image data set includes high-definition image data of all strip surfaces. The light and shadow data unit is connected to a high-definition camera through a network to collect a light and shadow data set, and the light and shadow data set includes high-definition light and shadow data of all strip surfaces. The data acquisition module transmits the image data set and the light and shadow data set to the intelligent detection module through the network; The intelligent detection module is composed of an image analysis unit and a light and shadow analysis unit. The image analysis unit is connected to the central control console through the network to obtain the unified reference image C and reference color value CYA of all strip surfaces, and then combined with the image data set, analyzes and generates a grayscale data group Hdsj and an appearance index Wzs to determine whether there are potholes and color differences on the strip surface. The calculation process of the grayscale data group Hdsj is as follows: Extract the high-definition image data of the surface of the i-th strip in the image data set and mark it as Ti x ; Ti x The front view image in is marked as Ti zm , Ti x The left view image in the figure is marked as Ti zc , Tl x The right view image in the figure is marked as Ti yc , Ti x The back view image in is marked as Ti bm , Ti x The top-down image in the image is marked as Ti fs , Ti x The upward-view image in is marked as Ti ys ; Identify Ti by image processing software zm 、Ti zc 、Ti yc 、Ti bm 、Ti fs and Ti ys The gray value of each pixel in Ti zm The gray value of the pixel is marked as HTi zm , Ti zc The gray value of the pixel is marked as HTi zc , Ti yc The gray value of the pixel is marked as HTi yc , Ti bm The gray value of the pixel is marked as HTi bm , Ti fs The gray value of the pixel is marked as HTi fs , Ti ys The gray value of the pixel is marked as HTi ys ; In the formula, Hdsj represents the grayscale data group, minHY zm ≤HTi zm ≤maxHY zm Indicates that Ti zm Gray value of the middle pixel compared to the gray threshold HY of the front view zm , minHY zc ≤HTi zc ≤maxHY zc Indicates that Ti zc Gray value of the middle pixel compared to the gray threshold HY of the left viewing angle zc , minHY yc ≤HTi yc ≤maxHY yc Indicates that Ti yc Gray value of the middle pixel compared to the gray threshold HY of the right viewing angle yc , minHY bm ≤HTi bm ≤maxHY bm Indicates that Ti bm The gray value of the middle pixel is compared with the gray threshold HY of the back view bm , minHY fs ≤HTi fs ≤maxHY fs Indicates that Ti fs Gray value of the middle pixel compared to the gray threshold HY of the top-down angle fs , minHY ys ≤HTi ys ≤maxHY ys Indicates that Ti ys Gray value of the middle pixel compared to the gray threshold HY of the upward viewing angle ys ; When the grayscale value of any pixel in the grayscale data set Hdsj exceeds the grayscale threshold HY, it indicates that there are potholes on the surface of the strip steel section; The calculation process of the appearance index Wzs is as follows: By using image processing software, Ti zm and C zm 、Ti zc and C zc 、Ti yc and C yc 、Ti bm and C bm 、Ti fs and C fs 、Ti ys and C ys Perform a positive film overlay and obtain a mixed color value YA after the positive film overlay of each group of images; In the formula, YA represents the mixed color value after a single set of images are superimposed, T ya represents the color value of the steel strip surface in the image dataset, C ya Represents the color value in the reference image, 255 represents the constant in the color value calculation formula, Indicates the mixed color value after a single set of strip surface images and a reference image are superimposed; In the formula, Wzs represents the appearance index, YA zm Ti zm and C zm The mixed color value after multiplication, YA zc Ti zc and C zc The mixed color value after multiplication, YA yc Ti yc and Cy c The mixed color value after multiplication, YA bm Ti bm and C bm The mixed color value after multiplication, YA fs Ti fs and C fs The mixed color value after multiplication, YA ys Ti ys and C ys The mixed color value after multiplication. The ratio of the sum of the mixed color values ​​of the six groups of images on the surface of the i-th strip after the positive film is superimposed to the reference color value is the appearance index; If the appearance index Wzs is not equal to 1, it means that there is color difference on the surface of the strip steel, and the judgment result is transmitted to the central control console. The light and shadow analysis unit connects to the central control console through the network to obtain a unified reference image YX of all strip steel surfaces, and then combines the light and shadow data set to judge whether there are texture defects on the strip steel surface, and transmits the judgment result to the central control console.

