Intelligent identification method for solid wood board defects based on 5G+ industrial internet

Abstract

The application discloses a kind of based on 5G+ industrial internet's solid wood board flaw intelligent identification method, the method comprises: obtaining first board image;The first board image is carried out denoising quality recovery processing, and target image is obtained;Through the flaw identification neural network model of trained good identification target image flaw and are labeled, and labeled image is obtained;The labeled image is returned to production line management system.The application is carried out denoising quality recovery processing to board image, then neural network is used to identify flaw and label, realizes the flaw identification technology of non-artificial solid wood board, improves the automation intelligent level of solid wood furniture production line.

Description

Technical Field

[0001] This invention relates to the field of image processing technology, and in particular to a method, device, terminal and storage medium for intelligent identification of defects in solid wood boards based on 5G+Industrial Internet. Background Technology

[0002] As living standards improve, solid wood furniture is becoming increasingly popular. Industrialized production of solid wood furniture is an inevitable way to move away from inefficient, workshop-style production and improve furniture quality. With the rise of 5G and the Industrial Internet of Things, automated and intelligent solid wood furniture production lines are becoming a development trend.

[0003] To ensure the quality of solid wood furniture and improve the utilization efficiency of solid wood raw materials, the first step is to screen the solid wood boards, especially to identify and mark the defects on the solid wood boards, including cracks and knots, which lays the foundation for the next process.

[0004] Traditionally, the selection of materials for solid wood furniture has relied primarily on manual labor. This increases labor costs and reduces the efficiency of solid wood furniture production lines, while also presenting challenges such as the difficulty in standardizing the accuracy of defect identification. Currently, there are no relevant technologies and systems to solve the problem of non-manual identification of defects in solid wood furniture boards, thus hindering the automation and intelligentization of solid wood furniture production.

[0005] Therefore, existing technologies still need to be improved and enhanced. Summary of the Invention

[0006] To address the aforementioned shortcomings of existing technologies, this invention provides an intelligent identification method for solid wood board defects based on 5G+Industrial Internet, aiming to solve the problem of inaccurate super-resolution image quality evaluation results in existing technologies.

[0007] To solve the above-mentioned technical problems, the technical solution adopted by the present invention is as follows:

[0008] In a first aspect, the present invention provides a method for intelligent identification of defects in solid wood panels based on 5G+Industrial Internet, the method comprising:

[0009] Obtain the image of the first board material;

[0010] The first plate image is subjected to noise reduction and quality restoration processing to obtain the target image;

[0011] The defect recognition neural network model is trained to identify and annotate defects in the target image, resulting in an annotated image;

[0012] The labeled image is then transmitted back to the production line management system.

[0013] The aforementioned intelligent defect recognition method for solid wood panels based on 5G+Industrial Internet, wherein, before acquiring the first panel image, includes:

[0014] Obtain the original image of the board material;

[0015] The original image of the board material is preprocessed to obtain the first image of the board material.

[0016] The aforementioned intelligent defect recognition method for solid wood panels based on 5G+Industrial Internet, wherein the denoising and quality restoration processing of the first panel image includes:

[0017] The target image is obtained by solving the objective optimization problem iteratively.

[0018] The objective optimization problem is:

[0019]

[0020] Where, the optimal solution of f at the end of the iteration is the target image, the initial value of g is the first board image, the first board image is an n×n board image, and α i The positive parameters are greater than 0, z and w are intermediate parameters for solving the problem, and D... i f is the matrix form of TV(f), β and γ are adjustment parameters, H is the degeneracy calculation, and μ is the positive factor;

[0021] The formula for calculating TV(f) is:

[0022]

[0023] Among them, D x and D y These represent the forward finite difference calculation formulas in the horizontal and vertical directions, respectively, σ x and σ y For the constant of its corresponding direction, [D x f] i For vector D x The i-th term of f, [D] y f] i For vector D y The i-th term of f.

