Method for measuring speed of casting blank by industrial CCD (charge coupled device) based on AI (artificial intelligence)

Abstract

The invention provides a method for measuring the casting blank speed through an industrial CCD based on AI, belongs to the technical field of casting blank speed measurement, and adopts scientific and technological means such as a computer technology, a laser technology and an information and image collecting and processing technology to measure the casting blank pulling speed. According to the basic principle, based on the following performance of mark features in a view field on casting blank movement and the selective transmission effect of an optical filter, an optical imaging sequence containing position information of the mark features at different moments is recorded through an imaging system, then the average displacement of the mark features is determined through an image processing and tracking positioning algorithm, and then the casting blank pulling speed is obtained. The technology mainly comprises the following steps of image partitioning, template correlation analysis, image matching and casting blank pulling speed calculation. According to the casting blank pulling speed measuring method, the measuring precision is improved, the operation process is greatly simplified, the skill requirement for operators is lowered, and high practical value and application prospects are achieved.

Description

Technical Field

[0001] The present invention relates to the technical field of billet speed measurement, and in particular to a method for accurately measuring billet speed by combining industrial CCD measurement with an AI model. Background Art

[0002] In the prior art, the measurement of billet speed mainly relies on traditional contact or non-contact methods.

[0003] Contact measurement, such as using a measuring roller, can cause wear due to direct contact with the billet, thus affecting measurement accuracy. Furthermore, this method is susceptible to surface conditions such as temperature, humidity, roughness, and slippage, all of which can interfere with measurement results.

[0004] Non-contact measurement methods, such as those using a straightening machine inverter to calculate speed, avoid direct contact with the ingot, but their accuracy is often affected by factors such as inverter performance, sensor accuracy, and signal transmission. Furthermore, these methods are susceptible to interference from various environmental factors at the production site, such as electromagnetic interference and vibration, which can reduce the accuracy of measurement results.

[0005] Especially in harsh environments such as high temperature, high humidity, and dust, the measurement accuracy and stability of existing technologies are even more difficult to guarantee.

[0006] Therefore, there is an urgent need in this field for a more accurate, stable and adaptable method for measuring the casting speed.

[0007] The information disclosed in this background technology section is only intended to enhance understanding of the overall background of the invention and should not be regarded as an admission or any form of suggestion that the information constitutes the prior art already known to a person skilled in the art. Summary of the Invention

[0008] The purpose of the present invention is to provide a more accurate, stable and adaptable method for measuring the casting speed.

[0009] To achieve the above object, the present invention provides the following solutions:

[0010] A method for measuring billet speed using an industrial CCD based on AI, comprising:

[0011] CCD collects images of the casting surface;

[0012] Divide the image into blocks;

[0013] Extract several small rectangular blocks of equal length and width from the entire image as templates. Use template matching to register them in the next frame of the image and determine the displacement of the template area in a specified time interval to obtain the motion speed of the billet. Specifically, the following steps are performed:

[0014] Read in two frames of images and extract each particle area for displacement calculation;

[0015] Calculate the mutual matrix using Fourier transform;

[0016] Peak detection;

[0017] Displacement calculation of selected calibration feature area;

[0018] Determine whether the displacement of each area is determined. If so, process the displacement data and calculate the casting speed;

[0019] If not, return to the step of calculating the correlation matrix using Fourier transform.

[0020] Optionally, the template matching is based on the grayscale cross-correlation registration method, assuming that the size of the template image T is M x ×M y , the search window S is of size N x ×N y When matching, the template image is superimposed on the reference image and translated. The search image covered by the template is called the sub-image S i,j , i, j are the coordinates of the lower left corner of the sub-image in the S image, which are called reference points. The value range of i, j is 0≤i, j≤NM. Compare T and S i,j If the two are consistent, then T and S i,j The difference is zero.

[0021] Optionally, the window size is determined by:

[0022] First, a higher frame rate is used to make a preliminary estimate of the pulling speed. Then, the size of the calculation correlation window is determined based on the relationship between the CCD sampling time and the image and the actual distance.

[0023] Optionally, image matching includes:

[0024] The normalized cross-correlation method is used to calculate the similarity between the marked features of the two frames. The degree of matching is determined by calculating the cross-correlation between the template image and the feature matching image, including:

[0025] First, the calculation area of ​​the image is normalized and preprocessed, and then the discrete cross-correlation method is used to calculate the best matching position;

[0026] The correlation function matrix is ​​obtained by inversely transforming the convolution result of Fourier transform, and then the displacement vector is determined by peak detection. The calculation of the absolute value of the displacement is similar to the slab width measurement. The instantaneous velocity of the casting is calculated based on the known CCD sampling time and speed definition.

[0027] Compared with the prior art, the present invention has the following beneficial effects:

[0028] The present invention measures billet motion by determining the position of a marker feature in two consecutive image frames. CCD imaging is used to record billet surface images bearing the marker feature, identify the marker feature, and perform cross-correlation processing on the marker features in the two consecutive image frames to calculate the velocity vector, ultimately achieving billet casting speed measurement.

