A combined template production quality detection method and system

Abstract

The application relates to the technical field of image detection, and discloses a combined template production quality detection method and system, wherein the method comprises the following steps: collecting a side surface image by adopting an RGB industrial camera in combination with multi-angle directional illumination and a signal-to-noise ratio self-feedback mechanism; performing channel selection, edge enhancement and dynamic narrowband ROI extraction; identifying interlayer boundary lines by adopting a direction angle filtering and line segment merging mechanism; establishing an offset vector function; and outputting offset and root mean square error results. Compared with the detection mode for judging the pressing quality of a multi-layer plywood in the prior art, which depends on manual visual inspection or a simple image threshold method, especially under actual production conditions that there is slight misalignment between the board layers and the edge texture contrast is insufficient, the technical problem that accurate identification and quantitative evaluation of the interlayer misalignment of the plywood side surface cannot be realized is solved; and due to the structured boundary extraction and quantitative vector modeling mechanism, the detection accuracy and the automation degree of slight misalignment defects are improved.

Description

Technical Field

[0001] This invention relates to the field of image detection technology, and in particular to a method and system for detecting the production quality of combined templates. Background Technology

[0002] Currently, multi-layer plywood, as an important structural composite formwork, is widely used in construction formwork, furniture manufacturing, vehicle flooring, and composite flooring. Multi-layer plywood is typically made of three or more veneers laminated together with adhesives in a cross-laminated pattern. Its quality stability largely depends on the bonding precision and alignment of each layer during pressing. However, in actual production, due to various factors such as human error during veneer loading, insufficient installation precision, veneer slippage in the initial pressing stage, uneven adhesive penetration, and hot-pressing displacement, slight but structurally risky misalignments often occur between multiple layers. These misalignments usually manifest as relative displacement of one or more intermediate layers in the length direction (X-axis) or thickness direction (Y-axis). Although the misalignment may only be on the order of millimeters, it can cause subsequent edge trimming errors, thickness fluctuations, surface bulging, and delamination risks, seriously affecting the quality grade and mechanical properties of the finished product. This is especially true in outdoor formwork or flooring splicing scenarios, where the requirements for interlayer flatness and structural overlap are even more stringent. Existing quality inspection methods are mainly divided into two categories: one is visual inspection that relies on human experience. However, due to the complex texture of the board layers, the blurred edges, and the fact that they are often obscured by glue residue or edge pressing, manual inspection is not only inefficient and has poor repeatability, but it is also almost impossible to accurately identify minor misalignments. The other category is physical measurement devices based on contact probes or line laser displacement sensors. Although these devices can obtain certain edge contour information, they have the risk of contact damage, sparse measurement points, limited coverage, and difficulty in achieving comprehensive characterization. Furthermore, they have high requirements for equipment rigidity and stability, making them unsuitable for high-speed continuous production lines.

[0003] Therefore, there is an urgent need for a quality inspection method for modular formwork production to solve the problems of unstable detection and inaccurate boundaries, and to improve the intelligence and standardization of quality inspection for modular formwork production. Summary of the Invention

[0004] To address the aforementioned technical shortcomings, the present invention aims to propose a method for quality inspection in the production of composite templates. This method addresses the technical problem that existing inspection methods, which rely on manual visual inspection or simple image thresholding to determine the bonding quality of multi-layer plywood, are unable to accurately identify and quantify the interlayer misalignment on the plywood side under actual production conditions where slight misalignment occurs between layers and edge texture contrast is insufficient.

[0005] To solve the above technical problems, the present invention adopts the following technical solution: The present invention provides a method for quality inspection in the production of composite templates.

[0006] The method for inspecting the production quality of the combined template includes:

[0007] Step S10: Using an RGB industrial camera installed in the side detection area of ​​the plywood, images of the plywood side are acquired while the plywood is stationary or being conveyed at a constant speed, employing a self-feedback control mechanism based on multi-angle directional illumination and signal-to-noise ratio. ;

[0008] Step S20: Based on the side image of the multi-layer plywood A phased image guidance mechanism is used to sequentially execute the channel selection task, the diffusion filtering task, and the dynamic narrowband ROI extraction task, and output the enhanced image G and the region of interest R;

[0009] Step S30: In the region of interest R, the adaptive Canny algorithm is used in combination with orientation angle filtering and line segment merging mechanism to perform edge extraction and layer boundary line segment recognition, and output the initial line segment set L and the boundary line segment set. ;

[0010] Step S40: Based on the set of boundary segments An offset vector function is established in physical space using a physical location fitting benchmark difference comparison method.

