Bobbin identification system and method based on laser vision

Abstract

The invention provides a bobbin identification system and method based on laser vision. According to the bobbin recognition system and method based on laser vision, the system comprises a laser projection device, a visual camera, an image processing module, a recognition control unit, a database and a model library, and the laser projection device forms a high-contrast laser mark on the surface of a bobbin by emitting a laser beam with the wavelength of 5nm. According to the bobbin recognition system and method based on laser vision, the laser projection device, the visual camera, the image processing module and the recognition control unit are introduced, and the response speed and accuracy of the system when multiple bobbins enter the recognition range at the same time are remarkably improved. According to the method, a parallel computing resource dynamic allocation mode is adopted, and the processing speed and the recognition accuracy can be effectively balanced in a complex background or a high-dynamic scene.

Description

Technical Field

[0001] The present invention relates to the technical field of thread bobbin recognition based on laser vision, and in particular to a thread bobbin recognition system and method based on laser vision. Background Art

[0002] The bobbin recognition system is mainly composed of a laser projection device, a visual camera, an image processing module and a recognition control unit. When the system is working, the laser projection device irradiates the bobbin or empty bobbin with a laser beam to form a clear laser mark. The visual camera captures the image after laser irradiation and transmits the image to the image processing module. The image processing module analyzes the collected image and identifies the specific position, type (whether it is an empty bobbin) and surface state of the bobbin by extracting feature points and contour information. The system uses the high precision of laser measurement combined with image processing technology to achieve accurate recognition and classification of bobbins. The whole process is based on the combination of the physical recognition principle of laser radar and the graphic comparison principle of visual images.

[0003] Although the system has certain advantages in accurately identifying thread reels, it still has some shortcomings. The system's response speed may be slow when dealing with complex backgrounds or high-dynamic scenes, especially when multiple thread reels enter the recognition range at the same time, and there is a problem in the balance between processing speed and accuracy. Summary of the invention

[0004] In view of the deficiencies in the prior art, the present invention provides a wire bobbin recognition system and method based on laser vision, which solves the problem of balancing the processing speed and accuracy when multiple wire bobbins enter the recognition range at the same time.

[0005] To achieve the above objectives, the present invention is implemented through the following technical solutions: a bobbin identification system and method based on laser vision, comprising:

[0006] The system includes a laser projection device, a visual camera, an image processing module, a recognition control unit, a database and a model library;

[0007] The laser projection device forms a high-contrast laser mark on the surface of the bobbin by emitting a laser beam with a wavelength of λ;

[0008] The visual camera captures the image of the marked bobbin in real time and transmits the image data to the image processing module for analysis;

[0009] The image processing module is based on the feature extraction formula of the convolutional neural network:

[0010]

[0011] Among them, f(x) is the extracted image feature, wi is the weight parameter, xi is the pixel value of the input image, b is the bias value, and σ is the activation function;

[0012] The image processing module calculates the similarity between feature vectors using the Euclidean distance formula:

[0013]

[0014] Among them, xi and y i are the corresponding components of the two feature vectors, and d is the similarity measure used to determine whether they are the same type of bobbins;

[0015] The identification control unit identifies the type and state of the bobbin according to the calculation results and the data in the model library; through the dynamic allocation of parallel computing resources, it is ensured that when multiple bobbins enter the identification range at the same time, the system response speed and accuracy are balanced; the resource allocation formula for parallel computing is:

[0016]

[0017] Among them, Rt is the resource allocated to each recognition task, T{total} is the total available resources, Nt is the number of bobbins currently entering the recognition range, and α is the resource allocation adjustment coefficient.

[0018] Preferably, the laser projection device can automatically adjust the brightness of the laser beam according to the external light conditions to adapt to different working environments. The emission angle θ of the laser projection device is calibrated in real time by the laser radar. The relationship between the laser emission angle and the laser distance is:

[0019]

[0020] Where L is the laser distance, h is the height of the bobbin, and θ is the laser emission angle.

