Lacquerware painting control method and system based on machine vision
By acquiring information about the lacquer body using machine vision technology, identifying polishing defects, and generating dynamic control strategies, the problem of unstable polishing in the traditional lacquer process has been solved, achieving efficient and refined lacquer surface treatment and improving production efficiency and quality.
Patent Information
- Authority / Receiving Office
- CN · China
- Patent Type
- Patents(China)
- Current Assignee / Owner
- SICHUAN AGRI UNIV
- Filing Date
- 2026-03-11
- Publication Date
- 2026-05-19
AI Technical Summary
In the existing technology, the degree of polishing in the lacquer application process relies on visual observation and experience to judge, which leads to unstable polishing results and affects the consistency and smoothness of the lacquer surface quality.
A machine vision-based control method is adopted. By acquiring information about the body manufacturing process, the number of key lacquer layers and the optimal polishing scheme are determined. Combined with image acquisition based on illumination parameters, polishing defects are identified, and a dynamic polishing control strategy is generated to achieve closed-loop control.
It has enabled refined management of lacquerware surfaces, reduced sandpaper usage by 40%, shortened repair cycles by 30%, improved production efficiency, and significantly enhanced the smoothness and quality consistency of lacquerware surfaces.
Smart Images
Figure CN121810694B_ABST
Abstract
Description
Technical Field
[0001] This invention relates to the field of lacquer control, and particularly to a machine vision-based method and system for controlling lacquer application. Background Technology
[0002] Lacquerware application is a unique and historically significant traditional Chinese craft, referring to the process of coating the surface of objects with natural raw lacquer to form a protective and decorative layer. The lacquerware process often requires multiple layers of lacquer. Because raw lacquer is a natural raw material and is harvested manually, it contains a certain amount of impurities. The laccase in raw lacquer has a drying effect; the stronger the laccase activity, the faster the lacquer dries. Raw lacquer cannot be filtered by machine; if centrifuges or other machines are used, the laccase in the raw lacquer will be destroyed, preventing it from drying. Therefore, raw lacquer undergoes simple manual filtration and refining to become black polished lacquer. However, when used for lacquering, black polished lacquer often still contains a small amount of impurities with relatively large particle sizes, affecting the smoothness of the lacquered surface and requiring polishing to refine it. Furthermore, when applying lacquer directly with a brush, brush marks are inevitably left on the substrate surface after lacquering, and these brush marks must also be removed by polishing.
[0003] In existing techniques, craftsmen rely on visual observation and tactile experience accumulated over long-term practice to judge the degree of polishing during the lacquer application process. This evaluation method is highly subjective, and different craftsmen have different judgment standards, making it difficult to guarantee the stability and consistency of the polishing effect. Even for the same craftsman, the judgment of the polishing effect may deviate at different times, environments, or emotional states, resulting in inconsistent surface quality of lacquerware and affecting the overall quality of the lacquer application.
[0004] Therefore, there is a need to provide machine vision-based methods and systems for controlling lacquer application to improve lacquer quality. Summary of the Invention
[0005] This invention provides a machine vision-based lacquerware lacquer application control method, comprising: acquiring the manufacturing information of the lacquerware body; determining multiple key lacquer application layers and an optimal polishing scheme based on the manufacturing information of the lacquerware body and the total number of lacquer application layers, wherein the optimal polishing scheme includes the polishing grit number corresponding to each key lacquer application layer; determining optimal lighting parameters based on the manufacturing information of the lacquerware body and the multiple key lacquer application layers; for each key lacquer application layer, acquiring an image of the lacquer layer corresponding to the key lacquer application layer based on the optimal lighting parameters; identifying polishing defect information corresponding to the key lacquer application layer based on the image of the lacquer layer corresponding to the key lacquer application layer; and generating a polishing control strategy for the next lacquer application layer based on the polishing defect information corresponding to the key lacquer application layer.