[0011] Step S50: Calculate the maximum offset based on the offset vector function and overall offset root mean square error Set the maximum offset threshold and root mean square error threshold ,like If the plywood is found to be of acceptable quality, then it is determined that the plywood has a potential misalignment.

[0012] Preferably, in step S10, a self-feedback control mechanism based on multi-angle directional illumination and signal-to-noise ratio is used to acquire images of the side surface of multi-layer plywood. The steps specifically include:

[0013] First, construct a directional lighting field: Two sets of oblique LED line light sources are arranged on both sides of the multi-layer plywood. These two sets of oblique LED line light sources include a first oblique LED line light source and a second oblique LED line light source; wherein, the illumination angle of the first oblique LED line light source is... Illumination angle with the second oblique LED line light source satisfy and ;Lighting angle and lighting angle This is used to control two sets of oblique LED line light sources to face the plywood layer direction respectively, so as to form a uniform gradient projection effect and enhance the light and dark difference of the interlayer misalignment boundary.

[0014] Then perform image signal-to-noise ratio evaluation and trigger self-feedback acquisition: based on illumination angle and lighting angle Start the RGB industrial camera to acquire initial images For the initial image Perform signal-to-noise ratio (SNR) evaluation and output an SNR evaluation score. , ;in, For the initial image Average brightness in the edge area of ​​multi-layer plywood For the initial image Standard deviation of luminance in the edge area of ​​multi-layer plywood; preset signal-to-noise ratio evaluation score threshold. ,like Then confirm the initial image. It is in a valid state; if This triggers the self-feedback control mechanism, restarting the RGB industrial camera to acquire the initial image. ;

[0015] Finally, a linear scaling method based on the imaging geometry calibration model is used to establish the mapping relationship between pixels and actual length. Based on this mapping relationship, the final side image of the multi-layer plywood is output. .

[0016] Preferably, in step S20, the image is based on the side view of the multilayer plywood. The process employs a phased image guidance mechanism, sequentially executing channel selection, diffusion filtering, and dynamic narrowband ROI extraction tasks, outputting an enhanced image G and a region of interest R. Specifically, this includes:

[0017] Step S201: Image of the side surface of the multi-layer plywood within a preset edge pre-selection area. Calculate the edge gradient response intensities of the three channels, including the edge gradient response intensities of the first, second, and third channels; determine the optimized channel image based on the edge gradient response intensities of the three channels. ;

[0018] Step S202: Based on the preset PM diffusion model, an edge response adaptive diffusion control mechanism is used to optimize the channel image. Perform iterative updates to obtain an edge-enhanced image;

[0019] Step S203: Extract dynamic narrowband regions based on the edge enhancement image using vertical gradient peak localization and bandwidth limitation control mechanism, and output the enhanced image G and the region of interest R.

[0020] Preferably, in step S202, an edge-response adaptive diffusion control mechanism is used to optimize the channel image based on a preset PM diffusion model. The formula used in the iterative update step is: ; This is a two-dimensional divergence operator used to describe the spatial distribution variation of the diffusion flow in the PM diffusion model; where, A time marker for the number of iterations used to perform the iterative update; To optimize channel images The rate of change along the time dimension of the iteration number; The diffusion control operator introduced for the edge response adaptive diffusion control mechanism; This is the diffusion control coefficient; This is the edge sensitivity control factor.

[0021] Preferably, step S203, which involves extracting a dynamic narrowband region based on the edge enhancement image using a vertical gradient peak localization and bandwidth limiting control mechanism, and outputting the enhanced image G and the region of interest R, specifically includes:

[0022] First, in the edge enhancement image, the vertical grayscale gradient magnitude of each row of pixels is calculated row by row along the vertical direction, and the total intensity of the corresponding row gradient is counted. The position with the largest total intensity of the vertical gradient is marked as the vertical center coordinate, which is used as the response center of the misaligned boundary in the edge enhancement image.

[0023] Next, a narrow-band window of fixed width is set based on the vertical center coordinates. The upper and lower boundaries of the narrow-band window are distanced from the vertical center coordinates. The bandwidth length determines the dynamic narrowband region;

[0024] Finally, the edge-enhanced image is output as the enhanced image G, and the dynamic narrowband region is output as the region of interest R.

[0025] Preferably, in step S30, an adaptive Canny algorithm is used in the region of interest R, combined with orientation angle filtering and line segment merging mechanism, to perform edge extraction and layer boundary line segment recognition, outputting an initial line segment set L and a boundary line segment set. The steps specifically include:

[0026] Step S301: Extract the edge map in the region of interest R using the adaptive Canny algorithm, and calculate the horizontal edge response based on the edge map. Horizontal edge response in edge map exist and The edge pixels between them form a set of edge segments after directional filtering;

[0027] Step S302: Based on the edge line segment set, the line segment detection algorithm LSD is used to extract the initial line segment set L. Each line segment in the initial line segment set L is represented in the form of a parameter pair, which includes the parameter pair consisting of the slope and intercept of the line segment.