[0021] Preferably, the visual camera is a high-resolution camera capable of capturing the bobbin in high-speed motion. The image processing module uses Fourier transform F(u,v) to convert the image from the spatial domain to the frequency domain to reduce background noise interference. The formula is:

[0022]

[0023] in, f (x, y) is the spatial domain pixel value of the image, u, v is the coordinate in the frequency domain, the system uses GPU acceleration technology for parallel calculation, when multiple bobbins enter the recognition range, the real-time recognition is ensured, the image processing module combines the light intensity distribution formula = 1_0\codot e ∧{-\mux} compensates for the light attenuation of different surface materials, I0 is the initial light intensity, μ is the material absorption coefficient, and x is the laser propagation distance.

[0024] Preferably, the identification control unit dynamically adjusts the processing strategy by calculating the task time Ts and the processing time Tp, and the calculation formula is:

[0025]

[0026] Where Ts is the processing time, Nt is the number of bobbins, and P is the parallel processing capability.

[0027] Preferably, the system uses a deep learning algorithm to enhance the recognition ability of the model using existing data.

[0028] Preferably, the deep learning algorithm adopts a transfer learning method to identify and classify the bobbin types through an existing image recognition model library.

[0029] Preferably, the system supports adaptive updating of the model library to continuously optimize the recognition and classification accuracy of the bobbins. The system also includes an early warning module that can automatically send an alarm signal when an abnormal bobbin is detected during the recognition process.

[0030] A method for identifying a bobbin based on laser vision, comprising:

[0031] a. A laser beam with a wavelength of λ is emitted to the surface of the wire barrel through a laser projection device, forming a high-contrast laser mark on the surface of the wire barrel;

[0032] b. Use a visual camera to capture the image of the thread barrel after laser marking in real time, and transmit the image data to the image processing module;

[0033] c. The image processing module processes the collected image data based on the convolutional neural network algorithm;

[0034] d. The image processing module calculates the similarity between feature vectors using the Euclidean distance formula to determine whether they are the same type of bobbins;

[0035] e. The recognition control unit identifies the state and type of the bobbin based on the feature similarity calculation results and combines them with the data in the model library to determine whether it is an empty bobbin.

[0036] The present invention provides a bobbin identification system and method based on laser vision, which has the following beneficial effects:

[0037] This wire bobbin recognition system and method based on laser vision, by introducing a laser projection device, a visual camera, an image processing module and a recognition control unit, significantly improves the response speed and accuracy of the system when multiple wire bobbins enter the recognition range at the same time. The present invention adopts a method of dynamic allocation of parallel computing resources, which can effectively balance the processing speed and recognition accuracy under complex backgrounds or high dynamic scenes. By introducing GPU acceleration technology, the system realizes efficient parallel computing, ensuring a high processing speed when multiple wire bobbins enter the recognition range at the same time, solving the problem of slow processing speed in the prior art. At the same time, the laser projection device can automatically adjust the brightness of the laser beam according to the external light conditions, ensuring the stable operation of the system in different lighting environments, and having stronger adaptability.

[0038] Another significant advantage of the present invention is its high-precision recognition capability. The system uses a convolutional neural network to extract features from the captured image data and calculates the similarity between image features using the Euclidean distance formula, thereby accurately determining the state and type of the bobbin. The system combines deep learning algorithms and transfer learning methods to continuously optimize and update the model library, further enhancing the accuracy of bobbin recognition. In addition, the image processing module converts the image from the spatial domain to the frequency domain through Fourier transform, effectively reducing background noise interference and improving recognition effects under complex backgrounds. The system is also equipped with an early warning module that can automatically detect abnormal bobbins and send alarms during the recognition process to ensure the safety and reliability of the production process. BRIEF DESCRIPTION OF THE DRAWINGS

[0039] Figure 1 It is a schematic diagram of the process of the present invention. DETAILED DESCRIPTION

[0040] 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.

