Building decorative plate production quality rapid detection method based on machine vision

Abstract

The invention relates to the technical field of building decorative plate production detection, and discloses a building decorative plate production quality rapid detection method based on machine vision. The method comprises the following steps: firstly, acquiring a surface image, an edge image and a texture feature sequence of each visual detection unit in each detection period; obtaining a feature deviation index of an abnormal point based on a correlation deviation condition of the abnormal point in the texture feature sequence, and further obtaining an average deviation index and a quality fluctuation degree at a quality fluctuation collection moment, thereby determining a surface quality coefficient; obtaining an edge quality coefficient based on the surface and edge images; and finally, according to the surface and edge quality coefficients, obtaining a detection parameter adjustment value which is used for correcting a detection parameter and a threshold value for template matching defect identification in the next period. According to the method, the dynamic adjustment of the detection parameters is realized, and the accuracy of defect identification is improved.

Description

Technical Field

[0001] The present invention relates to the technical field of production and detection of building decorative panels, and in particular to a method for rapid production quality detection of building decorative panels based on machine vision. Background Art

[0002] In the production process of architectural decorative panels, product quality inspection is a crucial step in ensuring their market competitiveness. As the construction industry's requirements for the appearance and performance of decorative materials continue to rise, traditional manual inspection methods are increasingly unable to meet the demands of large-scale industrial production. Manual inspection not only relies on the experience and commitment of inspectors, but also easily leads to fatigue after long hours of work, resulting in reduced inspection efficiency and increased false positives. This is particularly true when the surface textures of decorative panels are complex and the types of defects are diverse, significantly compromising the accuracy of manual inspection.

[0003] Currently, machine vision-based inspection technology has been applied in multiple industrial fields and is gradually being promoted in the inspection of building decorative panels. However, existing machine vision inspection methods still have many limitations in practical applications. Most inspection systems use fixed detection parameters and thresholds for defect identification, which makes it difficult to adapt to quality fluctuations caused by changes in material properties and environmental factors during the production process. For example, in the production of decorative panels, factors such as raw material batch differences, fluctuations in production line speeds, and changes in lighting conditions can cause the same type of defect to appear different in the image. Template matching with fixed parameters is prone to missed detections or false detections.

[0004] Existing techniques for analyzing the surface texture of decorative panels often rely solely on the extraction and assessment of single features, ignoring the correlations between texture features and the differences between local and overall textures. Texture defects on the surface of decorative panels can manifest as localized texture disturbances or abnormal fluctuations in feature values. These subtle variations are difficult to accurately capture using a single threshold. Furthermore, edge quality, a crucial indicator of the installation accuracy and appearance integrity of decorative panels, is a key factor influencing the appearance of the panels. Existing edge image processing methods focus on simple geometric parameter measurements and lack comprehensive analysis of the correlation between edge defects and surface quality, resulting in incomplete detection results.

[0005] During the continuous operation of a production line, the stability of the detection system directly impacts the reliability of the test results. When production conditions change, such as equipment wear or raw material replacement, the original detection parameters may no longer apply. Failure to adjust the detection thresholds and parameters in a timely manner can lead to a large number of substandard products entering the market or misjudgment of qualified products, increasing production costs and wasting resources. Traditional parameter adjustment methods often rely on regular manual calibration, which is not only slow to respond but also difficult to accurately adjust based on real-time production data, failing to meet the requirements of efficient production.

[0006] Furthermore, existing machine vision inspection methods are deficient in predicting and tracing quality fluctuations. When defects are detected, they are often limited to the current product, failing to identify potential risks of quality fluctuations in advance by analyzing trends in texture features in historical data. This hinders proactive quality control on the production line. Furthermore, there is a lack of effective quantitative metrics to assess the severity of detected abnormal data, making it difficult to prioritize quality issues and impacting the timeliness and relevance of production adjustments. Summary of the Invention

[0007] The purpose of the present invention is to provide a method for quickly detecting the production quality of building decorative panels based on machine vision, so as to solve the problems raised in the above background technology.

[0008] To achieve the above object, the present invention provides a method for rapid detection of production quality of building decorative panels based on machine vision, the method comprising the following steps:

[0009] Collecting the surface image and edge image of each visual detection unit at each collection moment in each detection cycle and each texture feature sequence of each visual detection unit in each detection cycle;

[0010] Based on the deviation between the abnormal points in the texture feature sequence, the gradient of the texture feature sequence and the deviation between the means of each local texture feature sequence, the characteristic deviation index of each abnormal point in each texture feature sequence of each visual detection unit in each detection cycle is obtained;

[0011] Based on the distribution of the characteristic deviation index, the average deviation index of each visual inspection unit at each quality fluctuation collection moment in each inspection cycle is obtained;

[0012] Based on the similarity and average deviation index between local texture feature data sequences, the quality fluctuation degree of each visual inspection unit at each quality fluctuation collection moment in each inspection cycle is obtained;

[0013] Obtaining the surface quality coefficient of each visual inspection unit in each inspection cycle based on the degree of quality fluctuation;

[0014] Obtaining edge quality coefficients of each visual inspection unit in each inspection cycle based on the surface image and the edge image;

[0015] Based on the surface quality coefficient and the edge quality coefficient, the adjustment value of the detection parameter of each visual inspection unit in each inspection cycle is obtained. The adjustment value of the detection parameter of each visual inspection unit in each inspection cycle is used as the value of the detection parameter of each visual inspection unit when using template matching for defect identification in the next adjacent cycle of each inspection cycle, and the detection threshold is corrected.

[0016] Preferably, the surface image and edge image include:

[0017] For each visual inspection unit, a surface image at each acquisition moment is acquired by a line array camera as the first surface image at each acquisition moment, and an edge image at each acquisition moment is acquired with the assistance of a ring light source as the first edge image at each acquisition moment; a surface image and an edge image at each acquisition moment are acquired by an area array camera as the second surface image and the second edge image at each acquisition moment.

