Real-time evaluation method for roughening effect of aluminum-plastic composite formwork

Abstract

The invention discloses a real-time evaluation method for the roughening effect of an aluminum-plastic composite formwork, and specifically relates to the technical field of aluminum-plastic composite formwork. In the acquisition stage, micro-vibration information is recorded in real time through image data from multiple angles and multiple light sources combined with a high-precision vibration sensor. Then, the image data and vibration data are preprocessed and feature extracted to generate texture distribution abnormality features and vibration frequency fluctuation features, and these features are input into a machine learning model for training to evaluate the accuracy of recognition. When the recognition accuracy is insufficient, the system re-acquires multiple frames of images, realizes inter-frame alignment and image superposition through vibration data, optimizes the quality of the synthesized image, and thus ensures accurate evaluation of the roughening effect. The invention effectively improves the accuracy of image recognition in a low-light and high-vibration environment, and ensures the stability and safety of the construction structure.

Description

Technical Field

[0001] The invention relates to the technical field of aluminum-plastic composite templates, and in particular to a real-time evaluation method for roughening effect of an aluminum-plastic composite template. Background Art

[0002] The real-time evaluation of the roughening effect of aluminum-plastic composite formwork is a detection process for the surface quality of building formwork, which aims to ensure the quality and performance of the formwork by real-time monitoring and evaluating the roughening effect. In construction, aluminum-plastic composite formwork is a common material, and its surface usually needs to be roughened to enhance the adhesion of concrete, thereby improving the use effect and safety of the formwork. The real-time evaluation system can monitor the roughening effect in real time through image recognition, sensor measurement or automated detection technology, and identify whether the surface roughness meets the predetermined standard. This can not only improve the detection efficiency, but also ensure the stability of construction quality and avoid problems such as poor concrete bonding caused by insufficient roughening effect.

[0003] The prior art has the following deficiencies:

[0004] In some complex construction sites (such as bridges and subway tunnels), there may be micro-vibrations or imperceptible ground vibrations. This vibration will cause the sensor to move or shake slightly, making the image slightly blurred, especially in low light conditions. And long-term micro-vibration blur will affect the accuracy of the image recognition system, making it difficult to accurately identify roughness details. This may cause the surface roughening effect to fail to meet the standard but be mistakenly considered qualified, seriously affecting the stability and safety of the structure. Summary of the invention

[0005] The purpose of the present invention is to provide a real-time evaluation method for the roughening effect of an aluminum-plastic composite template to solve the shortcomings of the background technology.

[0006] In order to achieve the above object, the present invention provides the following technical solution: a real-time evaluation method for roughening effect of aluminum-plastic composite template, comprising the following steps:

[0007] S1: Use a high-resolution camera to capture images of the roughened surface, and obtain image data from multiple angles and multiple light sources during the acquisition process. Install a high-precision vibration sensor on the camera to monitor the micro-vibration data to which the camera is subjected during the acquisition process.

[0008] S2: preprocessing the acquired multi-angle, multi-light source image data and micro-vibration data, performing feature extraction on the preprocessed image data and micro-vibration data, respectively extracting texture distribution abnormality features in the image data and vibration frequency fluctuation features in the micro-vibration data;

[0009] S3: Convert the abnormal texture distribution features and vibration frequency fluctuation features into comprehensive feature vectors, input them into the machine learning model for training, and determine the accuracy of the image recognition system in identifying the roughness details of the roughening effect based on the model output results;

[0010] S4: When the image recognition system has low accuracy in identifying the roughness details of the roughening effect, multiple frames of images are collected, and each frame of image is aligned and superimposed based on the vibration data of the vibration sensor, and the quality of the composite image is optimized to improve the accuracy of the roughening effect evaluation.

[0011] Preferably, in S2, a texture distribution anomaly index is generated according to the texture distribution anomaly features in the extracted image data, and the method for obtaining the texture distribution anomaly index is:

[0012] Grayscale the image and quantize it, quantize the image into L gray levels, construct the gray level co-occurrence matrix, and set the value of each pixel in the image is the gray-level co-occurrence matrix element, indicating that at the specified offset Under this condition, the co-occurrence frequency of gray levels between adjacent pixel pairs is normalized to the gray level co-occurrence matrix, and the gray level co-occurrence matrix is ​​normalized to Represents the normalized gray-level co-occurrence matrix: Extract texture features through the gray-level co-occurrence matrix and extract the contrast in texture features , reflects the size of the gray value difference, the expression is: ; Extract the homogeneity of texture features, which reflects the uniformity of grayscale distribution. The expression is: ; Extract entropy from texture features and measure the randomness of texture. The expression is: ; Where: ϵ is a small numerical compensation, which extracts the energy Energy in the texture feature, and the expression is: ; Calculate the texture distribution anomaly index, the expression is: ; In the formula, is the weight coefficient of the feature, is the texture distribution anomaly index.