[0024] The aforementioned intelligent defect identification method for solid wood panels based on 5G+Industrial Internet, wherein the iterative solution of the objective optimization problem includes:

[0025] Initialize, input H, μ>0, β, γ and {α i >0, i=1, K,n 2 The value of}, input the initial value of g. 0That is, the image of the first board material, f 0 =g 0 Iterative calculation of count k=0;

[0026] In the (k+1)th calculation:

[0027] Limit f to f k The optimal solution w for variable w is calculated based on the first formula. * Reassign w k+1 =w * ;

[0028] Limit f to f k The optimal solution z for variable z is calculated based on the second formula. * Reassign z k+1 =z * ;

[0029] Limit w k+1 =w * and z k+1 =z * The optimal solution f for variable f in this iteration is calculated based on the third formula. * Reassign f k+1 =f * ;

[0030] Determine whether variable f satisfies the convergence condition. If it does, terminate the calculation; otherwise, continue the calculation until the convergence condition is met.

[0031] The aforementioned intelligent defect identification method for solid wood panels based on 5G+Industrial Internet, wherein the first formula is:

[0032]

[0033] The second formula is:

[0034]

[0035] Where o is the point-to-point product, and the sign function sign(Hf-g) is:

[0036]

[0037] The intelligent defect identification method for solid wood boards based on 5G+Industrial Internet, wherein the third formula is:

[0038]

[0039] The aforementioned intelligent identification method for defects in solid wood boards based on 5G+Industrial Internet, wherein the labeled image includes the number, location coordinates, range, and type of defects in the target image.

[0040] A second aspect of the present invention provides a smart defect identification device for solid wood panels based on 5G+Industrial Internet, comprising:

[0041] Image acquisition module, the image acquisition module is used to acquire a first board image;

[0042] A noise reduction and quality restoration processing module is used to perform noise reduction and quality restoration processing on the first board image to obtain a target image;

[0043] The defect annotation module is used to identify and annotate defects in the target image using a trained defect recognition neural network model, thereby obtaining an annotated image.

[0044] A transmission module is used to transmit the labeled image back to the production line management system.

[0045] A third aspect of the present invention provides a terminal, the terminal including a processor and a computer-readable storage medium communicatively connected to the processor, the computer-readable storage medium being adapted to store a plurality of instructions, the processor being adapted to invoke the instructions in the computer-readable storage medium to execute the steps of implementing the intelligent identification method for solid wood board defects based on 5G+Industrial Internet as described in any of the above claims.

[0046] In a fourth aspect, the present invention provides a computer-readable storage medium storing one or more programs that can be executed by one or more processors to implement the steps of the intelligent identification method for solid wood board defects based on 5G+Industrial Internet as described in any of the preceding claims.

[0047] Compared with existing technologies, this invention provides a method for intelligent identification of defects in solid wood panels based on 5G+Industrial Internet. The method includes: acquiring a first image of the wood panel; performing denoising and quality restoration processing on the first image to obtain a target image; identifying and annotating defects in the target image using a trained defect recognition neural network model to obtain an annotated image; and transmitting the annotated image back to the production line management system. This invention achieves non-manual defect identification technology for solid wood panels by performing denoising and quality restoration processing on the panel image and then using a neural network to identify and annotate defects, thereby improving the automation and intelligence level of solid wood furniture production lines. Attached Figure Description

[0048] Figure 1 A flowchart illustrating an embodiment of the intelligent defect identification method for solid wood panels based on 5G+Industrial Internet provided by the present invention;

[0049] Figure 2 This is an overall framework diagram of an embodiment of the intelligent identification method for defects in solid wood panels based on 5G+Industrial Internet provided by the present invention.

[0050] Figure 3 A scenario flowchart illustrating an embodiment of the intelligent defect identification method for solid wood panels based on 5G+Industrial Internet provided by the present invention;

[0051] Figure 4 This is a noise reduction and quality recovery processing diagram of an embodiment of the intelligent identification method for defects in solid wood panels based on 5G+Industrial Internet provided by the present invention.