[0018] Preferably, the method for obtaining the characteristic deviation index is:

[0019] For each texture feature sequence of each visual detection unit in each detection cycle, a grayscale threshold segmentation algorithm is used to obtain all the highlight areas and dark areas in each texture feature sequence of each detection cycle as each outlier. For each outlier, the absolute value of the difference between the texture feature data of the outlier and the texture feature data of the previous adjacent acquisition moment is calculated as the forward deviation index of each outlier in each texture feature sequence of each visual detection unit in each detection cycle.

[0020] Obtain the clustering index of each abnormal point based on the gradient of the texture feature data sequence;

[0021] For each outlier point in each texture feature sequence of each visual detection unit in each detection cycle, a window of preset size is constructed with the outlier point as the center as the local window of each outlier point. The mean of all texture feature data before the outlier point in the local window is calculated, and the absolute value of the difference between the mean of all texture feature data after the outlier point in the local window is used as the mean deviation of each outlier point.

[0022] The characteristic deviation index is calculated as a weighted combination of the forward deviation index, the clustering index and the forward and backward mean deviation, and each weight is a preset adjustment coefficient.

[0023] Preferably, the clustering index is obtained by:

[0024] For each texture feature sequence of each visual detection unit in each detection cycle, the maximum value of the gradient amplitude of all elements in the texture feature sequence is calculated as the maximum gradient value of the texture feature sequence;

[0025] For each abnormal point in each texture feature sequence of each visual detection unit in each detection cycle, the absolute value of the difference between the texture feature data of the abnormal point and the gradient maximum value is calculated as the clustering index of each abnormal point.

[0026] Preferably, the method for obtaining the mean deviation index is:

[0027] For each abnormal point in each texture feature sequence of each visual detection unit in each detection cycle, when the texture feature data in all other types of texture feature sequences at the acquisition moment where the abnormal point is located are all abnormal points, the acquisition moment is used as the quality fluctuation acquisition moment, and the average of the characteristic deviation indexes of the abnormal points of all types of texture feature sequences at the quality fluctuation acquisition moment of each detection cycle is calculated as the average deviation index of each visual detection unit at each quality fluctuation acquisition moment of each detection cycle.

[0028] Preferably, the method for obtaining the degree of quality fluctuation is:

[0029] At each quality fluctuation collection moment in each detection cycle, the K-means clustering algorithm is used to cluster the average deviation index of all visual inspection units to obtain each cluster; the cluster where the average deviation index of each visual inspection unit is located is used as the target cluster of each visual inspection unit;

[0030] For the texture feature data of each visual inspection unit in each texture feature sequence in each inspection cycle at each quality fluctuation collection moment, a sequence consisting of all texture feature data in a local window of the texture feature data is used as a local window sequence at each quality fluctuation collection moment;

[0031] For each visual detection unit at each quality fluctuation collection moment in each detection cycle, calculate the mean of the similarities between the local window sequence of each texture feature sequence and the local window sequences of the same type of texture feature sequences of all other visual detection units in its target cluster, as the feature trend difference of each texture feature sequence of each visual detection unit at each quality fluctuation collection moment in each detection cycle, and take the mean of the feature trend differences of all types of texture feature sequences of each visual detection unit at each quality fluctuation collection moment in each detection cycle as the feature dissimilarity index of each visual detection unit at each quality fluctuation collection moment in each detection cycle;

[0032] The degree of quality fluctuation is calculated as a weighted combination of the average deviation index, the number of target cluster elements, and the characteristic dissimilarity index, with each weight being a preset adjustment factor.

[0033] Preferably, the surface quality coefficient is obtained by:

[0034] The quality fluctuation degree of each visual inspection unit at all quality fluctuation collection moments in each inspection cycle is used as the input of the Otsu threshold segmentation algorithm, and the optimal segmentation threshold is output. The quality fluctuation collection moment when the quality fluctuation degree is greater than or equal to the optimal segmentation threshold is regarded as the severe fluctuation collection moment, and the quality fluctuation collection moment when the quality fluctuation degree is less than the optimal segmentation threshold is regarded as the stable fluctuation collection moment;

[0035] For each visual inspection unit in each inspection cycle, the absolute value of the difference between the mean of the quality fluctuation degree at all severe fluctuation collection moments and the mean of the quality fluctuation degree at all stable fluctuation collection moments is calculated as the quality difference index of each visual inspection unit in each inspection cycle;

[0036] For each visual inspection unit in each inspection cycle, calculate the ratio of the number of stable fluctuation collection moments to the number of all collection moments in the inspection cycle, and use this as the stable duration ratio of each visual inspection unit in each inspection cycle.

[0037] For each visual inspection unit in each inspection cycle, the product of the mean of the quality fluctuation degree at all severe fluctuation collection moments and the quality difference index is calculated, and the ratio of the product to the proportion of the stable time length is used as the surface quality coefficient of each visual inspection unit in each inspection cycle.

[0038] Preferably, the edge quality coefficient is obtained by:

[0039] For each visual detection unit in each detection cycle, the average of the first surface image and the second surface image at each acquisition moment is calculated as the initial surface feature of each visual detection unit at each acquisition moment in each detection cycle, and the average of the first edge image and the second edge image at each acquisition moment is calculated as the initial edge feature of each visual detection unit at each acquisition moment in each detection cycle;

[0040] The edge quality coefficient is calculated by weighted combination of the initial edge feature change and the position sequence difference between adjacent severe fluctuation collection moments, and each weight is determined according to the number of severe fluctuation collection moments in the detection period.

[0041] Preferably, the method for obtaining the adjustment value of the detection parameter is:

[0042] The adjustment value of the detection parameter is calculated by a weighted combination of a preset initial value, a surface quality coefficient and an edge quality coefficient, and each weight is determined by a normalization function.