[0013] Preferably, in S2, a vibration frequency fluctuation index is generated according to the vibration frequency fluctuation characteristics in the extracted micro-vibration data, and the method for obtaining the vibration frequency fluctuation index is:

[0014] Get the time series signal collected by the vibration sensor, expressed as x(t), where t is time, denoise the vibration signal, and perform fast Fourier transform on the pre-processed vibration signal x(t). Transformation is performed to convert the signal from the time domain to the frequency domain. The resulting frequency domain signal is represented as X(f) and the calculation expression is: ; Identify the main frequency from the frequency domain signal X(f) and all frequency components with significant amplitude , main frequency is the frequency component with the largest amplitude, and the calculation expression is: ; Calculate each frequency component With main frequency The deviation value SF between , the vibration frequency fluctuation index is calculated using the deviation values ​​of all frequency components. The expression is: ; Where: N represents the number of all significant frequency components, It is the vibration frequency fluctuation index.

[0015] Preferably, in S3, the texture distribution anomaly index and the vibration frequency fluctuation index are converted into a comprehensive feature vector, and the comprehensive feature vector is used as the input of the machine learning model. The machine learning model uses each group of comprehensive feature vectors to predict the accuracy value label of the image recognition system in identifying the roughness details of the roughened effect as the prediction target, and takes minimizing the sum of the prediction errors of the accuracy value labels of the roughness details of the roughened effect identified by all image recognition systems as the training target. The machine learning model is trained until the sum of the prediction errors converges and the model training is stopped. The accuracy value of the image recognition system in identifying the roughness details of the roughened effect is determined according to the model output results, wherein the machine learning model is a polynomial regression model.

[0016] Preferably, in S3, the acquired accuracy value of the image recognition system for identifying the roughness details of the roughening effect is compared with the gradient standard threshold, the gradient standard threshold includes a first standard threshold and a second standard threshold, and the first standard threshold is less than the second standard threshold, and the accuracy value of the image recognition system for identifying the roughness details of the roughening effect is compared with the first standard threshold and the second standard threshold respectively;

[0017] If the accuracy value of the image recognition system in identifying the roughness details of the roughness effect is greater than the second standard threshold, it means that the image recognition system has high accuracy in identifying the roughness details of the roughness effect, and a high-accuracy recognition signal is generated at this time;

[0018] If the accuracy value of the image recognition system in identifying the roughness details of the roughened effect is greater than or equal to the first standard threshold and less than or equal to the second standard threshold, it means that the image recognition system has high accuracy in identifying the roughness details of the roughened effect, and a medium accuracy recognition signal is generated at this time;

[0019] If the accuracy value of the image recognition system in identifying the roughness details of the roughened effect is less than the first standard threshold, it means that the image recognition system has high accuracy in identifying the roughness details of the roughened effect, and a low accuracy recognition signal is generated.

[0020] Preferably, in S4, it is set to collect M frames of images within a fixed time period, and each frame of image is represented by ,in , x and y are the pixel position coordinates of the image. A high-precision vibration sensor is used to collect the vibration displacement information of the camera when each frame of the image is captured. For each frame , and its corresponding vibration displacement is , indicating the offset in the x and y directions;

[0021] Based on the displacement data of the vibration sensor, each frame of the image Displacement compensation is performed to align all frame images. The aligned images are expressed as , the expression is: ; Overlay and synthesize the aligned multiple frames to create a composite image It is obtained by averaging and superposition, and the expression is: .

[0022] Preferably, the Laplacian operator is used to enhance the edge details in the synthesized image, and the expression is: ;in, The Laplace operator represents the synthesized image, λ is the sharpening coefficient, and the final optimized image is enhanced by histogram equalization. The roughness details used by the image recognition system to re-evaluate the roughness effect, the synthesized and optimized image expression is: ; In the formula, the HistogramEqualization function is used to adjust the image contrast.