[0028] Step S303: For line segment pairs in the initial line segment set L that satisfy the preset collinearity condition, perform a merging and reconstruction operation to obtain the interlayer boundary line segment set. , ;in, This is the first interlayer boundary segment. This is the second interlayer boundary segment. This is the nth interlayer boundary segment, where n is the number of layers in the plywood minus 1;

[0029] Step S304: Output the initial set of line segments L and the set of boundary line segments. .

[0030] Preferably, in step S40, based on the set of boundary line segments... The steps for establishing an offset vector function in physical space using the physical location fitting benchmark difference comparison method specifically include:

[0031] Step S401: Select the set of boundary line segments Line segment at the middle and end position As a reference layer, construct the reference layer position function corresponding to the reference layer. And based on the reference layer position function Construct a set of boundary segments The Middle Position offset function of line segment ;

[0032] Step S402: Combine the position offset functions of all line segments to construct a two-dimensional offset vector field. , Where m represents the number of line segments in the set of boundary line segments; For the first The position offset function of a line segment. For the first The position offset function of the line segment; and the two-dimensional offset vector field Output as an offset vector function.

[0033] The present invention also provides a combined template production quality inspection system comprising:

[0034] The image acquisition module is used to acquire images of the side of the plywood using an RGB industrial camera installed on the side detection area of ​​the plywood, whether the plywood is stationary or being conveyed at a constant speed. The acquisition employs a self-feedback control mechanism based on multi-angle directional illumination and signal-to-noise ratio. ;

[0035] Image preprocessing and ROI extraction module for images based on the side views of multi-layer plywood. A phased image guidance mechanism is used to sequentially execute the channel selection task, the diffusion filtering task, and the dynamic narrowband ROI extraction task, and output the enhanced image G and the region of interest R;

[0036] The edge and boundary segment extraction module is used to perform edge extraction and layer boundary segment recognition in the region of interest R using an adaptive Canny algorithm combined with orientation angle filtering and segment merging mechanism, and outputs an initial segment set L and a boundary segment set. ;

[0037] The physical offset modeling module is used to model based on the set of boundary segments. An offset vector function is established in physical space using a physical location fitting benchmark difference comparison method.

[0038] The quality assessment module is used to calculate the maximum offset based on the offset vector function. and overall offset root mean square error Set the maximum offset threshold and root mean square error threshold ,like If the plywood is found to be of acceptable quality, then it is determined that the plywood has a potential misalignment.

[0039] The present invention also provides a composite template production quality inspection device, comprising: a memory, a processor, and a composite template production quality inspection program stored in the memory and executable on the processor, wherein the composite template production quality inspection program implements a composite template production quality inspection method when executed by the processor.

[0040] The present invention also provides a computer program product, including a composite template production quality inspection program, which, when executed by a processor, implements the composite template production quality inspection method.

[0041] The beneficial effects of this invention are as follows: By introducing an image recognition-driven multi-channel enhancement and dynamic region of interest extraction mechanism, combined with side images acquired by an RGB industrial camera, this invention can stably extract the interlayer edges of multilayer plywood under different lighting and combined template texture backgrounds, effectively avoiding edge recognition errors caused by natural texture interference.

[0042] This invention constructs a physical offset vector function for interlayer misalignment based on image recognition technology, and introduces quantitative evaluation indicators of maximum offset and root mean square error. It can automatically determine the alignment accuracy of multilayer plywood during the pressing process, realize high-precision detection and visualization analysis of misalignment defects, and improve detection efficiency and automation level. Attached Figure Description

[0043] To more clearly illustrate the technical solutions in the embodiments of the present invention or the prior art, the drawings used in the description of the embodiments or the prior art will be briefly introduced below. Obviously, the drawings described below are only some embodiments of the present invention. For those skilled in the art, other drawings can be obtained based on these drawings without creative effort.

[0044] Figure 1 This is a flowchart illustrating the first embodiment of a combined template production quality inspection method according to the present invention.

[0045] Figure 2 This is a schematic diagram of the equipment for a combined template production quality inspection method according to the present invention. Detailed Implementation

[0046] The technical solutions of the embodiments of the present invention will be clearly and completely described below with reference to the accompanying drawings. Obviously, the described embodiments are only some embodiments of the present invention, and not all embodiments. Based on the embodiments of the present invention, all other embodiments obtained by those skilled in the art without creative effort are within the scope of protection of the present invention.