[0006] Furthermore, the manufacturing information of the base material includes the type of substrate and the composition information of the ash in the ash-scraping process; based on the manufacturing information of the lacquerware base material and the total number of lacquer coats, several key lacquer coat numbers and optimal polishing schemes are determined, including: determining the target sample lacquerware type from multiple sample lacquerware types based on the manufacturing information of the base material and the total number of lacquer coats; determining several key lacquer coat numbers based on the lacquerware test data of the target sample lacquerware type; and determining the optimal polishing scheme based on the lacquerware test data of the target sample lacquerware type and the several key lacquer coat numbers.
[0007] Furthermore, the lacquer test data for the target sample lacquerware type includes images of the lacquer layer on the body of multiple sample lacquerware pieces corresponding to the target sample lacquerware type under preset polishing schemes and preset lighting parameters, for each number of lacquer layers. Based on the lacquer test data for the target sample lacquerware type, several key lacquer layer numbers are determined, including: for each lacquer layer number, based on the images of the lacquer layer on the body of multiple sample lacquerware pieces under preset polishing schemes and preset lighting parameters, determining key values for the lacquer layer number; and based on the key values for each lacquer layer number, determining several key lacquer layer numbers.
[0008] Furthermore, based on images of the lacquer layer of multiple sample lacquerware pieces under preset polishing schemes and preset lighting parameters, the key values for the number of lacquer layers are determined, including: for each sample lacquerware piece, based on images of the lacquer layer of the sample lacquerware piece under preset polishing schemes and preset lighting parameters, and images of the lacquer layer of adjacent lacquer layer pieces, determining the performance difference value corresponding to the number of lacquer layer pieces; based on the performance difference value corresponding to the number of lacquer layer pieces for each sample lacquerware piece, calculating the first difference value corresponding to the number of lacquer layer pieces; based on images of the lacquer layer of the sample lacquerware piece under preset polishing schemes and preset lighting parameters, calculating the second difference value corresponding to the number of lacquer layer pieces; and based on the first and second difference values, calculating the key value for the number of lacquer layer pieces.
[0009] Furthermore, the lacquer test data for the target sample lacquerware type includes images of the lacquer layer of multiple sample lacquerware pieces corresponding to the target sample lacquerware type under different polishing schemes and preset lighting parameters, for each key number of lacquer layers. Based on the lacquer test data for the corresponding sample lacquerware type and the multiple key lacquer layers, the optimal polishing scheme is determined, including: for each polishing scheme, calculating the priority value of the polishing scheme based on the images of the lacquer layer of multiple sample lacquerware pieces corresponding to the target sample lacquerware type under the polishing scheme and preset lighting parameters, for each key number of lacquer layers; and determining the optimal polishing scheme based on the priority value of each polishing scheme.
[0010] Furthermore, the lacquer test data for the target sample lacquerware type includes images of the lacquer layer of the body for each key lacquer layer number under optimal polishing schemes and different lighting parameters for multiple sample lacquerware corresponding to the target sample lacquerware type; based on the body manufacturing information and multiple key lacquer layer numbers, the optimal lighting parameters are determined, including: for each set of lighting parameters, based on the images of the body lacquer layer of the body for each key lacquer layer number under optimal polishing schemes and lighting parameters for multiple sample lacquerware corresponding to the target sample lacquerware type, calculating the effective value of the lighting parameters; and based on the effective value of each set of lighting parameters, determining the optimal lighting parameters.
[0011] Furthermore, based on the paint layer image corresponding to the key number of lacquer passes, the polishing defect information corresponding to the key number of lacquer passes is identified, including: dividing the paint layer image corresponding to the key number of lacquer passes into multiple region images; for each region image, determining the adaptive segmentation threshold corresponding to the region image; generating a binary image corresponding to the region image based on the adaptive segmentation threshold corresponding to the region image; and identifying the surface smoothness corresponding to the region image. The polishing defect information corresponding to the key number of lacquer passes includes the surface smoothness corresponding to each region image.