[0052] Figure 5 A schematic diagram of the structural principle of an embodiment of the intelligent identification device for defects in solid wood boards based on 5G+Industrial Internet provided by the present invention;

[0053] Figure 6 A schematic diagram illustrating the principle of an embodiment of the terminal provided by the present invention;

[0054] Figure 7 A flowchart illustrating an embodiment of the terminal provided by the present invention. Detailed Implementation

[0055] To make the objectives, technical solutions, and effects of this invention clearer and more explicit, the invention will be further described in detail below with reference to the accompanying drawings and embodiments. It should be understood that the specific embodiments described herein are merely illustrative of the invention and are not intended to limit the invention.

[0056] The intelligent identification method for solid wood board defects based on 5G+Industrial Internet provided by this invention can be applied to terminals with computing capabilities. The terminal can execute the intelligent identification method for solid wood board defects based on 5G+Industrial Internet provided by this invention to obtain the quality evaluation results of the target super-resolution image. The terminal can be, but is not limited to, various computers, mobile terminals, smart home appliances, wearable devices, etc.

[0057] Example 1

[0058] like Figure 1 As shown, one embodiment of the intelligent defect identification method for solid wood boards based on 5G+Industrial Internet includes the following steps:

[0059] S100, Obtain the image of the first board material.

[0060] Before acquiring the first board image, the process includes:

[0061] S110. Obtain the original image of the board material;

[0062] S120. Preprocess the original image of the board material to obtain the first board material image.

[0063] Reference Figure 2 and Figure 3 In this embodiment, the steps of acquiring the original image of the board and preprocessing the original image of the board can be performed in the production line management system.

[0064] Specifically, the production line management system includes a real-time imaging system. The cameras in this system are installed 1-2 meters above the placement positions of the solid wood panels during the solid wood furniture production line processing steps. The real-time imaging system features automatic adjustable 10x zoom, a resolution of at least 1080P, a frame rate of at least 50 frames per second, a dynamic range greater than 80dB, automatic reporting, automatic gain control, and white balance capabilities. It is also Python programmable, has a gimbal function, and can remotely and automatically control the shooting angle. This results in clearer and more comprehensive original images of the panels.

[0065] The real-time imaging system is used to acquire the original image of the board material.

[0066] After the camera acquires the original image of the board material, the original image of the board material is input into the embedded image preprocessing system to adjust the image angle and optimize the brightness, while also performing size cropping. Finally, the image data is compressed to obtain the first image of the board material.

[0067] Specifically, the performance requirements of the embedded image preprocessing system are as follows: CPU no less than 2GHz (quad-core), GPU no less than 256-core NVIDIA Maxwell GPU, memory no less than 8GB 64-bit LPDDR4, support for HDMI and USB 3.0, onboard storage of no less than 128GB solid-state drive, and support for a 5G data transmission module. This makes the image of the first board material clearer and facilitates defect identification.

[0068] Due to the large size of the image data, in this embodiment, a 5G data transmission system is used to transmit the obtained first board image to the defect recognition system in real time and quickly.

[0069] After the defect recognition system acquires the image of the first board material, it further includes:

[0070] S200. Perform noise reduction and quality restoration processing on the first plate image to obtain the target image.

[0071] While image recognition methods can theoretically be used to identify cracks and defects in parts and panels, there have been no successful applications in solid wood panels. One major reason is that traditional image recognition is mostly based on a stand-alone model and is generally only used for quality inspection. However, solid wood furniture production involves many steps, and each step of cutting may expose new defects not previously visible on the surface. Therefore, defect identification is needed at almost every step, and multiple production lines operate in parallel. Using a stand-alone model would be too costly; thus, a distributed image acquisition, high-speed data transmission, and server-based rapid identification model is needed to improve the efficiency of actual production. Secondly, defect identification in solid wood panels faces unique challenges in image noise removal. This is due to objective factors affecting defect identification, such as high dust levels, limited lighting conditions, and a large amount of sawdust in solid wood production workshops, which require technical solutions.

[0072] Therefore, after obtaining the image of the first board material, it is necessary to first perform noise reduction and quality restoration processing on the image of the first board material in order to obtain a target image that is easy to identify defects.