[0043] Preferably, the method for determining the preset size of the local window is:

[0044] For each texture feature sequence of each visual detection unit in each detection cycle, the product of the number of all acquisition moments in the detection cycle and the preset proportional coefficient is calculated, and the calculation result is used as the preset size of the local window.

[0045] Compared with the prior art, the present invention has the following beneficial effects:

[0046] A machine vision-based rapid production quality inspection method for building decorative panels uses multi-dimensional visual data collection to provide a more detailed analytical basis for quality inspection during the production process. This method collects surface images, edge images, and multiple texture feature sequences from each visual inspection unit during different inspection cycles, capturing quality information on decorative panels from multiple perspectives and avoiding the potential biased inspections caused by a single data source. This comprehensive data collection method provides a more realistic picture of the quality status of decorative panels during production, enabling subsequent analysis and judgment to be based on a richer set of information.

[0047] In terms of outlier analysis, this method comprehensively considers the deviation between outliers in the texture feature sequence, the gradient of the sequence, and the deviation from the local mean to calculate the feature deviation index. This multi-factor analysis method breaks through the limitations of traditional anomaly judgment based on a single threshold. It can more accurately identify outliers that truly affect quality and reduce misjudgments caused by accidental fluctuations. Furthermore, by analyzing the distribution of the feature deviation index to obtain the average deviation index at the time of quality fluctuation collection, the impact of outliers is further quantified, providing reliable basic data for subsequent quality fluctuation assessments.

[0048] To assess quality fluctuations, this method combines the similarity of local texture feature sequences and the mean deviation index to more comprehensively reflect quality stability at different acquisition times. This assessment approach not only focuses on the impact of individual outliers but also considers the changing trends of overall texture features, making the description of quality fluctuations more consistent with actual quality variations in production. The resulting surface quality coefficient truly reflects the quality status of decorative panels, providing an effective quantitative indicator for subsequent quality assessment.

[0049] In edge quality testing, the edge quality coefficient is derived from surface and edge images, integrating surface and edge quality into a unified analysis framework for a comprehensive assessment of decorative panel quality. This integrated assessment avoids potential oversights that might occur when evaluating either surface or edge quality separately, ensuring the integrity of quality judgments. Furthermore, combining the surface and edge quality coefficients to determine adjustment values ​​for testing parameters allows the system to optimize parameters in real time based on actual quality conditions, improving test adaptability.

[0050] A key feature of this method is its dynamic adjustment mechanism for detection parameters. Based on the quality analysis results of the current cycle, the detection parameters for the next cycle are automatically adjusted, enabling self-optimization of the detection system. This dynamic adjustment rapidly responds to changes in the production process, such as changes in raw material properties and fluctuations in equipment operating conditions, ensuring that detection thresholds remain within a reasonable range and reducing missed detections and false detections caused by fixed parameters. Furthermore, this parameter adjustment method, based on actual data, eliminates the need for manual intervention and reduces reliance on operator experience, making the detection process more automated and intelligent.

[0051] In actual production applications, this method can generate quality assessment results in real time and quickly feed them into the parameter adjustments of the inspection system, forming a closed-loop quality control process. This process not only improves the efficiency of quality inspections but also promptly identifies quality risks during the production process, allowing production departments to take timely adjustments and reduce the number of substandard products. Furthermore, through the continuous analysis of texture feature sequences, a large amount of quality data can be accumulated, providing valuable reference information for production process improvements, which will help to fundamentally improve the production quality of architectural decorative panels. BRIEF DESCRIPTION OF THE DRAWINGS

[0052] Figure 1 This is a working principle diagram of the method for rapid detection of production quality of building decorative panels based on machine vision according to the present invention;

[0053] Figure 2 Flowchart for the calculation of characteristic deviation index;

[0054] Figure 3 Flowchart for clustering index calculation;

[0055] Figure 4 Flowchart for the calculation of quality fluctuation degree. DETAILED DESCRIPTION

[0056] The following will clearly and completely describe the technical solutions in the embodiments of the present invention in conjunction with the accompanying drawings. Obviously, the described embodiments are only part of the embodiments of the present invention, not all of the embodiments. Based on the embodiments of the present invention, all other embodiments obtained by ordinary technicians in this field without making creative efforts are within the scope of protection of the present invention.

[0057] See also Figure 1-Figure 4 The present invention provides a method for rapid detection of production quality of building decorative panels based on machine vision. The method is described in detail below in conjunction with a specific process.

[0058] Step 1: Collect surface images and edge images from each visual inspection unit at each acquisition moment in each inspection cycle, as well as each texture feature sequence from each visual inspection unit during each inspection cycle. Visual inspection units deployed at different locations on the production line acquire images of the building decorative panels according to a preset inspection cycle. Each inspection cycle includes multiple consecutive acquisition moments, and at each acquisition moment, surface images, edge images, and multiple texture feature sequences are synchronously acquired. The texture feature sequences include, but are not limited to, gray-level co-occurrence matrix features, directional gradient histogram features, and other feature data that can reflect changes in the decorative panel's surface texture.

[0059] Step 2: Based on the deviation between the outliers in the texture feature sequence, the gradient of the texture feature sequence, and the deviation between the mean values ​​of each local texture feature sequence, obtain the characteristic deviation index of each outlier in each texture feature sequence of each visual detection unit in each detection cycle. For each texture feature sequence, first identify the outliers in the sequence, and then quantitatively calculate the characteristic deviation index corresponding to each outlier from three dimensions: the degree of mutual deviation between the outliers, the gradient change trend of the sequence as a whole, and the difference in the mean value of the texture features in the local range. This index is used to measure the degree of deviation of the outlier in the texture feature sequence.