[0012] Further, determining the adaptive segmentation threshold corresponding to the region image includes: determining the optimal segmentation threshold corresponding to multiple sample region images; determining multiple threshold influencing factors based on the optimal segmentation threshold corresponding to multiple sample region images; and determining the adaptive segmentation threshold corresponding to the region image based on the multiple threshold influencing factors and the optimal segmentation threshold corresponding to multiple sample region images.
[0013] Furthermore, based on the polishing defect information corresponding to the key paint layer number, a polishing control strategy for the next paint layer number is generated, including: determining the number of polishing times for the polishing area corresponding to each area image based on the surface flatness corresponding to each area image, wherein the polishing control strategy for the next paint layer number includes the number of polishing times for the polishing area corresponding to each area image.
[0014] This invention provides a machine vision-based lacquerware application control system, comprising: an information acquisition module for acquiring the manufacturing information of the lacquerware body; a scheme optimization module for determining multiple key lacquer application layers and an optimal polishing scheme based on the manufacturing information of the lacquerware body and the total number of lacquer application layers, wherein the optimal polishing scheme includes the polishing grit number corresponding to each key lacquer application layer; a parameter optimization module for determining optimal lighting parameters based on the manufacturing information of the lacquerware body and the multiple key lacquer application layers; and a lacquerware control module for, for each key lacquer application layer, acquiring an image of the lacquer layer corresponding to the key lacquer application layer based on the optimal lighting parameters, identifying polishing defect information corresponding to the key lacquer application layer based on the image of the lacquer layer corresponding to the key lacquer application layer, and generating a polishing control strategy for the next lacquer application layer based on the polishing defect information corresponding to the key lacquer application layer.
[0015] Compared with existing technologies, the machine vision-based lacquer application control method and system provided by this invention have at least the following beneficial effects:
[0016] 1. A closed-loop control system for substrate manufacturing information, paint application layers, and sanding processes was constructed using machine vision technology. Based on key parameters such as substrate type and ash layer composition, and combined with a sample database, the optimal sanding scheme (e.g., sandpaper grit combination) is matched. Process parameters can be dynamically adjusted for different material characteristics, avoiding over-sanding or under-sanding problems caused by traditional experience-based sanding. Simultaneously, by acquiring paint layer images in real time and identifying surface smoothness, the defect status after each paint application can be quantitatively evaluated, providing data support for subsequent sanding.
[0017] 2. The lacquer layer image is divided into multiple regions for independent analysis. Based on the surface smoothness assessment results of each region, a differentiated polishing strategy is generated. For example, areas with severe defects (such as areas with dense scratches) are polished twice more, while smooth areas are polished with less intervention to avoid over-polishing and damaging the lacquer layer. In the production of bodiless lacquerware, selective processing can be applied to the extremely thin lacquer layer (only 0.2mm thick), repairing defects while preserving the original lightness of the object. This refined management model reduces sandpaper usage by approximately 40% and shortens the repair cycle by 30%, significantly improving production efficiency.
[0018] 3. To address the challenges of complex surface textures and significant variations in reflective properties in lacquerware, this method innovatively employs adaptive segmentation thresholding technology. By analyzing the optimal segmentation threshold and its influencing factors in sample region images, a dynamic thresholding model is generated, enabling precise identification of micron-level defects. Attached Figure Description
[0019] This specification will be further described by way of exemplary embodiments, which will be described in detail with reference to the accompanying drawings. These embodiments are not limiting; in these embodiments, the same reference numerals denote the same structures, wherein:
[0020] Figure 1 This is a schematic flowchart of a machine vision-based lacquer application control method according to some embodiments of this specification;
[0021] Figure 2 This is a flowchart illustrating the key values for determining the number of lacquer coats according to some embodiments of this specification;
[0022] Figure 3 This is a schematic diagram of a machine vision-based lacquerware varnishing control system according to some embodiments of this specification. Detailed Implementation
[0023] To more clearly illustrate the technical solutions of the embodiments in this specification, the accompanying drawings used in the description of the embodiments will be briefly introduced below. Obviously, the drawings described below are merely some examples or embodiments of this specification. For those skilled in the art, these drawings can be applied to other similar scenarios without creative effort. Unless obvious from the context or otherwise specified, the same reference numerals in the drawings represent the same structures or operations.