[0023] In the above technical solution, the technical effects and advantages provided by the present invention are:

[0024] 1. The present invention realizes real-time and accurate evaluation of the roughening effect of aluminum-plastic composite formwork by introducing high-resolution image acquisition, vibration data monitoring and multi-angle and multi-light source acquisition methods. The abnormal texture features and vibration frequency fluctuation features of the image are extracted using grayscale co-occurrence matrix and fast Fourier transform, and converted into comprehensive feature vectors and input into the machine learning model for training. The model predicts the recognition accuracy value and judges the recognition accuracy level through the gradient standard threshold, ensuring the reliability of the recognition result. When the recognition accuracy is low, the system automatically enhances the image quality through multi-frame image acquisition, vibration compensation alignment and superposition optimization, thereby effectively improving the accuracy of roughening effect recognition.

[0025] 2. The present invention effectively addresses the image blurring problem caused by micro-vibration in complex construction environments, and optimizes the details of the synthesized image through Laplace operator sharpening and histogram equalization, significantly improving the system's ability to recognize roughness details. This efficient real-time evaluation method not only improves the monitoring accuracy of construction quality, reduces the misjudgment of unqualified roughening effects, but also ensures the stability and safety of the construction structure. BRIEF DESCRIPTION OF THE DRAWINGS

[0026] In order to more clearly illustrate the embodiments of the present application or the technical solutions in the prior art, the drawings required for use in the embodiments will be briefly introduced below. Obviously, the drawings described below are only some embodiments recorded in the present invention. For ordinary technicians in this field, other drawings can also be obtained based on these drawings.

[0027] Figure 1 The present invention is a flow chart of the method. DETAILED DESCRIPTION

[0028] In order to make the purpose, technical solution and advantages of the embodiments of the present invention clearer, the technical solution in the embodiments of the present invention will be clearly and completely described below in conjunction with the drawings in the embodiments of the present invention. Obviously, the described embodiments are part of the embodiments of the present invention, not all of the embodiments. Based on the embodiments of the present invention, all other embodiments obtained by ordinary technicians in this field without creative work are within the scope of protection of the present invention.

[0029] For examples, see Figure 1 As shown, the real-time evaluation method for roughening effect of aluminum-plastic composite template described in this embodiment includes the following steps:

[0030] S1: Use a high-resolution camera to capture images of the roughened surface, and obtain image data from multiple angles and multiple light sources during the acquisition process. Install a high-precision vibration sensor on the camera to monitor the micro-vibration data to which the camera is subjected during the acquisition process.

[0031] S2: preprocessing the acquired multi-angle, multi-light source image data and micro-vibration data, performing feature extraction on the preprocessed image data and micro-vibration data, respectively extracting texture distribution abnormality features in the image data and vibration frequency fluctuation features in the micro-vibration data;

[0032] S3: Convert the abnormal texture distribution features and vibration frequency fluctuation features into comprehensive feature vectors, input them into the machine learning model for training, and determine the accuracy of the image recognition system in identifying the roughness details of the roughening effect based on the model output results;

[0033] S4: When the image recognition system has low accuracy in identifying the roughness details of the roughening effect, multiple frames of images are collected, and each frame of image is aligned and superimposed based on the vibration data of the vibration sensor, and the quality of the composite image is optimized to improve the accuracy of the roughening effect evaluation.

[0034] In S1, a high-resolution camera is used to collect images of the roughened surface, and image data from multiple angles and multiple light sources are obtained during the collection process. A high-precision vibration sensor is installed on the camera to monitor the micro-vibration data to which the camera is subjected during the collection process. Specifically:

[0035] Use a camera with high pixel density for image acquisition to ensure that the fine roughness details of the roughened surface are captured. Adjust the camera angle (e.g. vertical, 45 degrees, horizontal, etc.) to capture images of the surface at different angles to obtain complete details of the roughness texture. A fixed robotic arm or a rotatable bracket can be used to gradually change the camera at a preset angle. Introduce different light source positions and lighting intensities to reduce interference from shadows and reflections. A combination of ring light, point light source, or diffuse light can be used to make the surface texture clearly present under various lighting conditions.

[0036] Install a high-precision vibration sensor on the camera device to monitor the micro-vibration of the camera in real time during the acquisition process. The resolution of the vibration sensor must be high enough to detect small changes in vibration amplitude and frequency. While each image is being acquired, the data of the current vibration state is obtained, including the amplitude (displacement), frequency, and direction of the vibration. These vibration parameters will be used for subsequent image alignment and blur compensation.

