CN108458668A - Slab edge and Head and Tail Shape automatic checkout system based on binocular vision and method - Google Patents

Abstract

The invention provides a binocular vision-based automatic detection system for the shape of the edge and head and tail of a slab, which includes an image acquisition module, an image processing module, a system communication module, a software interaction interface and a machine vision detection algorithm module, wherein: image acquisition The module is used for image acquisition of the side, head and tail of the slab; the image processing module is used for image preprocessing of the acquired real-time video images of the side and tail of the slab; the system communication module is used to collect the slab captured by the industrial camera The real-time video images of the edge and tail of the blank are sent to the client, and the acquisition control signal is sent to the industrial camera and the output of the recognition result signal of the machine vision detection algorithm module is controlled; the machine vision detection algorithm module includes a calculation module and a judgment module. The invention has simple equipment, flexible installation, does not affect the layout and operation of the original production line, has high equipment operation reliability and is easy to operate, can prevent workers from working in harsh environments, and reduce labor costs.

Description

Binocular vision-based automatic detection system and method for slab edge and head and tail shapes

technical field

The patent of the present invention relates to the field of slab edge and head and tail shape detection in slab hot rolling production, in particular to an automatic detection system and method for slab edge and head and tail shape based on binocular vision.

Background technique

Aluminum alloy thick plate hot-rolled slabs eliminate the defects of the casting structure due to large reduction rolling, and transform the casting structure into a deformed structure. Its continuation products have great advantages in deep drawing performance, surface quality and precision control. It can be widely used in important fields of the national economy such as aerospace, automobile, machinery manufacturing, shipbuilding and chemical industry. In the past ten years, the degree of automation of the production process has been greatly developed, but the intelligence is far from being realized, and most of the key technical links are still Rely on human experience.

One of the more critical issues is that the initial thickness of the hot-rolled slab of aluminum alloy thick plate is large (500-800mm), the rolling reduction cannot penetrate to the center, and the accumulated deformation is prone to double drums at the edge and head and tail. Defects. At present, in the actual industrial production process, the shape of the slab is controlled by using vertical roll hemming and head and tail cutting. However, due to the lack of effective detection methods, the timing of vertical roll input, rolling reduction, and head and tail cutting Quantity and other controls are mainly realized by manual inspection or production experience. Not only is the operating environment for workers harsh, but it also affects the formulation of rolling regulations, resulting in huge waste of slabs and greatly reducing the yield of finished products.

The machine vision technology developed in recent years has the advantages of high reliability, fast response, and low production cost, and can truly realize the intelligentization of production. Therefore, in the face of rising labor costs and increasing competitive pressure, how to use machine vision Technology, improving its production efficiency, product yield and product quality has become a key technical problem to be solved urgently.

Contents of the invention

The purpose of the present invention is to provide a simple structure, real-time detection and accurate detection device for the lack of slab shape detection device in the current production of aluminum alloy plate hot rolling and rough rolling, which leads to the problem that vertical roll input, head and tail cutting timing and numerical value cannot be accurately controlled. A highly efficient binocular vision slab edge and head and tail shape detection system and method.

The purpose of the present invention is achieved through the following technical solutions:

The invention provides a binocular vision-based automatic detection system for slab edge and head and tail shapes, which includes an image acquisition module, an image processing module, a system communication module, a software interaction interface and a machine vision detection algorithm module, wherein:

The image acquisition module is used for image acquisition of the side, head and tail of the slab. The image acquisition module includes a plurality of industrial cameras and positioning brackets. The positioning brackets are fixed on both sides of the roller table at the exit of the rolling mill. The industrial cameras are installed On the positioning bracket, it is used to obtain real-time video images of the edge and tail of the production line slab, and every two industrial cameras form a binocular system;

The image processing module is used to perform image preprocessing on the acquired real-time video images of the edge and tail of the slab, and the image preprocessing includes foreground extraction and image enhancement of the area of interest of the slab to obtain images of the edge and tail of the slab ;

The system communication module is used to collect the real-time video images of the edge and tail of the slab collected by the industrial camera and send them to the client, and send the acquisition control signal to the industrial camera and control the output of the recognition result signal of the machine vision detection algorithm module ;

The software interaction interface is used to display the real-time video images of the edge and tail of the slab and the operation results of the machine vision detection algorithm module, and realize the control of the operation parameters of the machine vision detection algorithm module;

The machine vision detection algorithm module includes a calculation module and a judgment module,

The calculation module analyzes the processed slab edge and head and tail image information to obtain shape parameters for detecting the slab side and head and tail,

The judging module compares the shape parameters of the detected slab side and head and tail with the preset pass deformation parameters, thereby judging whether vertical roll input, head and tail removal, rolling edge reduction, and head and tail removal are required.