[0024] Figure 1 This is a flowchart illustrating a machine vision-based lacquer application control method according to some embodiments of this specification, such as... Figure 1 As shown, the machine vision-based lacquer application control method may include the following steps.
[0025] Step 110: Obtain information on the fabrication of the embryo.
[0026] The manufacturing information of the substrate includes the type of substrate and the composition of the putty used in the puttying process.
[0027] Specifically, the type of substrate refers to the original material properties of the base material. Common substrates include wood, metal, ceramics, leather, bamboo weaving, and linoleum (a lightweight base with a linen frame). The physical properties (such as density, water absorption, and coefficient of thermal expansion) and chemical stability (such as acid and alkali resistance and corrosion resistance) of different substrates vary significantly, directly determining the suitability of the lacquering process. For example, wood bases, due to their high water absorption, require multiple coats of lacquer to form a dense lacquer layer to isolate moisture and prevent cracking; metal bases require a rust-preventive primer before a topcoat to ensure adhesion; and linoleum bases, due to their lightweight nature, require a thinner lacquer layer to avoid deformation.
[0028] The composition information of the ash used in the ash application process refers to the ash formula used for surface treatment of the substrate. Ash application is an important step before lacquer application. By applying ash (such as a mixture of raw lacquer and tile ash, brick powder, porcelain powder, etc.) to the surface of the substrate, it fills the pores, cracks, or uneven areas, forming a smooth and solid base layer, providing a uniform adhesion surface for subsequent lacquer application. The composition of the ash directly affects its hardness, fineness, and adhesion to the lacquer layer. For example, traditional lacquer craftsmanship often uses three ash application processes: coarse ash, medium ash, and fine ash. Coarse ash, mainly tile ash, has coarser particles and is used to quickly fill large pores; medium ash, mainly brick powder, has moderate particles and further refines the surface; and fine ash, mainly porcelain powder or talcum powder, has fine particles and ensures a smooth, mirror-like surface.
[0029] Step 120: Based on the production information of the lacquerware body and the total number of lacquer coats, determine several key lacquer coat numbers and the optimal polishing scheme.
[0030] The optimal sanding plan includes the sanding grit corresponding to each key paint coat. Specifically, the key paint coats are those crucial for paint layer thickness accumulation, material transition, or defect correction during the paint application process, such as the 3rd, 5th, and 8th coats. The optimal sanding plan includes specifying the sanding grit (sandpaper particle size) for each key coat. For example, the 3rd coat might use 240-400 grit sandpaper to quickly remove dust bumps and initially smooth the surface; the 5th coat upgrades to 600-800 grit to eliminate orange peel texture and enhance interlayer adhesion; and the 8th coat uses 1000-1500 grit ultrafine sandpaper to achieve a mirror-like finish.
[0031] In some embodiments, step 120 specifically includes:
[0032] Based on the manufacturing information of the body and the total number of lacquer coats, the target sample lacquerware type is determined from multiple sample lacquerware types;
[0033] Based on the lacquer test data of the target sample lacquerware type, several key lacquer layers were determined;
[0034] Based on the lacquer test data of the target sample lacquerware type and several key lacquer application layers, the optimal polishing scheme was determined.
[0035] Specifically, firstly, a sample database is constructed to record the production information (such as substrate type, composition and ratio of the ash layer) and the total number of lacquer coats for each sample lacquer type. Secondly, the weighted Euclidean distance method is used to calculate the similarity between the production information and the total number of lacquer coats of the body and the production information and the total number of lacquer coats of each sample lacquer type: the difference squared is calculated after scaling continuous parameters (such as ash ratio), and the Hamming distance is calculated after one-hot encoding of discrete parameters (such as substrate type). Finally, the comprehensive similarity score is obtained by weighting and summing the results using preset weights (such as the total number of lacquer coats accounting for 45%, the substrate type accounting for 35%, and the composition information accounting for 20%). Finally, the sample lacquer type with the highest comprehensive similarity score is selected from the sample database as the target sample lacquer type.