[0037] Each time a frame of image is acquired, the vibration data at the corresponding time point is recorded, and the image and vibration data are stored one by one. This synchronous acquisition method ensures the data relevance in subsequent processing, which facilitates accurate compensation and optimization of the image. Each set of images and vibration data is marked (for example, timestamp or frame number), and all data is stored in the database for subsequent feature extraction, vibration compensation and recognition analysis.

[0038] S2: Preprocess the acquired multi-angle, multi-light source image data and micro-vibration data, perform feature extraction on the preprocessed image data and micro-vibration data, and respectively extract texture distribution abnormality features in the image data and vibration frequency fluctuation features in the micro-vibration data.

[0039] Since images at different angles may have displacement and rotation differences, these images need to be aligned first for subsequent synthesis and analysis. Use image registration technology (such as feature point matching, phase correlation method) to align images at different angles. Find common feature points (such as edges or corners) in the images, calculate the relative offset and rotation angle of each image, and adjust the images to the same position.

[0040] Images collected using different light sources may have differences in brightness and shadows. Fusion processing can be used to reduce the interference caused by uneven lighting and ensure that surface details are clearly presented. The high dynamic range (HDR) synthesis method is used to fuse multi-light source images according to pixel brightness information to enhance brightness and contrast while keeping image details intact. HDR synthesis can reduce highlight reflections and shadow areas, making the rough texture more clearly visible.

[0041] Noise (such as high-frequency noise, light interference, etc.) may be generated during image acquisition, and denoising is required to improve image quality. Use denoising techniques such as Gaussian filtering and median filtering to reduce noise levels while retaining texture details, making the surface texture smoother and easier to identify subsequent details.

[0042] In order to enhance the edge clarity of the texture, the Laplacian operator or Unsharp Masking is applied to sharpen the image and enhance the edge details of the image. This can better highlight the rough texture and make the surface roughness details more obvious, which is convenient for accurate recognition.

[0043] The vibration data collected by the vibration sensor may contain noise or discontinuous high-frequency signals, which will affect the judgment of the vibration situation. Use a low-pass filter or Kalman filter to smooth the vibration data, remove meaningless high-frequency noise, retain the main trend information of the vibration (such as vibration frequency and amplitude), and ensure that the data is stable and reliable.

[0044] Analyze the vibration data and extract important features, such as the peak value, mean value and change of vibration amplitude of vibration frequency, to quantify the impact of vibration on image clarity. Record the main characteristic information of vibration and establish association with image data to facilitate targeted compensation in the subsequent image processing process.

[0045] Ensure that each set of vibration data is synchronized with the corresponding image frame so that vibration compensation technology can be accurately applied. Use timestamps or frame numbers to align the image and vibration data to ensure that the vibration information at each moment can be accurately applied during subsequent processing.

[0046] Feature extraction is performed on the preprocessed image data and microvibration data, and a texture distribution anomaly index is generated according to the texture distribution anomaly features in the extracted image data. The method for obtaining the texture distribution anomaly index is as follows:

[0047] Grayscale the image (if it is a color image) and quantize it. Quantize the image into L gray levels to reduce the matrix size and reduce the computational complexity, construct the gray level co-occurrence matrix, and set the value of each pixel of the image is the gray-level co-occurrence matrix element, indicating that at the specified offset In this case, the co-occurrence frequency of gray levels between adjacent pixel pairs is usually taken as In order to avoid the influence of image size on the results, the gray level co-occurrence matrix is ​​normalized. Represents the normalized gray level co-occurrence matrix: Multiple typical texture features are extracted through the gray level co-occurrence matrix, which can be used to evaluate the abnormality of texture distribution in the image. Extracting contrast in texture features , reflects the size of the gray value difference, corresponding to the roughness of the texture. Areas with high contrast usually indicate high texture complexity, and the expression is: ; Extract the homogeneity of texture features, which reflects the uniformity of grayscale distribution. The larger the value, the more uniform the texture. The expression is: ; Extract the entropy from the texture features and measure the randomness of the texture. The higher the entropy value, the more complex the texture. The expression is: ; where: ϵ is a small numerical compensation to prevent the logarithm from taking zero, extracting the energy Energy in the texture feature, reflecting the regularity or repeatability of the texture distribution. The higher the value, the more consistent the texture. The expression is: ; Calculate the texture distribution anomaly index, the expression is: ; In the formula, is the weight coefficient of the feature, and the appropriate weight is selected according to the specific application and data distribution. is the texture distribution anomaly index.