[0060] Step 3: Based on the distribution of the characteristic deviation index, obtain the average deviation index of each visual inspection unit at each quality fluctuation collection moment in each inspection cycle. Analyze the distribution of the characteristic deviation index at different collection moments, screen out collection moments with quality fluctuations, and calculate the average characteristic deviation index of all relevant outliers at these moments as the average deviation index, which comprehensively reflects the overall deviation level at that moment.

[0061] Step 4: Based on the similarity and mean deviation index between local texture feature data sequences, the degree of quality fluctuation for each visual inspection unit at each quality fluctuation collection moment in each inspection cycle is obtained. By comparing the similarity of local texture feature sequences and combining it with the mean deviation index, the intensity of fluctuation at each quality fluctuation collection moment is quantified to obtain the degree of quality fluctuation, which reflects the quality stability of the decorative board at that moment.

[0062] Step 5: Obtain the surface quality coefficient for each visual inspection unit during each inspection cycle based on the degree of quality fluctuation. Comprehensively analyze the degree of quality fluctuation at all quality fluctuation collection moments within the inspection cycle and process it using a specific algorithm to obtain the surface quality coefficient, which is used to characterize the overall surface quality of the building decorative panels.

[0063] Step 6: Obtain the edge quality coefficient for each visual inspection unit during each inspection cycle based on the surface and edge images. Perform image processing and feature extraction on the collected surface and edge images, analyzing edge integrity, smoothness, and other characteristics. Calculate the edge quality coefficient to reflect the quality of the decorative panel edge.

[0064] Step 7: Based on the surface quality coefficient and edge quality coefficient, the adjusted value of the detection parameter for each visual inspection unit in each inspection cycle is obtained. The adjusted value of the detection parameter for each visual inspection unit in each inspection cycle is used as the detection parameter value for each visual inspection unit when using template matching for defect identification in the next adjacent inspection cycle, thereby correcting the detection threshold. By fusing the surface quality coefficient and edge quality coefficient, the adjusted value of the detection parameter is calculated and applied to the template matching defect identification process in the next inspection cycle, realizing dynamic correction of the detection threshold to improve detection accuracy and adaptability.

[0065] Example 1:

[0066] Each visual inspection unit uses a collaborative approach of line scan cameras and area scan cameras to capture surface and edge images. Line scan cameras, with their line-by-line scanning capabilities, can capture surface images at high speeds at each capture moment on a production line where building decorative panels are continuously transported. These images are defined as first surface images. To enhance the imaging of edge areas, a ring light source is used to assist with edge image capture using the line scan camera. This ring light source evenly illuminates the edges of decorative panels from multiple angles, minimizing the impact of shadows and reflections on image quality. Edge images captured by the line scan camera at each capture moment with the aid of the ring light source are defined as first edge images.

[0067] An area scan camera can capture images of a larger area at once. At each capture moment, the camera simultaneously captures the surface and edge of the decorative panel, producing a second surface image and a second edge image, respectively. The first and second surface images differ in their imaging principles. The first surface image from a line scan camera captures the continuous texture of the decorative panel along its direction of motion, while the second surface image from an area scan camera reveals the overall regional characteristics of the decorative panel at a given moment. The combination of the two fully captures both surface details and overall condition.

[0068] The first and second edge images are also complementary. The first edge image, acquired by a line scan camera, excels in edge line continuity and is suitable for analyzing edge straightness and continuity. The second edge image, acquired by an area scan camera, more clearly shows the junction between the edge and the surface, facilitating observation of edge integrity and transitions.

[0069] In practice, the mounting positions of the line scan camera and area scan camera require precise calibration to ensure that the images captured by both cameras accurately correspond to the same area of ​​the decorative panel. During calibration, the camera's height, angle, and focal length are adjusted to ensure that the image areas captured by the two cameras at the same acquisition time completely overlap, avoiding image misalignment due to field of view differences. Furthermore, the acquisition frequency of both cameras must match the operating speed of the production line. As the line speed changes, the acquisition frequency is adjusted synchronously to ensure clear, unambiguous image data is captured at every acquisition moment.

[0070] The scanning frequency of the line scan camera is set based on the width of the decorative panel and the transmission speed, ensuring that the entire width can be scanned within a specified timeframe. The scanning frequency of the area scan camera is determined based on the required inspection accuracy, ensuring that data redundancy caused by excessively high frequencies is avoided while still meeting inspection requirements. The brightness of the ring light source also needs to be adjusted based on the reflective properties of the decorative panel. For high-gloss panels, the brightness should be appropriately reduced to minimize reflections; for panels with darker surfaces, the brightness should be increased to improve image contrast.

[0071] Through the collaborative work of line scan cameras and area scan cameras, and the assistance of a ring light source, the captured first surface image, second surface image, first edge image, and second edge image can comprehensively and accurately reflect the surface and edge characteristics of the building decorative panels at each acquisition moment, providing rich and reliable raw data for subsequent quality analysis and inspection parameter adjustment.

[0072] Example 2:

[0073] Obtaining the feature deviation index for each texture feature sequence in each inspection cycle for each visual inspection unit requires a series of steps. First, the texture feature sequence is processed using a grayscale threshold segmentation algorithm. This algorithm uses a grayscale threshold to divide the grayscale values ​​in the sequence that exceed the normal range into highlight and dark areas. These areas are identified as outliers in the texture feature sequence. The grayscale threshold setting must be combined with the normal value range of the texture feature and determined by analyzing historical normal texture data to ensure accurate distinction between normal areas and outliers.

[0074] For each outlier, the absolute difference between its texture feature data and the texture feature data from the immediately preceding acquisition moment is calculated. This value is the forward deviation index. The forward deviation index reflects the magnitude of the change in the outlier's texture features over time compared to the immediately preceding moment. If no data is available from the immediately preceding acquisition moment, the texture feature data from the first acquisition moment within the detection cycle is used as a reference.