[0047] Example 1: As Figure 1 The diagram shown is a flowchart of the first embodiment of the combined template production quality inspection method of the present invention, which presents the first embodiment of the combined template production quality inspection method of the present invention.

[0048] In the first embodiment, the method for quality inspection of the combined template production includes:

[0049] Step S10: Using an RGB industrial camera installed in the side detection area of ​​the plywood, images of the plywood side are acquired while the plywood is stationary or being conveyed at a constant speed, employing a self-feedback control mechanism based on multi-angle directional illumination and signal-to-noise ratio. ;

[0050] It should be noted that multi-angle directional lighting refers to setting multiple LEDs or linear light sources with different directions on the side of the plywood to highlight the contrast features of the plywood interlayer boundaries under different incident angles and enhance edge visibility; the signal-to-noise ratio self-feedback control mechanism refers to calculating the brightness contrast, edge sharpness and other indicators between the target area and the background area in the image in real time, and automatically adjusting the brightness ratio of each lighting source according to the feedback results to maximize the signal difference between the target boundary and the background and suppress texture noise interference.

[0051] Understandably, in images acquired using this mechanism, the edge contours of multi-layer plywood exhibit a steeper transition and clearer brightness boundaries in terms of grayscale changes. This significantly reduces false edge responses caused by background interference such as natural textures, knots, and color differences, which is beneficial for the accurate extraction of true interlayer boundaries in subsequent image recognition steps, thereby enhancing the detection process's ability to perceive minor misalignment issues.

[0052] It should be understood that, compared to traditional detection schemes that only use a fixed light source at a single angle, this invention, by constructing a multi-angle illumination field and combining it with a signal-to-noise ratio feedback adjustment mechanism, can dynamically adapt imaging conditions for multi-layered plywood with different materials, color tones, and texture complexities. Especially when there are color differences, knots, shadows, or reflective interference on the plywood surface, traditional methods often result in blurred or lost image edges due to local overexposure or underexposure, while the method of this invention can maintain a stable high level of boundary recognition accuracy.

[0053] For example, when testing a 12-layer plywood sample of a certain model, after image acquisition using a traditional fixed light source illumination scheme, only 8 to 9 interlayer boundaries could be continuously identified in the edge map, and at least 2 boundaries showed blurred breaks. However, after adopting the multi-angle directional illumination and signal-to-noise ratio self-feedback control mechanism of this invention, the number of effective boundary segments identified in the same plywood sample at the same location increased to more than 11, and the average signal-to-noise ratio of the boundary segments in the gradient response improved by about 34.8%. This indicates that the edge enhancement effect of this method is more significant in actual detection and can provide more accurate input images for subsequent misalignment calculations.

[0054] Step S20: Based on the side image of the multi-layer plywood A phased image guidance mechanism is used to sequentially execute the channel selection task, the diffusion filtering task, and the dynamic narrowband ROI extraction task, and output the enhanced image G and the region of interest R;

[0055] It should be noted that the phased image guidance mechanism refers to a task chain-like nested structure in the image processing, where the output of each stage serves as the input for the next stage, thus ensuring contextual relevance and goal consistency in image processing. Specifically, the first stage, channel selection, filters single-channel or fused-channel images with the strongest edge response features from RGB images; the second stage, diffusion filtering, employs an improved Perona-Malik (PM) nonlinear diffusion model to enhance structural details with edge preservation and noise reduction; the third stage, dynamic narrowband ROI extraction, extracts image regions that may contain misaligned boundaries based on local gradient peak clusters and directional consistency windows, forming the final enhanced image G and region of interest R. Each of these stages automatically triggers processing branches or weight adjustment strategies through preset image evaluation metrics (such as edge entropy, uniformity variance, etc.).

[0056] Understandably, by employing this staged image guidance mechanism, edge blurring and spurious responses caused by compression textures, background adhesive layers, or knot noise in the original image can be effectively suppressed, while enhancing the response of real interlayer structural boundaries, making edge segment detection in subsequent steps more stable and accurate. Especially when real images have multi-scale local texture perturbations, this method can significantly improve the structural discriminability of the image and reduce ROI localization errors.

[0057] It should be understood that, compared to existing coarse-grained edge detection methods that directly perform edge detection on the entire image, this invention introduces a three-stage collaborative mechanism—channel response optimization, structure-preserving diffusion, and local ROI localization—into the image processing flow, achieving more efficient target region focusing and feature preservation. In complex plywood contexts, traditional methods often suffer from high false detection rates due to mis-extracting of the adhesive layer texture. This invention, however, significantly improves the accuracy and error resistance of extracting true misaligned boundaries through a structure-guided approach.