[0048] The texture distribution anomaly index is used to measure the degree of abnormality of texture features in an image, and helps evaluate the accuracy of the image recognition system in identifying roughness details of the roughened effect. When the texture distribution anomaly index is large, it usually indicates that there is significant texture unevenness and complexity in the image, such as high texture contrast and strong randomness. This may mean that the texture features of the roughened surface are unstable, and the system is easily disturbed when identifying these details, so the recognition accuracy is low. A high anomaly index usually indicates that the system may have a risk of misjudgment, and the identified roughness value may deviate from the actual situation, affecting the accurate assessment of the roughened quality.

[0049] On the contrary, when the texture distribution anomaly index is small, it means that the texture distribution in the image is relatively uniform, and the texture features are stable and consistent. A low anomaly index indicates that the texture of the roughened surface is relatively smooth and low in contrast, or that the noise level after image processing is low, which provides the system with clearer and more stable texture data. At this time, the image recognition system can more accurately identify the roughness details with a low probability of misjudgment, thereby improving the accuracy of identifying the roughened effect. A low anomaly index means that the system's assessment of surface roughness is more reliable, which helps to more accurately determine whether the roughened effect meets the quality standards.

[0050] Feature extraction is performed on the preprocessed image data and micro-vibration data, and a vibration frequency fluctuation index is generated according to the vibration frequency fluctuation features in the extracted micro-vibration data. The method for obtaining the vibration frequency fluctuation index is:

[0051] Get the time series signal collected by the vibration sensor, expressed as x(t), where t is time, and denoise the vibration signal to reduce the impact of high-frequency noise. You can use a low-pass filter or a smoothing filter to preprocess the data to ensure signal quality. Perform fast Fourier transform on the preprocessed vibration signal x(t). Transformation is performed to convert the signal from the time domain to the frequency domain. The resulting frequency domain signal is represented as X(f) and the calculation expression is: ; Identify the main frequency from the frequency domain signal X(f) and all frequency components with significant amplitude , main frequency is the frequency component with the largest amplitude: The extracted frequency components are usually the main frequency and secondary frequency of the vibration signal, which represent the main frequency characteristics of the signal. Calculate each frequency component With main frequency The deviation value SF between The deviation value can reflect the fluctuation degree of each frequency component relative to the main frequency. The vibration frequency fluctuation index is calculated using the deviation values ​​of all frequency components. The expression is: ; Where: N represents the number of all significant frequency components, It is the vibration frequency fluctuation index.

[0052] When the vibration frequency fluctuation index is large, it indicates that the frequency components of the vibration signal fluctuate violently, and the frequency components such as the main frequency and the secondary frequency are unstable. The violent fluctuation of the frequency usually causes the camera to shake or shift slightly in different directions, making the collected image unclear. In this case, the image recognition system may not be able to accurately capture the subtle texture of the surface roughening effect, and the recognition results are prone to deviation, which reduces the recognition accuracy of the system.

[0053] On the contrary, when the vibration frequency fluctuation index is small, it indicates that the frequency component of the vibration signal is relatively stable and the vibration frequency does not change significantly. Small frequency fluctuations mean that the camera is less affected by vibrations, and the image clarity is higher, which helps the system to accurately identify the details of roughness. A lower vibration frequency fluctuation index usually makes the system more stable when acquiring images and has better image quality, thereby improving the accuracy and reliability of the image recognition system in identifying the roughness details of the roughening effect. Therefore, a low vibration frequency fluctuation index usually points to high recognition accuracy, while a high vibration frequency fluctuation index indicates that the recognition accuracy may be limited.

[0054] S3: The abnormal texture distribution characteristics and vibration frequency fluctuation characteristics are converted into comprehensive feature vectors, which are input into the machine learning model for training, and the accuracy of the image recognition system in identifying the roughness details of the roughening effect is determined based on the model output results.