[0075] To obtain the clustering index, the gradient amplitude of all elements in the texture feature sequence is first calculated. The gradient amplitude is calculated by subtracting the texture feature data from two adjacent acquisition moments. The maximum of all gradient amplitudes is taken as the maximum gradient of the texture feature sequence. For each outlier, the absolute value of the difference between its texture feature data and the maximum gradient is calculated. This value is the clustering index, which reflects the deviation of the outlier point from the maximum gradient change of the entire sequence.

[0076] When constructing a local window, each outlier is used as the center, the window size is a preset size, and all texture feature data within the window range are included in the analysis. The mean of all texture feature data before the outlier in the local window is calculated, and then the mean of all texture feature data after the outlier is calculated. The absolute value of the difference between the two means is the deviation from the mean before and after. The deviation from the mean before and after reflects the overall difference between the texture features of the two segments before and after within the local range where the outlier is located. If the outlier is at the starting position of the window, only the absolute value of the difference between the mean of the texture feature data after the outlier and the texture feature data of the outlier is calculated; if the outlier is at the end of the window, only the absolute value of the difference between the mean of the texture feature data before the outlier and the texture feature data of the outlier is calculated.

[0077] The characteristic deviation index is derived from a weighted combination of the forward deviation index, clustering index, and front-to-back mean deviation, with each weight being a preset adjustment coefficient. The value of the adjustment coefficient is determined based on the importance of different texture features in quality inspection. For example, for texture features related to surface flatness, the weight of the forward deviation index can be appropriately increased; for features related to texture consistency, the weight of the front-to-back mean deviation can be increased accordingly. In practical applications, the adjustment coefficient can be set by analyzing the inspection data of various decorative panel samples and can be dynamically adjusted based on quality feedback during the production process to ensure that the characteristic deviation index can accurately quantify the degree of deviation of outliers, providing a reliable basis for subsequent quality fluctuation analysis.

[0078] During the entire process, for each texture feature sequence, the feature deviation index of the outlier point needs to be calculated independently. The processing processes of different texture feature sequences are independent of each other, ensuring that the anomalies of each feature can be accurately captured and quantified.

[0079] Example 3:

[0080] For each outlier point in each texture feature sequence for each visual inspection unit during each inspection cycle, a determination is made as to whether the acquisition moment at that point in time represents a quality fluctuation acquisition moment. When all texture feature data in all types of texture feature sequences at a particular acquisition moment represent outliers, that acquisition moment is determined to be a quality fluctuation acquisition moment. After determining the quality fluctuation acquisition moment, the mean of the feature deviation indices for all types of texture feature sequences at that moment in time is calculated, and this is used as the average deviation index for each visual inspection unit at that quality fluctuation acquisition moment in that inspection cycle.

[0081] To calculate the degree of quality fluctuation, we first cluster the average deviation index of all visual inspection units using the K-means clustering algorithm at each quality fluctuation acquisition moment in each inspection cycle. Based on the distribution characteristics of the average deviation index, similar average deviation indices are grouped together, ultimately resulting in several clusters. The cluster containing the average deviation index of each visual inspection unit is identified as the target cluster, reflecting the similarity of the visual inspection unit's quality fluctuation degree with other units.

[0082] Next, for each visual inspection unit's texture feature data at each quality fluctuation acquisition moment in each texture feature sequence during each inspection cycle, a local window of a preset size is constructed with the outlier as the center. All texture feature data within the local window are chronologically sequenced to form the local window sequence at the quality fluctuation acquisition moment. The size of the local window is determined based on the frequency of change in the texture feature sequence and the length of the inspection cycle, ensuring that it covers sufficient historical and subsequent data to reflect the texture feature change trend before and after the outlier.

[0083] For each visual detection unit at each quality fluctuation collection moment in each detection cycle, the similarity between the local window sequence of each texture feature sequence and the local window sequence of the same type of texture feature sequence of all other visual detection units in its target cluster is calculated. The similarity calculation is achieved by comparing the overall change trends and numerical distributions of the two sequences. The closer the values ​​are, the more consistent the change trends are, and the higher the similarity is. The average of all similarity values ​​is taken as the feature trend difference of the texture feature sequence of the visual detection unit in the corresponding detection cycle and quality fluctuation collection moment. The smaller the feature trend difference, the more consistent the texture feature change of the visual detection unit is with other units in the target cluster; otherwise, it indicates that there is a more obvious difference.

[0084] The feature dissimilarity index is calculated by averaging the characteristic trend differences of all texture feature sequences for the visual inspection unit during the inspection cycle and at the time of quality fluctuation collection. The feature dissimilarity index comprehensively reflects the overall degree of difference between the visual inspection unit and other units in the target cluster in terms of multiple texture features.

[0085] Finally, the degree of quality fluctuation is calculated by a weighted combination of the average deviation index, the number of target cluster elements, and the feature dissimilarity index, as follows:

[0086] Z=a×P+b×Q+c×R

[0087] Among them, Z represents the degree of quality fluctuation, P represents the average deviation index, Q represents the number of target cluster elements, R represents the characteristic dissimilarity index, a, b, and c are the weights of the average deviation index, the number of target cluster elements, and the characteristic dissimilarity index, respectively, and all are preset adjustment factors.

[0088] In practice, the value of the adjustment factor is determined based on the inspection scenario and the quality characteristics of the decorative panels. When the mean deviation index significantly influences quality fluctuations, the value of a is increased accordingly. If the number of target cluster elements is large, indicating that the fluctuation pattern is universal, the value of b can be appropriately increased. When the characteristic dissimilarity index better reflects the impact of individual differences on quality, the value of c needs to be adjusted. This weighted combination allows the degree of quality fluctuation to fully integrate information from multiple dimensions, reflecting both the absolute level of quality fluctuation at that moment and its relative position and characteristic differences within the population, providing a quantitative basis for subsequent surface quality coefficient calculations.