[0073] Reference Figure 4 The first board image, i.e., the original image f(x,y), is affected by wood chips on the board, dust from the production workshop, and lighting, resulting in environmental noise n(x,y). Furthermore, in this embodiment, the first board image is transmitted via a 5G network. The acquired first board image g(x,y) is affected by the transmission medium, optical instruments, and storage medium during transmission and storage, referred to as the influence function h(x,y). Therefore, the image used for defect identification inevitably experiences quality degradation. Therefore, using… Figure 3 The process performs noise reduction and quality restoration processing on the received first board image to obtain a restored image.

[0074] The mathematical description of the first plate image is as follows:

[0075] g(x,y)=h(x,y)*f(x,y)+n(x,y)

[0076] Where * represents spatial convolution, f(x,y) is the high-quality original image to be obtained through processing, n(x,y) is the environmental noise, g(x,y) is the received image after being affected, i.e., the first board image, and h(x,y) is the influence function.

[0077] To put the above in another form:

[0078] g(x,y)=H[f(x,y)]+n(x,y)

[0079] Where H[f(x,y)] is the degradation calculation formula for the original image f(x,y). For simplicity, it is expressed as follows:

[0080] g = Hf + n

[0081] That is, H is the degradation calculation formula, f is the original image, n is the environmental noise, and g is the first board image.

[0082] In this embodiment, the goal is to obtain the restored image. Therefore, firstly, given that the environmental noise n is unknown, we determine an approximate image. Make Approaching g, mathematically expressed as

[0083]

[0084] Specifically, relevant information about the first board image g, H, and n is added to the model to obtain an approximate solution for the original image, i.e., the optimized model is:

[0085]

[0086] Among them, Q reg (f) is a regularization term containing prior knowledge of the image, E(f) is an image preservation term, and μ is used to adjust Q. reg The weights of E(f) and E(f) are given. The optimization model is transformed into:

[0087]

[0088] Where L is generally taken as the identity matrix I, f 0 This was the initially predicted optimal solution.

[0089] Furthermore, Q reg If (f) is replaced with TV(f), and E(f) is replaced with the 1-norm, then the optimization model is transformed into:

[0090]

[0091] In this embodiment, the formula for calculating TV(f) is:

[0092]

[0093] Among them, D x and D y These represent the forward finite difference calculation formulas in the horizontal and vertical directions, respectively, σ x and σ y D is a constant in its corresponding direction. x f represents the positive finite difference in the horizontal direction of image f, and D yf represents the positive finite difference in the vertical direction of image f, [D x f] i For vector D x The i-th term of f, [D] y f] i For vector D y The i-th term of f. In this embodiment, the original image is an n×n square image, then i = 1, K, n 2 .

[0094] Among them, the positive finite difference operator refers to expressing the derivative of a variable as a difference in value at different time or space points using a Taylor series expansion. x This involves performing a Taylor series expansion of f in the x-direction, then dividing the image domain into several grids according to spatial unit granularity, and using the difference approximation formed by the values ​​of the unknown function at the grid nodes to replace the derivatives of each order appearing in the partial differential equations used. (D) y The process involves performing a Taylor series expansion of f in the y direction, then dividing the image domain into several grids according to spatial unit granularity, and using the difference approximation formed by the values ​​of the unknown function at the grid nodes to replace the derivatives of each order that appear in the partial differential equations used.

[0095] Specifically, the forward finite difference operator

[0096] Forward Finite Difference Operator

[0097] Where S represents the unit granularity of image space division, and i and j are the index numbers of the unit granularity on x and y, respectively.

[0098] To facilitate timely response to the identification of defects in solid wood products leaving the water, faster image quality restoration is required. In this embodiment, without loss of generality, the original image is cropped into an n×n square image, i.e. Degenerate operator First, two variables, Hf-g and D, which are respectively close to the original variables, are introduced into the model. i f's z and w, where D i f is in matrix form of TV(f) and Here, the matrix consists of the values ​​obtained from the Taylor expansion. Instead of connecting them with plus signs, each value is listed as an element in the matrix. For example, the Taylor expansion f = m1 + m2 + m3 + m4… can be written as [m1 m2 m3 m4]. Then, a quadratic term is added, so the objective optimization problem becomes:

[0099]

[0100] Where, the optimal solution of f at the end of the iteration is the target image, the initial value of g is the first board image, the first board image is an n×n board image, and α i The positive parameters are greater than 0, z and w are intermediate parameters for solving the problem, and D... i f is the matrix form of TV(f), β and γ are adjustment parameters, H is the degeneracy calculation formula, and μ is the positive factor.