[0055] The texture distribution anomaly index and the vibration frequency fluctuation index are converted into comprehensive feature vectors, and the comprehensive feature vectors are used as the input of the machine learning model. The machine learning model uses each group of comprehensive feature vectors to predict the accuracy value label of the image recognition system in identifying the roughness details of the roughened effect as the prediction target, and takes minimizing the sum of the prediction errors of the accuracy value labels of the roughness details of the roughened effect identified by all image recognition systems as the training target. The machine learning model is trained until the sum of the prediction errors converges, and the model training is stopped. The accuracy value of the image recognition system in identifying the roughness details of the roughened effect is determined according to the model output results, wherein the machine learning model is a polynomial regression model.

[0056] The method for obtaining the accuracy value of the roughness details of the roughening effect by the image recognition system is as follows: from the comprehensive feature vector training data of the trained machine learning model, the corresponding function expression is obtained: ; In the formula, is the output function of the model, HK is the texture distribution anomaly index, GH is the vibration frequency fluctuation index, A value for the accuracy of the image recognition system in identifying roughness details of the brushed effect.

[0057] The obtained accuracy value of the image recognition system in identifying the roughness details of the roughening effect is compared with the gradient standard threshold, the gradient standard threshold includes a first standard threshold and a second standard threshold, and the first standard threshold is less than the second standard threshold, and the accuracy value of the image recognition system in identifying the roughness details of the roughening effect is compared with the first standard threshold and the second standard threshold respectively;

[0058] If the accuracy value of the image recognition system in identifying the roughness details of the roughness effect is greater than the second standard threshold, it means that the image recognition system has high accuracy in identifying the roughness details of the roughness effect, and a high-accuracy recognition signal is generated at this time;

[0059] If the accuracy value of the image recognition system in identifying the roughness details of the roughened effect is greater than or equal to the first standard threshold and less than or equal to the second standard threshold, it means that the image recognition system has high accuracy in identifying the roughness details of the roughened effect, and a medium accuracy recognition signal is generated at this time;

[0060] If the accuracy value of the image recognition system in identifying the roughness details of the roughened effect is less than the first standard threshold, it means that the image recognition system has high accuracy in identifying the roughness details of the roughened effect, and a low accuracy recognition signal is generated.

[0061] S4: When the image recognition system has low accuracy in identifying the roughness details of the roughening effect, multiple frames of images are collected, and each frame of image is aligned and superimposed based on the vibration data of the vibration sensor, and the quality of the composite image is optimized to improve the accuracy of the roughening effect evaluation.

[0062] It is assumed that M frames of images are collected within a fixed time period, and each frame of image is represented as ,in , x and y are the pixel position coordinates of the image. A high-precision vibration sensor is used to collect the vibration displacement information of the camera when each frame of the image is captured. , and its corresponding vibration displacement is , indicating the offset in the x and y directions.

[0063] Based on the displacement data of the vibration sensor, each frame of the image Displacement compensation is performed to align all frame images. The aligned image is represented as , the expression is: ; Overlay the aligned multiple frames to enhance image details, reduce noise and minimize vibration. Composite image It can be obtained by averaging. ; Optimize the sharpness and contrast of the composite image to further enhance the texture details of the roughened effect.

[0064] The Laplacian operator is used to enhance the edge details in the synthesized image. The expression is: ;in, The Laplace operator represents the synthesized image, λ is the sharpening coefficient, and the image texture is made clearer by histogram equalization. The final optimized image It provides higher detail clarity, which can be used in the image recognition system to re-evaluate the roughness details of the roughening effect, improving the evaluation accuracy. The synthesized and optimized image expression is: ; In the formula, the HistogramEqualization function is used to adjust the image contrast.

[0065] In this embodiment, a high-resolution camera is used to capture images of the roughened surface from multiple angles and multiple light sources, and a high-precision vibration sensor installed on the camera is used to record micro-vibration data. Next, the captured images and vibration data are pre-processed to extract the abnormal texture distribution characteristics and vibration frequency fluctuation characteristics of the images, respectively, and these characteristics are converted into comprehensive feature vectors and input into the machine learning model for training to evaluate the recognition accuracy of the image recognition system on the roughness details of the roughened effect. If the recognition accuracy is low, the system re-captures multiple frames of images, aligns and superimposes each frame of image based on the vibration data, and performs optimization processing such as sharpening and contrast enhancement on the composite image to improve the evaluation accuracy of the roughened effect, thereby improving the overall recognition accuracy of the image recognition system.

[0066] The above formulas are all dimensionless and numerical calculations. The formula is a formula for the most recent real situation obtained by collecting a large amount of data and performing software simulation. The preset parameters in the formula are set by technicians in this field according to actual conditions.