[0029] Compared with traditional methods, the present invention significantly improves measurement accuracy and real-time performance. Traditional methods often rely on manual observation or use relatively rough sensors, resulting in large fluctuations in measurement data and insufficient stability. The present invention utilizes high-precision CCD for image capture, combined with advanced image processing algorithms, to accurately identify marker features and achieve accurate calculation of displacement. In addition, the image matching algorithm of the present invention can operate stably under complex backgrounds and is not affected by factors such as changes in illumination and contamination of the billet surface, thereby ensuring the reliability of the measurement results. In terms of superiority, the billet casting speed measurement method of the present invention not only improves measurement accuracy, but also greatly simplifies the operating process, reduces the skill requirements for operators, and has high practical value and application prospects. BRIEF DESCRIPTION OF THE DRAWINGS

[0030] In order to more clearly illustrate the embodiments of the present invention or the technical solutions in the prior art, the following briefly introduces the drawings required for use in the embodiments. Obviously, the drawings described below are only some embodiments of the present invention. For ordinary technicians in this field, other drawings can be obtained based on these drawings without paying any creative work.

[0031] Figure 1 A schematic diagram of the casting billet casting speed measurement process provided in an embodiment of the present invention.

[0032] Figure 2 A schematic diagram of a rectangular small block template provided in an embodiment of the present invention.

[0033] Figure 3 A schematic diagram of an image matching algorithm provided by an embodiment of the present invention. DETAILED DESCRIPTION

[0034] The following will clearly and completely describe the technical solutions in the embodiments of the present invention in conjunction with the accompanying drawings. 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 making creative efforts are within the scope of protection of the present invention.

[0035] The purpose of the present invention is to provide a more accurate, stable and adaptable method for measuring the casting speed.

[0036] In order to make the above-mentioned objects, features and advantages of the present invention more obvious and easy to understand, the present invention is further described in detail below with reference to the accompanying drawings and specific embodiments.

[0037] Example 1:

[0038] The present invention provides a method for measuring billet speed using an industrial CCD based on AI, such as Figure 1 Shown, including:

[0039] CCD collects images of the casting surface;

[0040] Divide the image into blocks;

[0041] Extract several small rectangular blocks of equal length and width from the entire image as templates. Use template matching to register them in the next frame of the image and determine the displacement of the template area in a specified time interval to obtain the motion speed of the billet. Specifically, the following steps are performed:

[0042] Read in two frames of images and extract each particle area for displacement calculation;

[0043] Calculate the mutual matrix using Fourier transform;

[0044] Peak detection;

[0045] Displacement calculation of selected calibration feature area;

[0046] Determine whether the displacement of each area is determined. If so, process the displacement data and calculate the casting speed;

[0047] If not, return to the step of calculating the correlation matrix using Fourier transform.

[0048] This embodiment utilizes computer technology, laser technology, and information and image acquisition and processing techniques to measure slab casting speed. The basic principle is based on the ability of marker features in the field of view to follow the motion of the slab and the selective transmission effect of filters. An imaging system records an optical image sequence containing the position of the marker features at different times. Image processing and tracking algorithms are then used to determine the average displacement of the marker features, thereby obtaining the casting speed. This technology primarily involves the following steps: image segmentation, correlation analysis templates, image matching, and casting speed calculation.

[0049] (1) Image segmentation:

[0050] The thermal image of the billet surface collected by CCD contains a large number of randomly generated marker features with obvious characteristic information. However, since correlation analysis of the entire image will result in a huge consumption of time and is not conducive to real-time measurement, it is proposed to adopt the method of image segmentation. The specific method is: extract a number of small rectangular blocks of equal length and width from the entire image as templates, and use the template matching method to align them in the next frame of the image to determine the displacement of the template area in the specified time interval, thereby obtaining the motion speed of the billet. In the calculation, it is assumed that each small rectangular block can represent the current motion speed of the billet. Considering the situation of scaly marker features, the use of a single window matching may cause large fluctuations in the displacement calculation. Therefore, it is necessary to perform statistical analysis on the displacement calculation results of multiple template blocks to determine the maximum credible displacement. Then, the billet pulling speed is calculated to improve the measurement accuracy.

[0051] (2) Correlation analysis template:

[0052] Template matching is based on the grayscale cross-correlation registration method, the principle of which is as follows Figure 2 As shown, let the template image T (reference image) size be M x ×M y , the search window S (reference image) size is N x ×N y When matching, the template image is superimposed on the reference image and translated. The search image covered by the template is called the sub-image S i,j , i, j are the coordinates of the lower left corner of the sub-image in the S image, which is called the reference point. It is not difficult to know that the value range of i, j is 0≤i,j≤NM. Compare T and S i,j If the two are consistent, then T and S i,j The difference is zero.

[0053] (3) Image matching:

[0054] This technology uses the normalized cross-correlation method to calculate the similarity between the marked features of two frames of images, and determines the degree of matching by calculating the cross-correlation value between the template image and the feature matching image.