[0101] After obtaining the target optimization problem, in this embodiment, the algorithm flow for denoising and restoring the quality of solid wood images is shown in Table 1:

[0102]

[0103] Table 1

[0104] That is, the method of solving the objective optimization problem iteratively includes:

[0105] Initialize, input H, μ>0, β, γ and {α i >0, i=1, K,n 2 The value of}, input the initial value of g. 0 That is, the image of the first board material, f 0 =g 0 Iterative calculation of count k=0;

[0106] In the (k+1)th calculation:

[0107] Limit f to f k The optimal solution w for variable w is calculated based on the first formula. * Reassign w k+1 =w * ;

[0108] Limit f to f k The optimal solution z for variable z is calculated based on the second formula. * Reassign z k+1 =z * ;

[0109] Limit w k+1 =w * and z k+1 =z * The optimal solution f for variable f in this iteration is calculated based on the third formula. * Reassign f k+1 =f * ;

[0110] Determine whether variable f satisfies the convergence condition. If it does, terminate the calculation; otherwise, continue the calculation until the convergence condition is met. The convergence condition is that f does not change after 10 consecutive iterations.

[0111] The optimal solutions for variables w, z, and f are respectively solved using the following formulas:

[0112] The first formula is:

[0113]

[0114] The second formula is:

[0115]

[0116] Where o is the point-to-point product, and the sign function sign(Hf-g) is:

[0117]

[0118] The third formula is:

[0119]

[0120] In this embodiment, after the convergence condition is met and the target image is obtained, the method further includes the following steps:

[0121] S300. The defect recognition neural network model is trained to identify and annotate the defects in the target image, thereby obtaining an annotated image.

[0122] Before identifying defects in the target image, the method further includes:

[0123] Obtain the target training image set, and construct the defect recognition training dataset by manually annotating the location, range and type of defects in the target training image set.

[0124] Specifically, before identifying defects in the target image, a target training image set is obtained. This set consists of a group of original images that have undergone preprocessing and noise reduction / quality restoration. The target training image set is hierarchically numbered according to different types of solid wood boards and different images of the same solid wood board. The location, extent, and type of defects in some solid wood boards are manually labeled to form a defect recognition training dataset. The dataset size requirements are: no fewer than 30 images of the same board type, including data under different angles and lighting conditions; no fewer than 2000 samples of different boards, including different types, colors, textures, and defect degrees of solid wood boards, with no fewer than 100 images of each type of board.

[0125] The defect recognition training dataset is input into the defect recognition neural network model. A 64-layer neural network model based on GoogLeNet and Single Shot MultiBox Detector is used to train the model, resulting in a trained defect recognition neural network model.

[0126] The target images are numbered and input sequentially into a trained defect recognition neural network model. The trained defect recognition neural network model identifies defects in the target images and marks the location coordinates and range of the defects to obtain an annotated image. The annotated image contains the number, location coordinates, range, and type of defects in the target image.

[0127] S400: The labeled image is sent back to the production line management system.

[0128] The data of the labeled images is transmitted back to the solid wood furniture production line management system via a 5G network. One copy enters the production line wood selection system, and the other copy enters the solid wood board visual image acquisition system. The system then uses a voice broadcast system to inform the solid wood board of the number, location, and type of defects.

[0129] Based on the labeled image, determine the defects of the target board in the labeled image, determine the production line corresponding to the furniture type applicable to the target board based on the defects of the target board, and schedule the target board to the production line corresponding to the furniture type applicable to the target board.