Preferably, the industrial camera is a CCD camera.

Preferably, the two CCD cameras used for edge detection are installed in parallel, and the CCD cameras are used to collect images on the slab; the cameras for head and tail detection are installed at the head and tail positions of the slab to collect images of the head and tail, in order to collect good The image of the head and tail can be installed on the slide, so that it can move at the same speed as the aluminum alloy plate.

Preferably, a database generation module is also included, the database generation module is used to establish a process specification database according to the collected images and processing results, and record the change process of the shape of the edge and head and tail of the slab during the production process.

The present invention also provides a machine vision-based automatic detection method for the edge and head and tail shapes of a slab, comprising the following steps:

S1. Arrange the detection device and install the CCD camera;

S2. Image and video collection, through the CCD camera to collect the edge and head and tail image and video information of the aluminum alloy slab;

S3. Image preprocessing, image preprocessing includes foreground extraction and image enhancement of the area of interest of the slab;

S4. Calculate the shape parameters, obtain the relationship between the CCD camera coordinates, the image coordinates and the world coordinate system through the calibration of the CCD camera parameters and the system calibration, and then calculate the lower part of the world coordinate system and its head and tail shape contours through the binocular vision principle;

S5. Comparison and judgment of the parameters of the preset pass. The edge of the slab and its head and tail shape profile obtained by fitting are compared with the deformation parameters of the preset pass. After the deformation parameters of the preset pass are reached, the vertical roller is put into rolling Edge and cut head and tail.

Preferably, the specific steps for arranging the detection device in S1 are as follows:

S11. First, use a binocular CCD camera to monitor the slab, extract the area of interest of the slab through the foreground, and then highlight the edge and head and tail features of the slab through image enhancement;

S12, CCD camera calibration, the relationship between the CCD camera coordinate system, the image coordinate system and the world coordinate system is obtained through calibration;

S13. Using a binocular CCD camera to detect edge deformation parameters.

Preferably, in S4, calculating the coordinates of the lower part of the world coordinate system and its head and tail key points through the principle of binocular vision specifically includes the following steps:

S41. The imaging plane is set at the front focal length f of the optical center of the lens, the coordinate origin of the left and right imaging planes is at the intersections _O1 and _O2 of the optical axis of the CCD camera and the plane, and a certain point P in space is imaged on the left imaging plane and the right The corresponding coordinates in the plane are P _l (u _l , v _l ) and P _r (u _r , v _r ), assuming that the images of the CCD camera are on the same plane, that is, v _l = v _r , obtained from the geometric relationship:

In the above formula (x _c , y _c , z _c ) are the coordinates of point P in the coordinate system of the left CCD camera, b is the baseline distance, f is the focal length of the two CCD cameras, (u _l , v _l ) and (u _r , v _r ) is the coordinates of point P in the left and right images;

S42. Define parallax as the position difference of a certain point in corresponding points in two images:

From this, the coordinates of a point P in the space in the left CCD camera coordinate system can be calculated as:

S43. Make the world coordinate system coincide with the coordinate system of the left CCD camera, and the obtained (x _c , y _c , z _c ) are the coordinates of point P in the world coordinate system.

Preferably, it also includes S6. Generating a process specification database, the process specification database is established according to the collected images and processing results, and records the change process of the shape of the edge and head and tail of the slab during the production process.

Preferably, the pass deformation parameters at least include the maximum length of the side and head and tail depressions during the rolling process.

Compared with existing detection methods, the beneficial effects of the present invention are:

① The invention has simple equipment, flexible installation, does not affect the layout and operation of the original production line, and has high reliability and simple operation, which can prevent workers from working in harsh environments and reduce labor costs.

②The present invention adopts machine vision technology to collect images of the edge and head and tail of the slab online, and feed back the deformed contour shape of the edge and head and tail in real time, so as to control the input timing of the vertical roller and head and tail cutting, and realize the amount of rolling edge reduction and The precise control of the head and tail cutting amount can achieve the purpose of improving the yield and product quality of aluminum alloy slabs and increasing the economic benefits of production.