[0058] Step S30: In the region of interest R, the adaptive Canny algorithm is used in combination with orientation angle filtering and line segment merging mechanism to perform edge extraction and layer boundary line segment recognition, and output the initial line segment set L and the boundary line segment set. ;

[0059] It should be noted that the adaptive Canny algorithm is an edge detection method that dynamically adjusts high and low thresholds based on local statistical features of the image. Its core lies in automatically setting response thresholds for different image noise levels, brightness distributions, and texture densities, thereby enhancing the robustness of edge detection in complex industrial environments. In this step, the orientation angle filtering mechanism filters candidates based on the angle between the orientation angle of each edge segment and the horizontal direction, retaining only candidate segments with orientations close to the horizontal and excluding vertical or inclined segments unrelated to the natural wood texture, node cracks, etc. The segment merging mechanism merges and reconstructs multiple short segments with consistent orientation, continuous position, and close distance to eliminate the segmented structure caused by edge breaks and obtain a complete and continuous interlayer boundary.

[0060] Understandably, the boundary segment set extracted through the aforementioned combination mechanism possesses good structural consistency and directional stability, effectively representing the actual boundary positions between each layer of plywood and providing a clear and quantifiable basic structural input for subsequent interlayer misalignment measurement. Simultaneously, retaining the initial segment set L supports the tracking and analysis of potential anomalous structures, improving the interpretability and scalability of atypical defects.

[0061] It should be understood that, compared to traditional edge detection methods based on fixed thresholds and simple line fitting, the adaptive adjustment and orientation angle constraint mechanism introduced in this step significantly improves the accuracy of edge detection under conditions of natural lighting interference, image blurring, and complex textures in combined templates. In particular, the orientation angle filtering strategy can effectively exclude most unstructured edges without relying on deep learning models, greatly reducing the false recognition rate and thus enhancing the consistency and reliability of boundary recognition results.

[0062] For example, for a set of multi-layer plywood sample images containing 8 layers, under the condition that the original image contrast is low and the background noise is high, after extracting the boundary line segments using the algorithm described in this step, the average recognition rate of complete boundary line segments per layer can reach 96.3%, which is about 18 percentage points higher than the traditional method using a fixed Canny threshold and least squares fitting. The average integrity of the boundary structure is improved to over 94%, enhancing the reliability of subsequent misalignment measurement calculations.

[0063] Step S40: Based on the set of boundary segments An offset vector function is established in physical space using a physical location fitting benchmark difference comparison method.

[0064] It should be noted that the physical position fitting benchmark difference comparison method refers to the method of converting the image coordinate information of each boundary line segment into spatial position information in actual physical length units based on the set of boundary line segments identified in the image coordinate system through the pixel-to-physical unit mapping relationship, and constructing an ideal boundary benchmark curve as a reference. Based on this reference, the actual position offset of each boundary line segment is calculated, ultimately forming a set of vector functions containing multiple position offset vectors. The offset vector functions are used to quantitatively represent the spatial deviation of each layer boundary from the ideal reference alignment state.

[0065] It should be understood that, compared to traditional manual visual inspection or deviation assessment methods based on fixed templates, this step overcomes the bottlenecks of low resolution, strong subjectivity, and inability to continuously monitor manually by combining the structural extraction of boundary line segments with physical modeling. This method can accurately extract the offset direction, magnitude, and positional trend of each layer under non-destructive conditions, and is particularly suitable for early warning and identification of potential quality problems that occur at the micron level during the production process but have not yet caused appearance defects.

[0066] Step S50: Calculate the maximum offset based on the offset vector function and overall offset root mean square error Set the maximum offset threshold and root mean square error threshold ,like If the plywood is found to be of acceptable quality, then it is determined that the plywood has a potential misalignment.

[0067] It should be noted that the maximum offset refers to the maximum value of the offset modulus among all boundary points covered by the offset vector function, reflecting the degree of misalignment at extreme positions; the overall offset root mean square error refers to the root mean square value of all offset vector moduli, reflecting the overall distribution uniformity of inter-layer offset. These two indicators together constitute the offset characteristic quantity system used for quality assessment, which can be preset according to the product specifications and process requirements of the plywood.

[0068] Understandably, this step, by extracting statistics representing the degree of boundary offset, not only achieves quantitative assessment of misalignment but also compares it with a threshold, thus completing closed-loop control from spatial modeling to quality judgment. Compared to methods that only use a single coordinate point or local samples for judgment, this invention provides a quality assessment scheme based on global offset distribution.

[0069] It should be understood that, compared to the problem that traditional visual inspection or simple edge alignment detection methods cannot detect subtle interlayer misalignments, this invention can identify structural misalignment problems that are difficult to detect with the naked eye but actually exist by establishing a continuous offset vector function and calculating key statistical indicators, thereby improving detection accuracy and anomaly identification capabilities.