[0067] The above embodiments can be implemented in whole or in part by software, hardware, firmware or any other combination. When implemented by software, the above embodiments can be implemented in whole or in part in the form of a computer program product. The computer program product includes one or more computer instructions or computer programs. When the computer instructions or computer programs are loaded or executed on a computer, the process or function described in the embodiment of the present application is generated in whole or in part. The computer can be a general-purpose computer, a special-purpose computer, a computer network, or other programmable device. The computer instructions can be stored in a computer-readable storage medium, or transmitted from one computer-readable storage medium to another computer-readable storage medium. For example, the computer instructions can be transmitted from one website site, computer, server or data center to another website site, computer, server or data center by wired or wireless (e.g., infrared, wireless, microwave, etc.). The computer-readable storage medium can be any available medium that a computer can access or a data storage device such as a server or data center that contains one or more available media sets. The available medium can be a magnetic medium (e.g., a floppy disk, a hard disk, a tape), an optical medium (e.g., a DVD), or a semiconductor medium. The semiconductor medium can be a solid-state hard disk.

[0068] It should be understood that the term "and / or" in this article is only a description of the association relationship of associated objects, indicating that there may be three relationships. For example, A and / or B can represent: A exists alone, A and B exist at the same time, and B exists alone, where A and B can be singular or plural. In addition, the character " / " in this article generally indicates that the objects associated with each other are in an "or" relationship, but it may also indicate an "and / or" relationship, which can be understood by referring to the context. A person of ordinary skill in the art can appreciate that the units and algorithm steps of each example described in conjunction with the embodiments disclosed in this article can be implemented in electronic hardware, or a combination of computer software and electronic hardware. Whether these functions are performed in hardware or software depends on the specific application and design constraints of the technical solution. Professional and technical personnel can use different methods to implement the described functions for each specific application, but such implementation should not be considered to exceed the scope of this application.

[0069] The above description is only a specific implementation manner of the present application, but the protection scope of the present application is not limited thereto. Any technician familiar with the technical field can easily think of changes or substitutions within the technical scope disclosed in the present application, which should be included in the protection scope of the present application.

Claims

1. A real-time evaluation method for roughening effect of aluminum-plastic composite formwork, characterized by: The steps include: S1: Use a high-resolution camera to capture images of the roughened surface, and obtain image data from multiple angles and multiple light sources during the acquisition process. Install a high-precision vibration sensor on the camera to monitor the micro-vibration data to which the camera is subjected during the acquisition process. S2: preprocessing the acquired multi-angle, multi-light source image data and micro-vibration data, performing feature extraction on the preprocessed image data and micro-vibration data, respectively extracting texture distribution abnormality features in the image data and vibration frequency fluctuation features in the micro-vibration data; S3: Convert the abnormal texture distribution features and vibration frequency fluctuation features into comprehensive feature vectors, input them into the machine learning model for training, and determine the accuracy of the image recognition system in identifying the roughness details of the roughening effect based on the model output results; S4: When the image recognition system has low accuracy in identifying the roughness details of the roughening effect, multiple frames of images are collected, and each frame of image is aligned and superimposed based on the vibration data of the vibration sensor, and the quality of the composite image is optimized to improve the accuracy of the roughening effect evaluation.

2. The real-time evaluation method for roughening effect of aluminum-plastic composite formwork according to claim 1 is characterized in that: In S2, a texture distribution anomaly index is generated according to the texture distribution anomaly features in the extracted image data, and the method for obtaining the texture distribution anomaly index is: Grayscale the image and quantize it. Quantize the image into L gray levels, construct a gray-level co-occurrence matrix, and set each pixel value of the image to is the gray-level co-occurrence matrix element, indicating that at the specified offset Under this condition, the co-occurrence frequency of gray levels between adjacent pixel pairs is normalized to the gray level co-occurrence matrix, and the gray level co-occurrence matrix is ​​normalized to Represents the normalized gray-level co-occurrence matrix: Extract texture features through the gray-level co-occurrence matrix and extract the contrast in texture features , reflects the size of the gray value difference, the expression is: ; Extract the homogeneity of texture features, which reflects the uniformity of grayscale distribution. The expression is: ; Extract entropy from texture features and measure the randomness of texture. The expression is: ; Where: ϵ is a small numerical compensation, which extracts the energy Energy in the texture feature, and the expression is: ; Calculate the texture distribution anomaly index, the expression is: ; In the formula, is the weight coefficient of the feature, is the texture distribution anomaly index.