[0130] In summary, this embodiment provides an intelligent defect recognition method for solid wood panels based on 5G+Industrial Internet. The method preprocesses the original image captured by the camera to obtain a first panel image, performs noise reduction and quality restoration processing on the first panel image to obtain a target image, and then inputs the target image into a trained defect recognition neural network. The trained defect recognition neural network model identifies and annotates the defects in the target image to obtain an annotated image. Subsequently, the annotated image is transmitted back to the production line management system, realizing non-manual defect recognition technology for solid wood panels and improving the automation and intelligence level of solid wood furniture production lines.

[0131] It should be understood that although the steps in the flowcharts shown in the accompanying drawings are displayed sequentially according to the arrows, these steps are not necessarily executed in the order indicated by the arrows. Unless explicitly stated herein, there is no strict order restriction on the execution of these steps, and they can be executed in other orders. Moreover, at least some steps in the flowchart may include multiple sub-steps or multiple stages. These sub-steps or stages are not necessarily completed at the same time, but can be executed at different times. The execution order of these sub-steps or stages is not necessarily sequential, but can be performed alternately or in turn with other steps or at least a portion of the sub-steps or stages of other steps.

[0132] Those skilled in the art will understand that all or part of the processes in the methods of the above embodiments can be implemented by a computer program instructing related hardware. The computer program can be stored in a non-volatile computer-readable storage medium. When executed, the computer program can include the processes of the embodiments of the above methods. Any references to memory, storage, databases, or other media used in the embodiments provided by this invention can include non-volatile and / or volatile memory. Non-volatile memory may include read-only memory (ROM), programmable ROM (PROM), electrically programmable ROM (EPROM), electrically erasable programmable ROM (EEPROM), or flash memory. Volatile memory may include random access memory (RAM) or external cache memory. By way of illustration and not limitation, RAM is available in a variety of forms, such as static RAM (SRAM), dynamic RAM (DRAM), synchronous DRAM (SDRAM), dual data rate SDRAM (DDRSDRAM), enhanced SDRAM (ESDRAM), synchronous link DRAM (SLDRAM), RAMbus direct RAM (RDRAM), direct memory bus dynamic RAM (DRDRAM), and memory bus dynamic RAM (RDRAM).

[0133] Example 2

[0134] Based on the above embodiments, the present invention also provides a smart defect identification device for solid wood boards based on 5G+Industrial Internet, such as... Figure 5 As shown, the intelligent defect identification device for solid wood panels based on 5G+Industrial Internet includes:

[0135] An image acquisition module is used to acquire an image of the first board material, as described in Embodiment 1.

[0136] A noise reduction and quality restoration processing module is used to perform noise reduction and quality restoration processing on the first board image to obtain a target image, as described in Embodiment 1.

[0137] The defect annotation module is used to identify and annotate defects in the target image through a trained defect recognition neural network model to obtain an annotated image, as described in Embodiment 1.

[0138] The transmission module is used to transmit the labeled image back to the production line management system, as described in Embodiment 1.

[0139] Example 3

[0140] Based on the above embodiments, the present invention also provides a terminal, such as... Figure 6 As shown, the terminal includes a processor 10 and a memory 20. Figure 6 Only some of the terminal components are shown; however, it should be understood that it is not required to implement all of the components shown, and more or fewer components may be implemented instead.

[0141] In some embodiments, the memory 20 may be an internal storage unit of the terminal, such as a hard drive or memory. In other embodiments, the memory 20 may be an external storage device of the terminal, such as a plug-in hard drive, smart media card (SMC), secure digital card (SD), flash card, etc. Further, the memory 20 may include both internal and external storage devices. The memory 20 is used to store application software and various types of data installed on the terminal. The memory 20 can also be used to temporarily store data that has been output or will be output. In one embodiment, the memory 20 stores a 5G+Industrial Internet-based intelligent identification program 30 for solid wood board defects. This 5G+Industrial Internet-based intelligent identification program 30 for solid wood board defects can be executed by the processor 10, thereby realizing the 5G+Industrial Internet-based intelligent identification method for solid wood board defects in this application.

[0142] In some embodiments, the processor 10 may be a central processing unit (CPU), a microprocessor, or other chip, used to run program code stored in the memory 20 or process data, such as executing the super-resolution image quality evaluation method.