[0089] Throughout the entire process, the degree of quality fluctuation is independently calculated for each inspection cycle and each quality fluctuation collection moment, ensuring real-time capture of dynamic quality changes during production. Quality fluctuations between different visual inspection units can be directly compared, facilitating the identification of abnormal areas or equipment on the production line and providing targeted reference information for quality control. Furthermore, the division of target clusters and the calculation of feature similarity are based on actual collected data, eliminating interference from subjective factors and making the calculated quality fluctuation results more objective and reliable.

[0090] Example 4:

[0091] When obtaining the surface quality coefficient, the quality fluctuation degree of each visual inspection unit at all quality fluctuation collection moments during each inspection cycle is used as input data for the Otsu threshold segmentation algorithm. The Otsu threshold segmentation algorithm automatically determines an optimal segmentation threshold by analyzing the grayscale distribution characteristics of the quality fluctuation degree. This threshold can classify the quality fluctuation degree into two different categories. Among them, quality fluctuation collection moments with a quality fluctuation degree greater than or equal to the optimal segmentation threshold are classified as severe fluctuation collection moments; quality fluctuation collection moments with a quality fluctuation degree less than the optimal segmentation threshold are classified as stable fluctuation collection moments.

[0092] For each visual inspection unit in each inspection cycle, it is necessary to calculate the mean of the quality fluctuation degree at the moment of severe fluctuation collection and the moment of smooth fluctuation collection respectively. Specifically, collect the quality fluctuation degree of all severe fluctuation collection moments of the visual inspection unit in the current inspection cycle, calculate the arithmetic mean of these values, and obtain the severe fluctuation mean; similarly, collect the quality fluctuation degree of all smooth fluctuation collection moments, calculate the arithmetic mean, and obtain the smooth fluctuation mean. Then, calculate the absolute value of the difference between the severe fluctuation mean and the smooth fluctuation mean, and use the absolute value as the quality difference index of the visual inspection unit in the current inspection cycle. The size of the quality difference index reflects the gap between severe fluctuation and smooth fluctuation. The larger the difference index, the more obvious the difference in the quality fluctuation degree under the two fluctuation states.

[0093] At the same time, the number of stable and fluctuating acquisition moments within the current inspection cycle and the total number of acquisition moments within the inspection cycle are counted. The ratio of the number of stable and fluctuating acquisition moments to the total number is calculated to obtain the stable duration ratio. The stable duration ratio ranges from 0 to 1. A larger ratio indicates that the decorative panel spent a greater proportion of its time in a stable and fluctuating state during the entire inspection cycle.

[0094] Finally, multiply the severe fluctuation mean by the quality difference index to obtain a product result, and then divide the product result by the proportion of stable time. The quotient obtained is the surface quality coefficient of the visual inspection unit in the current inspection cycle.

[0095] To obtain the edge quality coefficient, the image data from each visual inspection unit at each acquisition moment during each inspection cycle is first processed. For each acquisition moment, the first surface image captured by the line scan camera and the second surface image captured by the area scan camera are averaged to obtain the initial surface features at that acquisition moment. Similarly, the first edge image captured by the line scan camera using a ring light source and the second edge image captured by the area scan camera are averaged to obtain the initial edge features at that acquisition moment. Both the initial surface features and the initial edge features are presented as image pixel values ​​and reflect the basic characteristics of the decorative panel's surface and edge at that moment.

[0096] For each visual inspection unit in each detection cycle, severely fluctuating acquisition moments are selected from all acquisition moments and sorted chronologically. Next, two adjacent severely fluctuating acquisition moments are selected, and the initial edge feature change and position difference between them are calculated. The initial edge feature change refers to the absolute difference between the initial edge feature of the subsequent severely fluctuating acquisition moment and the initial edge feature of the previous severely fluctuating acquisition moment; the position difference refers to the difference between the position sequence number of the subsequent severely fluctuating acquisition moment within the detection cycle and the position sequence number of the previous severely fluctuating acquisition moment.

[0097] The initial edge feature change and the position difference are weightedly combined to obtain the edge quality sub-coefficient corresponding to adjacent severe fluctuation collection moments. For all adjacent severe fluctuation collection moments within the same detection cycle, the edge quality sub-coefficient is calculated in the same manner as above, and then the average of all edge quality sub-coefficients is taken as the edge quality coefficient of the visual inspection unit in the current detection cycle. The weights used in the weighted combination are determined according to the number of severe fluctuation collection moments within the detection cycle. When the number of severe fluctuation collection moments is large, the weight of the position difference will be adjusted accordingly to balance the impact of edge feature changes at different intervals on the overall edge quality.

[0098] Example 5:

[0099] Adjustment values ​​for inspection parameters are determined by combining preset initial values, surface quality coefficients, and edge quality coefficients. These values ​​are determined based on the type and material of the building decorative panel, as well as the underlying inspection standards. Different types of decorative panels correspond to different preset initial values. For example, the preset initial values ​​for wood and stone decorative panels are differentiated based on their surface characteristics and common defect types. The surface quality coefficient and edge quality coefficient are derived from the quality analysis results of the current inspection cycle and reflect the surface and edge quality of the decorative panel, respectively.

[0100] When calculating the adjustment value of the detection parameter, the preset initial value, surface quality coefficient and edge quality coefficient need to be weighted and combined. Each weight is determined by a normalization function, which processes the surface quality coefficient and edge quality coefficient according to their numerical ranges and converts them into corresponding weight ratios. Specifically, the normalization function first standardizes the surface quality coefficient and edge quality coefficient to eliminate the impact of different dimensions, and then assigns weights based on their relative importance in quality inspection. The weight of the preset initial value is 1 minus the sum of the weights of the surface quality coefficient and the edge quality coefficient to ensure that the total weight of the three is 1. Through such a weighted combination, the obtained detection parameter adjustment value can comprehensively reflect the benchmark requirements and the current quality status, and use it as the value of the detection parameter for each visual inspection unit when using template matching for defect identification in the next adjacent cycle, so as to realize the correction of the detection threshold so that the detection threshold can adapt to quality fluctuations in the production process.