[0070] For example, in a batch of actual samples, the maximum offset threshold was set to 0.30 mm, and the overall root mean square error threshold was set to 0.15 mm. Test sample A had a maximum offset of 0.22 mm and a root mean square error of 0.12 mm, and was judged to be qualified; sample B had a maximum offset of 0.35 mm and a root mean square error of 0.18 mm, and was judged to have potential misalignment. Subsequent cross-sectional verification revealed that there was indeed misalignment at the edge of the composite template, verifying the effectiveness of this method.

[0071] Example 2: Furthermore, the present invention provides a combined template production quality inspection system that employs a combined template production quality inspection method described in the above embodiments, which can solve a technical problem related to the quality inspection of combined template production. Compared with the prior art, the beneficial effects of the combined template production quality inspection system provided by the present invention are the same as those of the combined template production quality inspection method described in the above embodiments, and other technical features of the combined template production quality inspection system are the same as those disclosed in the methods of the above embodiments, and will not be repeated here.

[0072] Example 3: This invention provides a quality inspection device for combined template production. Please refer to... Figure 2A composite template production quality inspection device includes: at least one processor; and a memory communicatively connected to the at least one processor; wherein the memory stores instructions executable by the at least one processor, which, when executed by the at least one processor, enable the at least one processor to perform a composite template production quality inspection method as described in Embodiment 1 above. The composite template production quality inspection device in this embodiment may include, but is not limited to, mobile terminals such as mobile phones, laptops, digital radio receivers, PDAs (Personal Digital Assistants), PADs (Portable Application Description), PMPs (Portable Media Players), and vehicle terminals (e.g., vehicle navigation terminals), as well as fixed terminals such as digital TVs and desktop computers. This composite template production quality inspection device is merely an example and should not impose any limitations on the functionality and scope of use of the embodiments of this invention. A composite template production quality inspection device may include a processing device 1001 (e.g., a central processing unit, a graphics processing unit, etc.), which can perform various appropriate actions and processes according to a program stored in a read-only memory 1002 or a program loaded from a storage device 1003 into a random access memory 1004. Random access memory 1004 also stores various programs and data required for the operation of a composite template production quality inspection device. Processing device 1001, read-only memory 1002, and random access memory 1004 are interconnected via bus 1005. I / O interface 1006 is also connected to the bus. Typically, the following systems can be connected to I / O interface 1006: input devices 1007 including, for example, touchscreens, touchpads, keyboards, mice, image sensors, microphones, accelerometers, gyroscopes, etc.; output devices 1008 including, for example, liquid crystal displays (LCDs), speakers, vibrators, etc.; storage devices 1003 including, for example, magnetic tapes, hard disks, etc.; and communication devices 1009. Communication device 1009 allows a composite template production quality inspection device to communicate wirelessly or wiredly with other devices to exchange data. Although a composite template production quality inspection device with various systems is shown in the figures, it should be understood that it is not required to implement or possess all the systems shown. More or fewer systems may be implemented alternatively.

[0073] Example 4: This invention also provides a computer program product, including a computer program that, when executed by a processor, implements the steps of the above-described method for quality inspection of combined template production. The computer program product provided by this invention can solve a technical problem related to quality inspection of combined template production. Compared with the prior art, the beneficial effects of the computer program product provided by this invention are the same as those of the combined template production quality inspection method provided in the above embodiments, and will not be repeated here.

[0074] In particular, according to the embodiments disclosed in this invention, the processes described above with reference to the flowcharts can be implemented as computer software programs. For example, embodiments of this invention include a computer program product comprising a computer program carried on a computer-readable medium, the computer program containing program code for performing the methods shown in the flowcharts. In such embodiments, the computer program can be downloaded and installed from a network via a communication device, or installed from storage device 1003, or installed from read-only memory 1002. When the computer program is executed by processing device 1001, it performs the functions defined in the methods of the embodiments disclosed in this invention.

[0075] It should be understood that the various parts disclosed in this invention can be implemented using hardware, software, firmware, or a combination thereof. In the description of the above embodiments, specific features, structures, materials, or characteristics can be combined in any suitable manner in one or more embodiments or examples.

[0076] Obviously, those skilled in the art can make various modifications and variations to this invention without departing from its spirit and scope. Therefore, if these modifications and variations fall within the scope of the claims of this invention and their equivalents, this invention also intends to include these modifications and variations.