[0055] During the measurement of the billet casting speed, the surface temperature of the billet will drop to a certain extent at different times as it moves. In order to reduce the impact of changes in image grayscale values ​​on related measurement results, the normalized cross-correlation matching criterion is adopted in the calculation. First, the calculation area of ​​the image is normalized and preprocessed, and then the discrete cross-correlation method is used to calculate the best matching position.

[0056] Fast Fourier transform (FFT) increases the operation speed. If the discrete mutual method is used directly through the formula:

[0057]

[0058] The cross-correlation operation is performed on the normalized image area, and the order of magnitude of the operation can reach O[N 4 ]. Considering the cross-correlation theory, the cross-correlation operation in the spatial domain is equivalent to the convolution operation in the frequency domain, which can reduce the order of magnitude of the operation and simplify the operation. This means that a two-dimensional Fourier transform is performed on each recognition window. This operation will reduce the order of magnitude of the operation from O[N 4 ] is reduced to O[N 2 ], the algorithm diagram is as follows Figure 3 As shown, Fourier transform is required. The correlation function matrix is ​​obtained by inverse transforming the convolution result of Fourier transform. Then, the displacement vector is determined by peak detection. The calculation of the absolute value of displacement is the same as that of slab width measurement. The instantaneous velocity of the slab is calculated based on the known CCD sampling time and speed definition.

[0059] The calculation of the casting speed adopts the image correlation method to calculate the displacement vector of the casting image, and then realizes the conversion from displacement to casting speed according to the definition of speed. The algorithm flow is as follows: Figure 1 shown.

[0060] During the billet drawing speed measurement process, the billet's trajectory remains unchanged, only its speed changes. Simultaneously, a large number of randomly generated marker features (such as iron oxide scale) are distributed on the billet's surface and move synchronously with the billet. These marker features themselves are stable within the measurement system's field of view. Therefore, billet motion measurement is achieved by determining the position of these marker features in two consecutive image frames. CCD imaging is used to record images of the billet surface with these marker features, identify these marker features, and perform cross-correlation processing on the two consecutive image frames to calculate the velocity vector, ultimately achieving billet drawing speed measurement.

[0061] The various embodiments in this specification are described in a progressive manner, with each embodiment focusing on the differences from other embodiments. Reference can be made to the common and similar parts between the various embodiments. For the systems disclosed in the embodiments, since they correspond to the methods disclosed in the embodiments, the description is relatively simple, and the relevant parts can be referred to the method description.

[0062] This document uses specific examples to illustrate the principles and implementation methods of the present invention. The above examples are only intended to help understand the method and core concept of the present invention. At the same time, those skilled in the art will find that the specific implementation methods and application scopes may vary based on the concept of the present invention. In summary, the contents of this specification should not be construed as limiting the present invention.

Claims

1. A method for measuring billet speed using an industrial CCD based on AI, characterized in that: include: CCD collects images of the casting surface; Divide the image into blocks; Extract several small rectangular blocks of equal length and width from the entire image as templates. Use template matching to register them in the next frame of the image and determine the displacement of the template area in a specified time interval to obtain the motion speed of the billet. Specifically, the following steps are performed: Read in two frames of images and extract each particle area for displacement calculation; Calculate the mutual matrix using Fourier transform; Peak detection; Displacement calculation of selected calibration feature area; Determine whether the displacement of each area is determined. If so, process the displacement data and calculate the casting speed; If not, return to the step of calculating the correlation matrix using Fourier transform.

2. The method for measuring billet speed using an AI-based industrial CCD according to claim 1, characterized in that: The template matching is based on the grayscale cross-correlation registration method, assuming that the size of the template image T is M x ×M y , the search window S is of size N x ×N y When matching, the template image is superimposed on the reference image and translated. The search image covered by the template is called the sub-image S i,j , i, j are the coordinates of the lower left corner of the sub-image in the S image, which are called reference points. The value range of i, j is 0≤i, j≤NM. Compare T and S i,j If the two are consistent, then T and S i,j The difference is zero.

3. The method for measuring casting speed using an AI-based industrial CCD according to claim 2, characterized in that: The method for determining the window size is: First, a higher frame rate is used to make a preliminary estimate of the pulling speed. Then, the size of the calculation correlation window is determined based on the relationship between the CCD sampling time and the image and the actual distance.

4. The method for measuring billet speed using an AI-based industrial CCD according to claim 1, wherein: Image matching includes: The normalized cross-correlation method is used to calculate the similarity between the marked features of the two frames. The degree of matching is determined by calculating the cross-correlation between the template image and the feature matching image, including: First, the calculation area of ​​the image is normalized and preprocessed, and then the discrete cross-correlation method is used to calculate the best matching position; The correlation function matrix is ​​obtained by inversely transforming the convolution result of Fourier transform, and then the displacement vector is determined by peak detection. The calculation of the absolute value of the displacement is similar to the slab width measurement. The instantaneous velocity of the casting is calculated based on the known CCD sampling time and speed definition.

Images