[0143] In one embodiment, reference is made to Figure 7The flowchart illustrates the following steps when processor 10 executes the 5G+Industrial Internet-based intelligent identification program for solid wood board defects stored in memory 20:

[0144] Obtain the image of the first board material;

[0145] The first plate image is subjected to noise reduction and quality restoration processing to obtain the target image;

[0146] The defect recognition neural network model is trained to identify and annotate defects in the target image, resulting in an annotated image;

[0147] The labeled image is then transmitted back to the production line management system.

[0148] Prior to obtaining the first board image, the process includes:

[0149] Obtain the original image of the board material;

[0150] The original image of the board material is preprocessed to obtain the first image of the board material.

[0151] The step of performing noise reduction and quality restoration processing on the first board image includes:

[0152] The target image is obtained by solving the objective optimization problem iteratively.

[0153] The objective optimization problem is:

[0154]

[0155] Where, the optimal solution of f at the end of the iteration is the target image, the initial value of g is the first board image, the first board image is an n×n board image, and α i The positive parameters are greater than 0, z and w are intermediate parameters for solving the problem, and D... i f is the matrix form of TV(f), β and γ are adjustment parameters, H is the degeneracy calculation, and μ is the positive factor;

[0156] The formula for calculating TV(f) is:

[0157]

[0158] Among them, D x and D y These represent the forward finite difference calculation formulas in the horizontal and vertical directions, respectively, σ x and σ y For the constant of its corresponding direction, [D x f] i For vector D x The i-th term of f, [D] y f]i For vector D y The i-th term of f.

[0159] The method of solving the objective optimization problem iteratively includes:

[0160] Initialize, input H, μ>0, β, γ and {α i >0, i=1, K,n 2 The value of}, input the initial value of g. 0 That is, the image of the first board material, f 0 =g 0 Iterative calculation of count k=0;

[0161] In the (k+1)th calculation:

[0162] Limit f to f k The optimal solution w for variable w is calculated based on the first formula. * Reassign w k+1 =w * ;

[0163] Limit f to f k The optimal solution z for variable z is calculated based on the second formula. * Reassign z k+1 =z * ;

[0164] Limit w k+1 =w * and z k+1 =z * The optimal solution f for variable f in this iteration is calculated based on the third formula. * Reassign f k+1 =f * ;

[0165] Determine whether variable f satisfies the convergence condition. If it does, terminate the calculation; otherwise, continue the calculation until the convergence condition is met.

[0166] The first formula is:

[0167]

[0168] The second formula is:

[0169]

[0170] Where o is the point-to-point product, and the sign function sign(Hf-g) is:

[0171]

[0172] The third formula is as follows:

[0173]

[0174] The labeled image includes the number, location coordinates, range, and type of defects in the target image.

[0175] Example 4

[0176] The present invention also provides a computer-readable storage medium storing one or more programs, which can be executed by one or more processors to implement the steps of the intelligent identification method for solid wood board defects based on 5G+Industrial Internet as described above.

[0177] Finally, it should be noted that the above embodiments are only used to illustrate the technical solutions of the present invention, and not to limit them; although the present invention has been described in detail with reference to the foregoing embodiments, those skilled in the art should understand that modifications can still be made to the technical solutions described in the foregoing embodiments, or equivalent substitutions can be made to some of the technical features; and these modifications or substitutions do not cause the essence of the corresponding technical solutions to deviate from the spirit and scope of the technical solutions of the embodiments of the present invention.

Claims

1. A method for intelligent identification of defects in solid wood panels based on 5G + Industrial Internet, characterized in that, The method includes: Obtain the image of the first board material; The first plate image is subjected to noise reduction and quality restoration processing to obtain the target image; The defect recognition neural network model is trained to identify and annotate defects in the target image, resulting in an annotated image; The labeled image is then transmitted back to the production line management system. The denoising and quality restoration processing of the first board image includes: The target image is obtained by solving the objective optimization problem iteratively. The objective optimization problem is: Where, the optimal solution of f at the end of the iteration is the target image, the initial value of g is the first board image, the first board image is an n×n board image, and α i The positive parameters are greater than 0, z and w are intermediate parameters for solving the problem, and D... i f is the matrix form of TV(f), β and γ are adjustment parameters, H is the degeneracy calculation, and μ is the positive factor; The formula for calculating TV(f) is: Among them, D x and D y These represent the forward finite difference calculation formulas in the horizontal and vertical directions, respectively, σ x and σ y As a constant for its corresponding direction, [D x f] i For vector D x The i-th term of f, [D] y f] i For vector D y The i-th term of f.