Claims

1. A method for quality inspection in the production of composite templates, characterized in that, The methods include: Step S10: Using an RGB industrial camera installed in the side detection area of ​​the plywood, images of the plywood side are acquired while the plywood is stationary or being conveyed at a constant speed, employing a self-feedback control mechanism based on multi-angle directional illumination and signal-to-noise ratio. ; Step S20: Based on the side image of the multi-layer plywood A phased image guidance mechanism is used to sequentially execute the channel selection task, the diffusion filtering task, and the dynamic narrowband ROI extraction task, and output the enhanced image G and the region of interest R; Step S30: In the region of interest R, the adaptive Canny algorithm is used in combination with orientation angle filtering and line segment merging mechanism to perform edge extraction and layer boundary line segment recognition, and output the initial line segment set L and the boundary line segment set. ; Step S40: Based on the set of boundary segments An offset vector function is established in physical space using a physical location fitting benchmark difference comparison method; wherein, based on the boundary line segment set... The steps for establishing an offset vector function in physical space using the physical location fitting benchmark difference comparison method specifically include: Selecting a set of boundary segments Line segment at the middle and end position As a reference layer, construct the reference layer position function corresponding to the reference layer. And based on the reference layer position function Construct a set of boundary segments The Middle Position offset function of line segment ; Construct a two-dimensional offset vector field by combining the position offset functions of all line segments. , Where m represents the number of line segments in the set of boundary line segments; For the first The position offset function of a line segment. For the first The position offset function of the line segment; and the two-dimensional offset vector field Output as an offset vector function; Step S50: Calculate the maximum offset based on the offset vector function and overall offset root mean square error Set the maximum offset threshold and root mean square error threshold ,like If the plywood is found to be of acceptable quality, then it is determined that the plywood has a potential misalignment.

2. The method for quality inspection of combined template production as described in claim 1, characterized in that, In step S10, a self-feedback control mechanism based on multi-angle directional illumination and signal-to-noise ratio is used to acquire side images of multi-layer plywood. The steps specifically include: First, construct a directional lighting field: Two sets of oblique LED line light sources are arranged on both sides of the multi-layer plywood. These two sets of oblique LED line light sources include a first oblique LED line light source and a second oblique LED line light source; wherein, the illumination angle of the first oblique LED line light source is... Illumination angle with the second oblique LED line light source satisfy and ;Lighting angle and lighting angle This is used to control two sets of oblique LED line light sources to face the plywood layer direction respectively, so as to form a uniform gradient projection effect and enhance the light and dark difference of the interlayer misalignment boundary. Then perform image signal-to-noise ratio evaluation and trigger self-feedback acquisition: based on illumination angle and lighting angle Start the RGB industrial camera to acquire initial images For the initial image Perform signal-to-noise ratio (SNR) evaluation and output an SNR evaluation score. , ;in, For the initial image Average brightness in the edge area of ​​multi-layer plywood For the initial image Standard deviation of luminance in the edge area of ​​multi-layer plywood; preset signal-to-noise ratio evaluation score threshold. ,like Then confirm the initial image. It is in a valid state; if This triggers the self-feedback control mechanism, restarting the RGB industrial camera to acquire the initial image. ; Finally, a linear scaling method based on the imaging geometry calibration model is used to establish the mapping relationship between pixels and actual length. Based on this mapping relationship, the final side image of the multi-layer plywood is output. .

3. The method for quality inspection of combined template production as described in claim 1, characterized in that, In step S20, based on the side image of the multi-layer plywood The process employs a phased image guidance mechanism, sequentially executing channel selection, diffusion filtering, and dynamic narrowband ROI extraction tasks, outputting an enhanced image G and a region of interest R. Specifically, this includes: Step S201: Image of the side surface of the multi-layer plywood within a preset edge pre-selection area. Calculate the edge gradient response intensities of the three channels, including the edge gradient response intensities of the first, second, and third channels; determine the optimized channel image based on the edge gradient response intensities of the three channels. ; Step S202: Based on the preset PM diffusion model, an edge response adaptive diffusion control mechanism is used to optimize the channel image. Perform iterative updates to obtain an edge-enhanced image; Step S203: Extract dynamic narrowband regions based on the edge enhancement image using vertical gradient peak localization and bandwidth limitation control mechanism, and output the enhanced image G and the region of interest R.

4. The method for quality inspection of combined template production as described in claim 3, characterized in that, In step S202, an edge-response adaptive diffusion control mechanism is used to optimize the channel image based on a preset PM diffusion model. The formula used in the iterative update step is: ; This is a two-dimensional divergence operator used to describe the spatial distribution variation of the diffusion flow in the PM diffusion model; where, A time marker for the number of iterations used to perform the iterative update; To optimize channel images The rate of change along the time dimension of the iteration number; The diffusion control operator introduced for the edge response adaptive diffusion control mechanism; This is the diffusion control coefficient; This is the edge sensitivity control factor.