[0041] like Figure 1 As shown, the embodiment of the present invention provides a bobbin recognition system and method based on laser vision, including: the system includes a laser projection device, a visual camera, an image processing module, a recognition control unit, a database and a model library, the laser projection device can automatically adjust the brightness of the laser beam according to the external light conditions to adapt to different working environments, the emission angle θ of the laser projection device is calibrated in real time by a laser radar, and the relationship between the laser emission angle and the laser distance is:

[0042]

[0043] Where L is the laser distance, h is the height of the bobbin, and θ is the laser emission angle.

[0044] The laser projection device forms a high-contrast laser mark on the surface of the bobbin by emitting a laser beam with a wavelength of λ;

[0045] The visual camera captures the image of the marked bobbin in real time and transmits the image data to the image processing module for analysis. The visual camera is a high-resolution camera that can capture the bobbin in high-speed motion. The image processing module uses Fourier transform F(u,v) to convert the image from the spatial domain to the frequency domain to reduce background noise interference. The formula is:

[0046]

[0047] in, f (x, y) is the spatial domain pixel value of the image, u, v is the coordinate in the frequency domain, the system uses GPU acceleration technology for parallel calculation, when multiple bobbins enter the recognition range, it ensures the real-time recognition, the image processing module combines the light intensity distribution formula = 1_0\codot e ∧ {-\mux} compensates for the light attenuation of different surface materials, I0 is the initial light intensity, μ is the material absorption coefficient, and x is the laser propagation distance. The feature extraction formula of the image processing module based on the convolutional neural network is:

[0048]

[0049] Among them, f(x) is the extracted image feature, wi is the weight parameter, xi is the pixel value of the input image, b is the bias value, and σ is the activation function.

[0050] The image processing module calculates the similarity between feature vectors using the Euclidean distance formula:

[0051]

[0052] Among them, xi and y i are the corresponding components of the two eigenvectors, and d is the similarity measure used to determine whether they are the same type of bobbins.

[0053] The recognition control unit identifies the type and status of the bobbin based on the calculation results and the data in the model library. By dynamically allocating parallel computing resources, it ensures that the system response speed and accuracy are balanced when multiple bobbins enter the recognition range at the same time. The resource allocation formula for parallel computing is:

[0054]

[0055] Among them, Rt is the resource allocated for each recognition task, T{total} is the total available resources, Nt is the number of bobbins currently entering the recognition range, and α is the resource allocation adjustment coefficient. The recognition control unit dynamically adjusts the processing strategy by calculating the task time Ts and the processing time Tp. The calculation formula is:

[0056]

[0057] Among them, Ts is the processing time, Nt is the number of bobbins, and P is the parallel processing capability. The system uses a deep learning algorithm to enhance the recognition ability of the model using existing data. The deep learning algorithm uses a transfer learning method to identify and classify bobbin types through an existing image recognition model library. The system supports adaptive updating of the model library to continuously optimize the recognition and classification accuracy of bobbins. The system also includes an early warning module that can automatically send an alarm signal when an abnormal bobbin is detected during the recognition process.

[0058] A method for identifying a bobbin based on laser vision, comprising:

[0059] a. A laser beam with a wavelength of λ is emitted to the surface of the bobbin through a laser projection device, forming a high-contrast laser mark on the surface of the bobbin.

[0060] b. Use a visual camera to capture the image of the thread barrel after laser marking in real time and transmit the image data to the image processing module.

[0061] c. The image processing module processes the collected image data based on the convolutional neural network algorithm.

[0062] d. The image processing module calculates the similarity between feature vectors using the Euclidean distance formula to determine whether they are bobbins of the same type.

[0063] e. The recognition control unit identifies the state and type of the bobbin based on the feature similarity calculation results and combines them with the data in the model library to determine whether it is an empty bobbin.