③ The present invention can establish a process specification database according to the collected images and processing results, and record the shape change process of the edge and head and tail of the slab during the production process, which can be used as reference materials for technical personnel training and process specification optimization.

Description of drawings

Fig. 1 is a structural schematic block diagram of the present invention;

Fig. 2 is a schematic structural diagram of an image acquisition module of the present invention;

Fig. 3 is the detection schematic diagram of the binocular vision of the present invention;

Fig. 4 is a schematic diagram of a three-dimensional structure of the present invention;

Fig. 5 is a schematic diagram of the workflow of the present invention;

Fig. 6 is the threshold segmentation image in Embodiment 1 of the present invention;

Fig. 7 is the part of the calculation area connection in Embodiment 1 of the present invention;

Fig. 8 is the profile feature of the aluminum alloy extracted in Example 1 of the present invention;

Fig. 9 is the graph after the convex hull operation in embodiment 1 of the present invention;

Fig. 10 is the "double drum" gap contour extraction method in Embodiment 1 of the present invention;

Fig. 11 is a schematic diagram of deformation of the actual rolled slab in the factory in Embodiment 2 of the present invention;

12 is a schematic diagram of images collected by the left camera in Embodiment 2 of the present invention;

Fig. 13 is a schematic diagram of images collected by the right camera in Embodiment 2 of the present invention;

FIG. 14 is a schematic diagram of a disparity map in Embodiment 2 of the present invention;

FIG. 15 is a schematic diagram of matching scores in Embodiment 2 of the present invention.

Detailed ways

Exemplary embodiments, features, and aspects of the present invention will be described in detail below with reference to the accompanying drawings. The same reference numbers in the figures indicate functionally identical or similar elements. While various aspects of the embodiments are shown in drawings, the drawings are not necessarily drawn to scale unless specifically indicated.

Below in conjunction with accompanying drawing, the specific embodiment of patent of the present invention is described further:

The present invention provides a binocular vision-based automatic detection system for slab edge and head and tail shapes, as shown in Figure 1, Figure 2 and Figure 4, which includes an image acquisition module 10, an image processing module 11, and a system communication module 12 , software interaction interface 13 and machine vision detection algorithm module 14, wherein:

The image acquisition module 10 is used for image acquisition of the side, head and tail of the slab. The image acquisition module 10 includes a plurality of industrial cameras and positioning brackets 6. The positioning brackets 6 are fixed on both sides of the roll table at the exit of the rolling mill, and the industrial cameras are installed on the positioning brackets 6. It is used to obtain real-time video images of the edge and tail of the slab on the production line, and every two industrial cameras form a binocular system.

The image processing module 11 is used to perform image preprocessing on the acquired real-time video images of the edge and tail of the slab. The image preprocessing includes foreground extraction and image enhancement of the region of interest of the slab to obtain images of the edge and tail of the slab.

The system communication module 12 is used to collect the real-time video images of the edge and tail of the slab collected by the industrial camera and send it to the client 15, and send the collection control signal to the industrial camera and control the output of the recognition result signal of the machine vision detection algorithm module .

The software interaction interface 13 is used to display the images of the edge and tail of the slab, the results of the algorithm operation, and realize the control of the algorithm operation parameters.

The machine vision detection algorithm module 14 is used to analyze the edge and tail images of the slab, and the machine vision detection algorithm module 14 includes a calculation module 141 and a judgment module 142 .

The calculation module 141 analyzes the collected slab edge and head and tail image information to obtain the shape parameters of the detected slab side and head and tail.

The judging module 142 compares the calculation result with the preset deformation parameters of the pass, so as to judge whether vertical roll input, head and tail cutting, and rolling edge reduction and head and tail cutting are needed.

The pass deformation parameter is the maximum critical length of the "sag" of the side and head and tail during the rolling process, and there are other parameters, but the length is the most obvious, and other parameters can be used as auxiliary, such as the width of the notch in the "sag" area during the rolling process . Assuming that the maximum critical length of "sag" is set to 200mm, when the actual length is greater than 200mm, it is considered to have an impact on the subsequent rolling yield of the entire slab, so hemming and head and tail cutting operations are required. Simply speaking, the deformation parameter of the pass is to set a value artificially based on actual experience. When this value is reached during the rolling process, the judgment of rolling edge and cutting head and tail will be issued.

Preferably, a database generation module is also included, the database generation module is used to establish a process specification database according to the collected images and processing results, and record the change process of the shape of the edge and head and tail of the slab during the production process.