2. The intelligent identification method for defects in solid wood panels based on 5G+Industrial Internet according to claim 1, characterized in that, Before acquiring the first board image, the process includes: Obtain the original image of the board material; The original image of the board material is preprocessed to obtain the first image of the board material.

3. The intelligent identification method for defects in solid wood panels based on 5G+Industrial Internet according to claim 1, characterized in that, The method of solving the objective optimization problem using iteration includes: Initialize, input H, μ>0, β, γ and {α i >0, i=1, K,n 2 The value of}, input the initial value of g. 0 That is, the image of the first board material, f 0 =g 0 Iterative calculation of count k=0; In the (k+1)th calculation: Limit f to f k The optimal solution w for variable w is calculated based on the first formula. * Reassign w k+1 =w * ; Limit f to f k The optimal solution z for variable z is calculated based on the second formula. * Reassign z k+1 =z * ; Limit w k+1 =w * and z k+1 =z * The optimal solution f for variable f in this iteration is calculated based on the third formula. * Reassign f k +1 =f * ; Determine whether variable f satisfies the convergence condition. If it does, terminate the calculation; otherwise, continue the calculation until the convergence condition is met.

4. The intelligent identification method for defects in solid wood panels based on 5G+Industrial Internet according to claim 3, characterized in that, The first formula is: The second formula is: Where o is the point-to-point product, and the sign function sign(Hf-g) is:

5. The intelligent identification method for defects in solid wood panels based on 5G+Industrial Internet according to claim 3, characterized in that, The third formula is:

6. The intelligent identification method for defects in solid wood panels based on 5G+Industrial Internet according to claim 1, characterized in that, The labeled image contains the number, location coordinates, range, and type of defects in the target image.

7. A smart defect identification device for solid wood boards based on 5G+Industrial Internet, characterized in that, include: Image acquisition module, the image acquisition module is used to acquire an image of the first board material; A noise reduction and quality restoration processing module is used to perform noise reduction and quality restoration processing on the first board image to obtain a target image; The defect annotation module is used to identify and annotate defects in the target image using a trained defect recognition neural network model, thereby obtaining an annotated image. A transmission module is used to transmit the labeled image back to the production line management system; The 5G+Industrial Internet-based intelligent identification device for solid wood board defects is also used to solve the target optimization problem using an iterative method to obtain the target image; The objective optimization problem is: Where, the optimal solution of f at the end of the iteration is the target image, the initial value of g is the first board image, the first board image is an n×n board image, and α i The positive parameters are greater than 0, z and w are intermediate parameters for solving the problem, and D... i f is the matrix form of TV(f), β and γ are adjustment parameters, H is the degeneracy calculation, and μ is the positive factor; The formula for calculating TV(f) is: Among them, D x and D y These represent the forward finite difference calculation formulas in the horizontal and vertical directions, respectively, σ x and σ y As a constant for its corresponding direction, [D x f] i For vector D x The i-th term of f, [D] y f] i For vector D y The i-th term of f.

8. A terminal, characterized in that, The terminal includes: a processor and a computer-readable storage medium communicatively connected to the processor. The computer-readable storage medium is adapted to store multiple instructions, and the processor is adapted to call the instructions in the computer-readable storage medium to execute the steps of the intelligent identification method for solid wood board defects based on 5G+Industrial Internet as described in any one of claims 1-6.

9. A computer-readable storage medium, characterized in that, The computer-readable storage medium stores one or more programs, which can be executed by one or more processors to implement the steps of the intelligent identification method for solid wood board defects based on 5G+Industrial Internet as described in any one of claims 1-6.

Images