5. The method for quality inspection of combined template production as described in claim 3, characterized in that, Step S203, which involves extracting a dynamic narrowband region based on the vertical gradient peak localization and bandwidth limiting control mechanism of the edge enhancement image, and outputting the enhanced image G and the region of interest R, specifically includes: First, in the edge enhancement image, the vertical grayscale gradient magnitude of each row of pixels is calculated row by row along the vertical direction, and the total intensity of the corresponding row gradient is counted. The position with the largest total intensity of the vertical gradient is marked as the vertical center coordinate, which is used as the response center of the misaligned boundary in the edge enhancement image. Next, a narrow-band window of fixed width is set based on the vertical center coordinates. The upper and lower boundaries of the narrow-band window are distanced from the vertical center coordinates. The bandwidth length determines the dynamic narrowband region; Finally, the edge-enhanced image is output as the enhanced image G, and the dynamic narrowband region is output as the region of interest R.

6. The method for quality inspection of combined template production as described in claim 1, characterized in that, In step S30, within the region of interest R, an adaptive Canny algorithm is used in conjunction with orientation angle filtering and a line segment merging mechanism to perform edge extraction and layer boundary line segment recognition, outputting an initial line segment set L and a boundary line segment set. The steps specifically include: Step S301: Extract the edge map in the region of interest R using the adaptive Canny algorithm, and calculate the horizontal edge response based on the edge map. Horizontal edge response in edge map exist and The edge pixels between them form a set of edge segments after directional filtering; Step S302: Based on the edge line segment set, the line segment detection algorithm LSD is used to extract the initial line segment set L. Each line segment in the initial line segment set L is represented in the form of a parameter pair, which includes the parameter pair consisting of the slope and intercept of the line segment. Step S303: For line segment pairs in the initial line segment set L that satisfy the preset collinearity condition, perform a merging and reconstruction operation to obtain the interlayer boundary line segment set. , ;in, This is the first interlayer boundary segment. This is the second interlayer boundary segment. This is the nth interlayer boundary segment, where n is the number of layers in the plywood minus 1; Step S304: Output the initial set of line segments L and the set of boundary line segments. .

7. A composite template production quality inspection system, applied to the composite template production quality inspection method according to any one of claims 1 to 6, characterized in that, The combined template production quality inspection system includes: The image acquisition module is used to acquire images of the side of the plywood using an RGB industrial camera installed on the side detection area of ​​the plywood, whether the plywood is stationary or being conveyed at a constant speed. The acquisition employs a self-feedback control mechanism based on multi-angle directional illumination and signal-to-noise ratio. ; Image preprocessing and ROI extraction module for images based on the side views of multi-layer plywood. A phased image guidance mechanism is used to sequentially execute the channel selection task, the diffusion filtering task, and the dynamic narrowband ROI extraction task, and output the enhanced image G and the region of interest R; The edge and boundary segment extraction module is used to perform edge extraction and layer boundary segment recognition in the region of interest R using an adaptive Canny algorithm combined with orientation angle filtering and segment merging mechanism, and outputs an initial segment set L and a boundary segment set. ; The physical offset modeling module is used to model based on the set of boundary segments. An offset vector function is established in physical space using a physical location fitting benchmark difference comparison method; wherein, based on the boundary line segment set... The steps for establishing an offset vector function in physical space using the physical location fitting benchmark difference comparison method specifically include: Selecting a set of boundary segments Line segment at the middle and end position As a reference layer, construct the reference layer position function corresponding to the reference layer. And based on the reference layer position function Construct a set of boundary segments The Middle Position offset function of line segment ; Construct a two-dimensional offset vector field by combining the position offset functions of all line segments. , Where m represents the number of line segments in the set of boundary line segments; For the first The position offset function of a line segment. For the first The position offset function of the line segment; and the two-dimensional offset vector field Output as an offset vector function; The quality assessment module is used to calculate the maximum offset based on the offset vector function. and overall offset root mean square error Set the maximum offset threshold and root mean square error threshold ,like If the plywood is found to be of acceptable quality, then it is determined that the plywood has a potential misalignment.

8. A quality inspection device for combined template production, characterized in that, The combined template production quality inspection equipment includes: a memory, a processor, and a combined template production quality inspection program stored in the memory and executable on the processor. When the combined template production quality inspection program is executed by the processor, it implements a combined template production quality inspection method according to any one of claims 1 to 6.

9. A computer program product, characterized in that, The computer program product includes a modular template production quality inspection program, which, when executed by a processor, implements a modular template production quality inspection method according to any one of claims 1 to 6.

Images