3. The real-time evaluation method for roughening effect of aluminum-plastic composite formwork according to claim 2 is characterized in that: In S2, a vibration frequency fluctuation index is generated according to the vibration frequency fluctuation characteristics in the extracted micro-vibration data. The method for obtaining the vibration frequency fluctuation index is: Get the time series signal collected by the vibration sensor, expressed as x(t), where t is time, denoise the vibration signal, and perform fast Fourier transform on the pre-processed vibration signal x(t). Transformation is performed to convert the signal from the time domain to the frequency domain. The resulting frequency domain signal is represented as X(f) and the calculation expression is: ; Identify the dominant frequency from the frequency domain signal X(f) and all frequency components with significant amplitude , main frequency is the frequency component with the largest amplitude, and the calculation expression is: ; Calculate each frequency component With main frequency The deviation value SF between , the vibration frequency fluctuation index is calculated using the deviation values ​​of all frequency components. The expression is: ; Where: N represents the number of all significant frequency components, It is the vibration frequency fluctuation index.

4. The real-time evaluation method for roughening effect of aluminum-plastic composite formwork according to claim 3 is characterized in that: In S3, the texture distribution anomaly index and the vibration frequency fluctuation index are converted into comprehensive feature vectors, and the comprehensive feature vectors are used as inputs of the machine learning model. The machine learning model predicts the accuracy value labels of the roughness details of the roughened effect predicted by the image recognition system for each group of comprehensive feature vectors as prediction targets, and minimizes the sum of prediction errors of the accuracy value labels of the roughness details of the roughened effect recognized by all image recognition systems as training targets. The machine learning model is trained until the sum of prediction errors converges, and the model training is stopped. The accuracy value of the image recognition system in recognizing the roughness details of the roughened effect is determined according to the model output results, wherein the machine learning model is a polynomial regression model.

5. The real-time evaluation method for roughening effect of aluminum-plastic composite formwork according to claim 4 is characterized in that: In S3, the accuracy value of the image recognition system in identifying the roughness details of the roughening effect is compared with the gradient standard threshold, the gradient standard threshold includes a first standard threshold and a second standard threshold, and the first standard threshold is less than the second standard threshold, and the accuracy value of the image recognition system in identifying the roughness details of the roughening effect is compared with the first standard threshold and the second standard threshold respectively; If the accuracy value of the image recognition system in identifying the roughness details of the roughness effect is greater than the second standard threshold, it means that the image recognition system has high accuracy in identifying the roughness details of the roughness effect, and a high-accuracy recognition signal is generated at this time; If the accuracy value of the image recognition system in identifying the roughness details of the roughened effect is greater than or equal to the first standard threshold and less than or equal to the second standard threshold, it means that the image recognition system has high accuracy in identifying the roughness details of the roughened effect, and a medium accuracy recognition signal is generated at this time; If the accuracy value of the image recognition system in identifying the roughness details of the roughened effect is less than the first standard threshold, it means that the image recognition system has high accuracy in identifying the roughness details of the roughened effect, and a low accuracy recognition signal is generated.

6. The real-time evaluation method for roughening effect of aluminum-plastic composite formwork according to claim 1 is characterized in that: In S4, it is assumed that M frames of images are collected within a fixed time period, and each frame of image is represented as ,in , x and y are the pixel position coordinates of the image. A high-precision vibration sensor is used to collect the vibration displacement information of the camera when each frame of the image is captured. For each frame , and its corresponding vibration displacement is , represents the offset in the x and y directions; Based on the displacement data of the vibration sensor, each frame of the image Displacement compensation is performed to align all frame images. The aligned images are expressed as , the expression is: ; Overlay and synthesize the aligned multiple frames to create a composite image It is obtained by averaging and superposition, and the expression is: .

7. The real-time evaluation method for roughening effect of aluminum-plastic composite formwork according to claim 6 is characterized in that: The Laplacian operator is used to enhance the edge details in the synthesized image. The expression is: ;in, The Laplace operator represents the synthesized image, λ is the sharpening coefficient, and the final optimized image is enhanced by histogram equalization. The roughness details used by the image recognition system to re-evaluate the roughness effect, the synthesized and optimized image expression is: ; In the formula, the HistogramEqualization function is used to adjust the image contrast.

Images