[0101] The preset size of the local window is determined for each texture feature sequence in each inspection cycle for each visual inspection unit. This determination begins by counting all acquisition moments within the inspection cycle. The number of acquisition moments is set based on the production line's operating speed and inspection accuracy requirements. For faster speeds or higher inspection accuracy requirements, the number of acquisition moments is increased accordingly. The number of acquisition moments is then multiplied by a preset scaling factor. The resulting product becomes the preset size of the local window for that texture feature sequence.

[0102] The value of the preset scaling factor must take into account the changing characteristics of the texture feature sequence. For texture features that change frequently, such as surface density variations, the preset scaling factor can be increased appropriately to allow the local window to cover more acquisition moments and capture a more comprehensive range of texture change trends. For texture features that change more slowly, such as edge smoothness, the preset scaling factor can be decreased appropriately to improve the specificity of the local analysis. The specific value of the preset scaling factor can be determined based on historical inspection data and the production process of the decorative panels. The preset scaling factor can be different for different inspection cycles and different texture feature sequences to accommodate diverse inspection needs.

[0103] In actual applications, the preset size of the local window will be dynamically adjusted according to the detection cycle and texture feature sequence. For example, in a certain detection cycle, if the number of acquisition moments of a certain texture feature sequence is 100 and the preset proportional coefficient is 0.2, then the preset size of the local window of the texture feature sequence is 20, that is, each local window contains texture feature data from 20 acquisition moments. The local window size determined in this way can match the changing pattern of texture features, provide a suitable analysis range for steps such as the calculation of the deviation of the front and back mean of the outlier point, ensure that the texture feature data in the local window can accurately reflect the local environment where the outlier point is located, thereby improving the reliability of calculation results such as the feature deviation index.

[0104] The adjustment of the detection parameters and the determination of the preset size of the local window are independent of each other, but are both based on the actual data of the current detection cycle. This ensures that the entire detection method can be dynamically adjusted according to the actual production situation and adapt to different production conditions and quality conditions, providing flexible and reliable technical support for the rapid detection of the production quality of building decorative panels.

[0105] It should be noted that, in this document, relational terms such as first and second, etc., are used only to distinguish one entity or operation from another entity or operation, and do not necessarily require or imply any actual relationship or order between these entities or operations. Moreover, the terms "comprises," "comprising," or any other variations thereof are intended to cover non-exclusive inclusion, such that a process, method, article, or apparatus that includes a list of elements includes not only those elements but also other elements not explicitly listed, or elements inherent to such process, method, article, or apparatus.

[0106] While embodiments of the present invention have been shown and described, it will be appreciated by those skilled in the art that various changes, modifications, substitutions, and variations may be made to these embodiments without departing from the principles and spirit of the invention, and that the scope of the invention is defined by the appended claims and their equivalents.

Claims

1. A rapid detection method for production quality of building decorative panels based on machine vision, characterized in that: The method comprises the following steps: Collecting the surface image and edge image of each visual detection unit at each collection moment in each detection cycle and each texture feature sequence of each visual detection unit in each detection cycle; Based on the deviation between the abnormal points in the texture feature sequence, the gradient of the texture feature sequence and the deviation between the means of each local texture feature sequence, the characteristic deviation index of each abnormal point in each texture feature sequence of each visual detection unit in each detection cycle is obtained; Based on the distribution of the characteristic deviation index, the average deviation index of each visual inspection unit at each quality fluctuation collection moment in each inspection cycle is obtained; Based on the similarity and average deviation index between local texture feature data sequences, the quality fluctuation degree of each visual inspection unit at each quality fluctuation collection moment in each inspection cycle is obtained; Obtaining the surface quality coefficient of each visual inspection unit in each inspection cycle based on the degree of quality fluctuation; Obtaining edge quality coefficients of each visual inspection unit in each inspection cycle based on the surface image and the edge image; Based on the surface quality coefficient and the edge quality coefficient, the adjustment value of the detection parameter of each visual inspection unit in each inspection cycle is obtained. The adjustment value of the detection parameter of each visual inspection unit in each inspection cycle is used as the value of the detection parameter of each visual inspection unit when using template matching for defect identification in the next adjacent cycle of each inspection cycle, and the detection threshold is corrected.

2. The method for rapid detection of production quality of building decorative panels based on machine vision according to claim 1, characterized in that: The surface image and edge image include: For each visual inspection unit, a surface image at each acquisition moment is acquired by a line array camera as the first surface image at each acquisition moment, and an edge image at each acquisition moment is acquired with the assistance of a ring light source as the first edge image at each acquisition moment; a surface image and an edge image at each acquisition moment are acquired by an area array camera as the second surface image and the second edge image at each acquisition moment.

3. The method for rapid detection of production quality of building decorative panels based on machine vision according to claim 1, characterized in that: The method for obtaining the characteristic deviation index is: For each texture feature sequence of each visual detection unit in each detection cycle, a grayscale threshold segmentation algorithm is used to obtain all the highlight areas and dark areas in each texture feature sequence of each detection cycle as each outlier. For each outlier, the absolute value of the difference between the texture feature data of the outlier and the texture feature data of the previous adjacent acquisition moment is calculated as the forward deviation index of each outlier in each texture feature sequence of each visual detection unit in each detection cycle. Obtain the clustering index of each abnormal point based on the gradient of the texture feature data sequence; For each outlier point in each texture feature sequence of each visual detection unit in each detection cycle, a window of preset size is constructed with the outlier point as the center as the local window of each outlier point. The mean of all texture feature data before the outlier point in the local window is calculated, and the absolute value of the difference between the mean of all texture feature data after the outlier point in the local window is used as the mean deviation of each outlier point. The characteristic deviation index is calculated as a weighted combination of the forward deviation index, the clustering index and the forward and backward mean deviation, and each weight is a preset adjustment coefficient.