The present invention also provides a machine vision-based automatic detection method for the edge and head and tail shapes of a slab, comprising the following steps:

S1. Arrange the detection device and install the CCD camera;

S2. Image and video collection, through the CCD camera to collect the edge and head and tail image and video information of the aluminum alloy slab;

S3. Image preprocessing, image preprocessing includes foreground extraction and image enhancement of the area of interest of the slab;

S4. Calculate the shape parameters, obtain the relationship between the CCD camera coordinates, the image coordinates and the world coordinate system through the calibration of the CCD camera parameters and the system calibration, and then calculate the lower part of the world coordinate system and its head and tail key point coordinates through the binocular vision principle , get the slab edge and its head and tail shape profile by fitting;

S5. Comparison and judgment of the deformation parameters of the preset pass. The edge of the slab and its head and tail shape profile obtained by fitting are compared with the deformation parameters of the preset pass. After the deformation parameters of the preset pass are reached, the vertical roller is put into Piping and cutting head and tail.

S6. Generate a process specification database. The process specification database is established according to the collected images and processing results, and records the shape change process of the edge and head and tail of the slab during the production process.

Each pass of the rolling process will have different deformation parameters, and the database can record these deformation parameters for subsequent analysis. For example, the length and contour of the "sag" deformation at the edge and head and tail of each pass can be recorded, and digital image processing methods such as feature matching can be used for analysis and self-learning, and finally the edge and tail of the slab during the rolling process can be realized. Online real-time prediction of head and tail "sag" profile features to optimize process.

According to the schematic diagram of Figure 2, the structural connection of the various components of the device is carried out. Figure 2 and Figure 4 include the aluminum alloy slab to be inspected 2, the left CCD camera 3, the right CCD camera 5 and the positioning bracket 6.

The operation process of the slab edge and head and tail shape automatic detection system based on binocular machine vision is as follows:

(1) The arrangement of the device is shown in Figure 2 and Figure 4. First, the binocular CCD camera is used to monitor the aluminum alloy slab, and the region of interest (moving aluminum alloy slab) is extracted through the foreground, and then the aluminum alloy plate is highlighted through image enhancement The edge features of the billet are convenient for later detection.

(2) CCD camera calibration. The relationship between the CCD camera coordinate system, the image coordinate system and the world coordinate system is obtained through calibration.

(4) Obtain the disparity map of the image by using the principle of parallel binocular vision imaging, and use the disparity map to calculate the concave distance of the aluminum alloy edge.

(5) Compare the calculated length of the edge depression with the length of the concave contour set by the preset pass, and when the calculated length reaches the preset length, put in the hemming and cut the head and tail.

Accompanying drawing 3 is the schematic diagram of binocular parallel vision imaging. In fact, the imaging plane of the CCD camera is behind the optical center of the lens. For the convenience of calculation, the imaging plane is set at f (focal length) in front of the optical center of the lens. The coordinate origin of the left and right imaging surfaces is at the intersections _O1 and _O2 of the optical axis of the CCD camera and the plane. The corresponding coordinates of a point P in space in the left and right imaging planes are P _l (u _l , v _l ) and P _r (u _r , v _r ). If the internal parameters of the two CCD cameras are the same and the installation positions are on the same plane, then the two imaging planes are on the same plane. Now assume that the images of the CCD camera are on the same plane, that is, v _l =v _r , obtained from the geometric relationship:

In the above formula (x _c , y _c , z _c ) are the coordinates of point P in the coordinate system of the left CCD camera, b is the baseline distance, f is the focal length of the two CCD cameras, (u _l , v _l ) and (u _r , v _r ) are the coordinates of point P in the left and right images.

Parallax is defined as the positional difference of a point at corresponding points in two images:

From this, the coordinates of a point P in the space in the left CCD camera coordinate system can be calculated as:

Let the world coordinate system coincide with the coordinate system of the left CCD camera, and the obtained (x _c , y _c , z _c ) is the coordinate of point P in the world coordinate system. On the premise that the internal and external parameters of the two CCD cameras are consistent, for any point on the image plane of a certain CCD camera, as long as the corresponding matching point is found on the plane of another CCD camera, the three-dimensional coordinates of the point can be calculated.

Example 1: Measurement of the head and tail of the aluminum alloy slab.

(1) Area of interest setting.