[0064] 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. A bobbin recognition system based on laser vision, characterized in that: include: The system includes a laser projection device, a visual camera, an image processing module, a recognition control unit, a database and a model library; The laser projection device forms a high-contrast laser mark on the surface of the bobbin by emitting a laser beam with a wavelength of λ; The visual camera captures the image of the marked bobbin in real time and transmits the image data to the image processing module for analysis; The image processing module is based on the feature extraction formula of the convolutional neural network: Among them, f(x) is the extracted image feature, wi is the weight parameter, xi is the pixel value of the input image, b is the bias value, and σ is the activation function; The image processing module calculates the similarity between feature vectors using the Euclidean distance formula: Among them, xi and y i are the corresponding components of the two feature vectors, and d is the similarity measure used to determine whether they are the same type of bobbins; The identification control unit identifies the type and state of the bobbin according to the calculation results and the data in the model library; through the dynamic allocation of parallel computing resources, it is ensured that when multiple bobbins enter the identification range at the same time, the system response speed and accuracy are balanced; the resource allocation formula for parallel computing is: Among them, Rt is the resource allocated to each recognition task, T{total} is the total available resources, Nt is the number of bobbins currently entering the recognition range, and α is the resource allocation adjustment coefficient.

2. The laser vision-based bobbin identification system according to claim 1, characterized in that: The laser projection device can automatically adjust the brightness of the laser beam according to the external light conditions to adapt to different working environments. The emission angle θ of the laser projection device is calibrated in real time by the laser radar. The relationship between the laser emission angle and the laser distance is: Where L is the laser distance, h is the height of the bobbin, and θ is the laser emission angle.

3. The laser vision-based bobbin identification system according to claim 1, characterized in that: The visual camera is a high-resolution camera that can capture the bobbin in high-speed motion. The image processing module uses Fourier transform F(u,v) to convert the image from the spatial domain to the frequency domain to reduce background noise interference. The formula is: in, f (x, y) is the spatial domain pixel value of the image, u, v is the coordinate in the frequency domain, the system uses GPU acceleration technology for parallel calculation, when multiple line barrels enter the recognition range, the real-time recognition is ensured, the image processing module combines the light intensity distribution formula = 1_0\codot e ∧ {-\mux} compensates for the light attenuation of different surface materials, I0 is the initial light intensity, μ is the material absorption coefficient, and x is the laser propagation distance.

4. The laser vision-based bobbin identification system according to claim 1, characterized in that: The identification control unit dynamically adjusts the processing strategy by calculating the task time Ts and the processing time Tp. The calculation formula is: Where Ts is the processing time, Nt is the number of bobbins, and P is the parallel processing capability.

5. The laser vision-based bobbin identification system according to claim 1, characterized in that: The system uses deep learning algorithms to enhance the recognition capabilities of the model using existing data.

6. The laser vision-based bobbin identification system according to claim 1, characterized in that: The deep learning algorithm adopts a transfer learning method to identify and classify bobbin types through an existing image recognition model library.

7. The laser vision-based bobbin identification system according to claim 1, characterized in that: The system supports adaptive updating of the model library to continuously optimize the recognition and classification accuracy of the bobbins. The system also includes an early warning module that can automatically send an alarm signal when an abnormal bobbin is detected during the recognition process.

8. A method for identifying bobbins based on laser vision, characterized in that: include: a. A laser beam with a wavelength of λ is emitted to the surface of the wire barrel through a laser projection device, forming a high-contrast laser mark on the surface of the wire barrel; b. Use a visual camera to capture the image of the thread barrel after laser marking in real time, and transmit the image data to the image processing module; c. The image processing module processes the collected image data based on the convolutional neural network algorithm; d. The image processing module calculates the similarity between feature vectors using the Euclidean distance formula to determine whether they are the same type of bobbins; e. The recognition control unit identifies the state and type of the bobbin based on the feature similarity calculation results and combines them with the data in the model library to determine whether it is an empty bobbin.

Images