4. The method for rapid detection of production quality of building decorative panels based on machine vision according to claim 3, characterized in that: The clustering index is obtained as follows: For each texture feature sequence of each visual detection unit in each detection cycle, the maximum value of the gradient amplitude of all elements in the texture feature sequence is calculated as the maximum gradient value of the texture feature sequence; For each abnormal point in each texture feature sequence of each visual detection unit in each detection cycle, the absolute value of the difference between the texture feature data of the abnormal point and the gradient maximum value is calculated as the clustering index of each abnormal point.

5. The method for rapid detection of production quality of building decorative panels based on machine vision according to claim 1, characterized in that: The method for obtaining the mean deviation index is: For each abnormal point in each texture feature sequence of each visual detection unit in each detection cycle, when the texture feature data in all other types of texture feature sequences at the acquisition moment where the abnormal point is located are all abnormal points, the acquisition moment is used as the quality fluctuation acquisition moment, and the average of the characteristic deviation indexes of the abnormal points of all types of texture feature sequences at the quality fluctuation acquisition moment of each detection cycle is calculated as the average deviation index of each visual detection unit at each quality fluctuation acquisition moment of each detection cycle.

6. The method for rapid detection of production quality of building decorative panels based on machine vision according to claim 3, characterized in that: The method for obtaining the quality fluctuation degree is: At each quality fluctuation collection moment in each detection cycle, the K-means clustering algorithm is used to cluster the average deviation index of all visual inspection units to obtain each cluster; the cluster where the average deviation index of each visual inspection unit is located is used as the target cluster of each visual inspection unit; For the texture feature data of each visual inspection unit in each texture feature sequence in each inspection cycle at each quality fluctuation collection moment, a sequence consisting of all texture feature data in a local window of the texture feature data is used as a local window sequence at each quality fluctuation collection moment; For each visual detection unit at each quality fluctuation collection moment in each detection cycle, calculate the mean of the similarities between the local window sequence of each texture feature sequence and the local window sequences of the same type of texture feature sequences of all other visual detection units in its target cluster, as the feature trend difference of each texture feature sequence of each visual detection unit at each quality fluctuation collection moment in each detection cycle, and take the mean of the feature trend differences of all types of texture feature sequences of each visual detection unit at each quality fluctuation collection moment in each detection cycle as the feature dissimilarity index of each visual detection unit at each quality fluctuation collection moment in each detection cycle; The degree of quality fluctuation is calculated as a weighted combination of the average deviation index, the number of target cluster elements, and the characteristic dissimilarity index, with each weight being a preset adjustment factor.

7. The method for rapid detection of production quality of building decorative panels based on machine vision according to claim 2, characterized in that: The method for obtaining the surface quality coefficient is: The quality fluctuation degree of each visual inspection unit at all quality fluctuation collection moments in each inspection cycle is used as the input of the Otsu threshold segmentation algorithm, and the optimal segmentation threshold is output. The quality fluctuation collection moment when the quality fluctuation degree is greater than or equal to the optimal segmentation threshold is regarded as the severe fluctuation collection moment, and the quality fluctuation collection moment when the quality fluctuation degree is less than the optimal segmentation threshold is regarded as the stable fluctuation collection moment; For each visual inspection unit in each inspection cycle, the absolute value of the difference between the mean of the quality fluctuation degree at all severe fluctuation collection moments and the mean of the quality fluctuation degree at all stable fluctuation collection moments is calculated as the quality difference index of each visual inspection unit in each inspection cycle; For each visual inspection unit in each inspection cycle, calculate the ratio of the number of stable fluctuation collection moments to the number of all collection moments in the inspection cycle, and use this as the stable duration ratio of each visual inspection unit in each inspection cycle. For each visual inspection unit in each inspection cycle, the product of the mean of the quality fluctuation degree at all severe fluctuation collection moments and the quality difference index is calculated, and the ratio of the product to the proportion of the stable time length is used as the surface quality coefficient of each visual inspection unit in each inspection cycle.

8. The method for rapid detection of production quality of building decorative panels based on machine vision according to claim 7, characterized in that: The method for obtaining the edge quality coefficient is: For each visual detection unit in each detection cycle, the average of the first surface image and the second surface image at each acquisition moment is calculated as the initial surface feature of each visual detection unit at each acquisition moment in each detection cycle, and the average of the first edge image and the second edge image at each acquisition moment is calculated as the initial edge feature of each visual detection unit at each acquisition moment in each detection cycle; The edge quality coefficient is calculated by weighted combination of the initial edge feature change and the position sequence difference between adjacent severe fluctuation collection moments, and each weight is determined according to the number of severe fluctuation collection moments in the detection period.

9. The method for rapid detection of production quality of building decorative panels based on machine vision according to claim 1, characterized in that: The method for obtaining the adjustment value of the detection parameter is: The adjustment value of the detection parameter is calculated by a weighted combination of a preset initial value, a surface quality coefficient and an edge quality coefficient, and each weight is determined by a normalization function.

10. The method for rapid production quality inspection of building decorative panels based on machine vision according to claim 3, characterized in that: The method for determining the preset size of the local window is: For each texture feature sequence of each visual detection unit in each detection cycle, the product of the number of all acquisition moments in the detection cycle and the preset proportional coefficient is calculated, and the calculation result is used as the preset size of the local window.

Images