ROI is the abbreviation of RegionOfInterest, that is, the region of interest. For the image that needs to extract features, our area of interest is the aluminum alloy plate, and we don’t care about other things such as factory background, rolls, roller tables, etc., so the set area of interest only needs to include the aluminum alloy plate. It should be noted that the image collected on site is a color image, and the color image needs to be converted into a grayscale image before the ROI area is set.

(2) Image threshold segmentation

The basic idea of threshold segmentation is: for a given grayscale image f(x,y), for a given grayscale range [Z ₁ ,Z ₂ ], t∈[Z ₁ ,Z ₂ ], if:

Then f _t (x, y) is called the binary image of the image f(x, y) with t as the threshold. The threshold threshold used in this paper is 97, and the image after threshold segmentation is shown in Figure 6 below.

(3) Contour feature extraction of aluminum alloy thick plate

After threshold segmentation, calculate the ranges connected together after thresholding, the result is shown in Figure 7, different ranges are represented by different colors, and then select the area with the target feature. The target feature in this paper is the area area. It is obvious from Figure 7 that the area of the aluminum alloy thick plate contour that needs to be extracted is much larger than the area of other areas, so the target feature is set to an area greater than 100,000 pixels (the specific value is adjusted according to the program), and the extraction is obtained The contour characteristics of the aluminum alloy are shown in Fig. 8.

(4), the measurement of the length of "double drum"

After the feature extraction, the length of the "double drum" needs to be measured. The specific idea is to perform a convex hull calculation on the extracted aluminum alloy profile, and then calculate the difference between the obtained graph and the extracted contour features to obtain the notch shape of the "double drum" , and then find the minimum circumscribed rectangle of the gap, the width of the minimum circumscribed rectangle is the length of the "double drum".

(4.1) Convex hull operation

The convex hull is the basic structure of computer geometry and has been widely used in many graphics and image related fields. A convex polygon is a polygon without any depression. After the convex hull operation, the concave part is filled with graphics. The graph obtained after the convex hull operation is shown in Figure 9.

3.2 "Double drum" notch contour extraction and measurement

The graph obtained by the convex hull is subtracted from the extracted contour features, as shown in Figure 10, and the contour of the "double drum" gap is obtained.

The width of the minimum circumscribed rectangle of the obtained notch outline is the length of the "double drum", and the length value is 174.065 pixels. Through camera calibration, it can be measured that the real distance of the "double drum" is 200mm.

Embodiment 2: Measurement of the edge of the aluminum alloy slab.

The edge deformation of the aluminum alloy thick plate during the actual rolling process is shown in Figure 11. The middle groove of the aluminum profile is used to simulate the edge deformation during the rolling process.

Figure 12 and Figure 13 are images collected by binocular cameras, Figure 14 is the generated disparity map, Figure 15 is the matching score, and the software used is halocn13. The depth of the middle groove of the aluminum profile can be calculated by the gray value of the obtained disparity map, which represents the deformation depth of the hot-rolled edge of the aluminum alloy slab. When the depth reaches the set value, it is put into the vertical roll hemming.

Finally, it should be noted that: the above-described embodiments are only used to illustrate the technical solutions of the present invention, rather than to limit them; although the present invention has been described in detail with reference to the foregoing embodiments, those of ordinary skill in the art should understand : It is still possible to modify the technical solutions described in the foregoing embodiments, or perform equivalent replacements to some or all of the technical features; and these modifications or replacements do not make the essence of the corresponding technical solutions deviate from the technical solutions of the various embodiments of the present invention range.

Claims (9)