2. The image processing method based on strip surface detection according to claim 1 is characterized in that: The central control console is electrically connected to a high-definition camera, a projector and a conveyor belt. The projector is used to project a light beam pattern onto the surface of the steel strip. Two transparent placement tables are provided, and the top surface of each transparent placement table is set as a curved surface. The conveyor belt is set around a white detection bin and a black detection bin. The inner wall of the white detection bin is set as a white background, and the black detection bin is set as a black background.

3. The image processing method based on strip surface detection according to claim 2 is characterized in that: The expression of the image dataset is {T1 x 、T2 x 、T3 x , ..., Tn x }, T1 x To Tn x They are respectively the high-definition image data of the surface of the strip from the first section to the nth section, x represents the high-definition image data of the surface of the strip, and the high-definition image data includes front perspective image, left perspective image, right perspective image, back perspective image, top perspective image and bottom perspective image.

4. The image processing method based on strip surface detection according to claim 3 is characterized in that: The expression of the light and shadow data set is {G1 y , G2 y , G3 y , ..., Gn y }, G1 y To Gn y They are the high-definition light and shadow data of the surface of the strip from the first to the nth section respectively, y represents the high-definition light and shadow data of the surface of the strip, and the high-definition light and shadow data include front perspective light and shadow, left perspective light and shadow, right perspective light and shadow, back perspective light and shadow, top perspective light and shadow and upward perspective light and shadow.

5. The image processing method based on strip surface detection according to claim 4 is characterized in that: The reference image C includes a front view reference image C zm , left view reference image C zc , right view reference image C yc , back view reference image C bm , top-down reference image C fs and the upward reference image C ys The image analysis unit identifies C through image processing software zm , C zc , C yc , C bm , C fs and C ys The grayscale value of each pixel in the image is calculated, and the grayscale threshold is generated according to the minimum and maximum grayscale values ​​in each image. The grayscale threshold includes the front view grayscale threshold HY zm , left viewing angle grayscale threshold HY zc , right viewing angle grayscale threshold HY yc , back viewing angle grayscale threshold HY bm , Top-down viewing angle grayscale threshold HY fs and the grayscale threshold HY ys .

6. The image processing method based on strip surface detection according to claim 5 is characterized in that: The reference image YX includes a front view reference image YX zm , left view reference image YX zc , right view reference image YX yc , back view reference image YX bm , top-down reference image YX fs and the upward reference image YX ys .

7. The image processing method based on strip surface detection according to claim 6 is characterized in that: The light and shadow analysis unit determines whether there are texture defects on the strip surface based on the reference image YX and the light and shadow data set. The judgment process is as follows: Extract the light and shadow data of the i-th strip surface in the light and shadow data set and mark it as Gi y ; Will Gi y The front view light and shadow in the image is marked as Gi zm , will Gi y The left perspective light and shadow in the image is marked as Gi zc , will Gi y The right perspective light and shadow in the image is marked as Gi yc , will Gi y The back view light and shadow in the image is marked as Gi bm , will Gi y The top-down perspective light and shadow in the image is marked as Gi fs , will Gi y The upward perspective light and shadow in the image is marked as Gi ys ; Through video processing software zm and YX zm 、Gi zc and YX zc 、Gi yc and YX yc 、Gi bm and YX bm 、Gi fs and YX fs 、Gi ys and YX ys A frame-by-frame comparison is performed. If any group of images is different, it means that there is a texture defect on the surface of the i-th section of the strip.

Images