1. A slab edge and head and tail shape automatic detection system based on binocular vision, is characterized in that: it comprises image acquisition module, image processing module, system communication module, software interactive interface and machine vision detection algorithm module, wherein : The image acquisition module is used for image acquisition of the side, head and tail of the slab. The image acquisition module includes a plurality of industrial cameras and positioning brackets. The positioning brackets are fixed on both sides of the roller table at the exit of the rolling mill. The industrial cameras are installed On the positioning bracket, it is used to obtain real-time video images of the edge and tail of the production line slab, and every two industrial cameras form a binocular system; The image processing module is used to perform image preprocessing on the acquired real-time video images of the edge and tail of the slab, and the image preprocessing includes foreground extraction and image enhancement of the area of interest of the slab to obtain images of the edge and tail of the slab ; The system communication module is used to collect the real-time video images of the edge and tail of the slab collected by the industrial camera and send them to the client, and send the acquisition control signal to the industrial camera and control the output of the recognition result signal of the machine vision detection algorithm module ; The software interaction interface is used to display the real-time video images of the edge and tail of the slab and the operation results of the machine vision detection algorithm module and realize the control of the operation parameters of the machine vision detection algorithm module; The machine vision detection algorithm module includes a calculation module and a judgment module, and the calculation module analyzes the processed slab edge and head and tail image information to obtain the shape parameters of the detected slab side and head and tail; The judging module compares the shape parameters of the detected slab side and head and tail with the preset pass deformation parameters, thereby judging whether vertical roll input, head and tail removal, rolling edge reduction, and head and tail removal are needed. 2. The binocular vision-based automatic inspection system for the edge and head and tail shapes of slabs according to claim 1, wherein the industrial camera is a CCD camera. 3. The binocular vision-based slab edge and head and tail shape automatic detection system according to claim 2, characterized in that: the two CCD cameras are installed in parallel, and the CCD cameras are used to collect aluminum alloy slabs The edge and head and tail image video information. 4. The slab edge and head and tail shape automatic detection system based on binocular vision according to claim 1, characterized in that: it also includes a database generation module, and the database generation module is used to establish The process specification database records the change process of the shape of the edge and head and tail of the slab during the production process. 5. A slab edge and head and tail shape automatic detection method based on binocular vision, is characterized in that: comprise the following steps: S1. Arrange the detection device and install the CCD camera; S2. Image and video collection, through the CCD camera to collect the edge and head and tail image and video information of the aluminum alloy slab; S3. Image preprocessing, image preprocessing includes foreground extraction and image enhancement of the area of interest of the slab; S4. Calculate the shape parameters, obtain the relationship between the CCD camera coordinates, the image coordinates and the world coordinate system through the calibration of the CCD camera parameters and the system calibration, and then calculate the lower part of the world coordinate system and its head and tail shape contours through the binocular vision principle; S5. Comparison and judgment of the deformation parameters of the preset pass. The edge of the slab and its head and tail shape profile obtained by machine vision measurement are compared with the deformation parameters of the preset pass. After the deformation parameters of the preset pass are reached, the vertical roll is put into the Do the piping and trim the ends. 6. The binocular vision-based slab edge and head and tail shape automatic detection method according to claim 5, characterized in that: the specific steps of arranging the detection device in S1 are as follows: S11. First, use a binocular CCD camera to monitor the slab, extract the area of interest of the slab through the foreground, and then highlight the edge and head and tail features of the slab through image enhancement; S12, CCD camera calibration, the relationship between the CCD camera coordinate system, the image coordinate system and the world coordinate system is obtained through calibration; S13, using the binocular CCD camera to measure the deformation of the edge of the slab and the monocular camera to measure the deformation of the head and tail. 7. The binocular vision-based slab edge and head and tail shape automatic detection method according to claim 5, characterized in that: in S4, the lower edge of the world coordinate system and the key points of the head and tail are calculated by the principle of binocular vision Coordinates specifically include the following steps: S41. The imaging plane is set at the front focal length f of the optical center of the lens, the coordinate origin of the left and right imaging planes is at the intersections _O1 and _O2 of the optical axis of the CCD camera and the plane, and a certain point P in space is imaged on the left imaging plane and the right The corresponding coordinates in the plane are P _l (u _l , v _l ) and P _r (u _r , v _r ), assuming that the images of the CCD camera are on the same plane, that is, v _l = v _r , obtained from the geometric relationship: In the above formula (x _c , y _c , z _c ) are the coordinates of point P in the coordinate system of the left CCD camera, b is the baseline distance, f is the focal length of the two CCD cameras, (u _l , v _l ) and (u _r , v _r ) is the coordinates of point P in the left and right images; S42. Define parallax as the position difference of a certain point in corresponding points in two images: From this, the coordinates of a point P in the space in the left CCD camera coordinate system can be calculated as: S43. Make the world coordinate system coincide with the coordinate system of the left CCD camera, and the obtained (x _c , y _c , z _c ) are the coordinates of point P in the world coordinate system. 8. The binocular vision-based slab edge and head and tail shape automatic detection method according to claim 5, characterized in that: it also includes S6, generating a process specification database, and the process specification database is based on the collected images and processing results Establish and record the shape change process of the edge and head and tail of the slab during the production process. 9. The binocular vision-based automatic detection method for the edge and head and tail shapes of slabs according to claim 5, characterized in that: the preset pass deformation parameters at least include the maximum side and head and tail depressions during the rolling process length.