[0036] In some embodiments, the lacquer test data for the target sample lacquerware type includes images of the lacquer layer of each lacquer layer for multiple sample lacquerware corresponding to the target sample lacquerware type under a preset polishing scheme and preset lighting parameters.
[0037] Specifically, for the selected target sample type, 3-5 representative lacquerware samples of the same type need to be selected and the lacquering process tracked under unified experimental conditions: a preset progressive polishing scheme (such as polishing in stages from 240 grit to 2000 grit sandpaper) is adopted, combined with standard lighting parameters (such as D65 light source, 500lx illuminance, 45° incident angle), and a high-resolution industrial camera is used to image the lacquer layer of the body after each lacquering.
[0038] In some embodiments, based on lacquer test data of the target sample lacquerware type, several key lacquer coat numbers are determined, including:
[0039] For each lacquer layer, the key value of the lacquer layer is determined based on the images of the lacquer layer on the body of multiple sample lacquerware under preset polishing schemes and preset lighting parameters.
[0040] Based on the key value of each lacquer layer number, multiple key lacquer layer numbers are determined.
[0041] Figure 2 This is a flowchart illustrating the key values for determining the number of lacquer coats according to some embodiments of this specification, such as... Figure 2 As shown, in some embodiments, based on images of the lacquer layer on the body of multiple sample lacquerware pieces under preset polishing schemes and preset lighting parameters, key values for the number of lacquer layers are determined, including:
[0042] For each sample lacquerware, based on the images of the body lacquer layer with the number of lacquer layers and the body lacquer layer images of adjacent lacquer layers, the performance difference value corresponding to the number of lacquer layers is determined.
[0043] Based on the performance difference value of each sample lacquerware corresponding to the number of lacquer layers, calculate the first difference value corresponding to the number of lacquer layers;
[0044] Based on the lacquer layer image of the body of each sample lacquerware under the preset polishing scheme and preset lighting parameters, the second difference value corresponding to the number of lacquer layers is calculated.
[0045] Based on the first and second difference values, the key value for the number of lacquer layers is calculated.
[0046] Specifically, for each image of the lacquer layer, key visual features such as surface smoothness, gloss, and orange peel texture density are identified. The number of adjacent lacquer passes can be the number of lacquer passes adjacent to the current lacquer pass number. For example, if the current lacquer pass number is 5, then the adjacent lacquer passes could be the 3rd, 4th, 6th, and 7th lacquer passes. The Euclidean distance between the key visual features of each adjacent lacquer pass number and the key visual features of the current lacquer pass number is calculated. The average of these Euclidean distances is then used as the performance difference value for the sample lacquerware corresponding to the current lacquer pass number.
[0047] The mean of the performance differences for each sample of lacquerware corresponding to the number of lacquer layers is taken as the first difference value corresponding to the number of lacquer layers.
[0048] For any two sample lacquerware pieces, calculate the Euclidean distance between the key visual features of the two sample lacquerware pieces corresponding to the current number of lacquer lines. Calculate the mean of the Euclidean distances between the key visual features of the two sample lacquerware pieces corresponding to the current number of lacquer lines, and use this as the second difference value corresponding to the current number of lacquer lines.
[0049] The average of the first and second difference values corresponding to the number of lacquer coats is used as the key value for the number of lacquer coats.
[0050] Multiple lacquer application lines can be sorted based on the key value of each lacquer application line, and the top n lacquer application lines (e.g., 3, 5, etc.) can be selected as the key lacquer application lines.
[0051] Based on the calculation of Euclidean distance using multi-dimensional visual features, the differences in process performance between adjacent paint layers can be quantified, accurately identifying nodes with significant changes in visual features during the paint layer accumulation process, providing a scientific basis for optimizing the polishing rhythm. Secondly, by calculating the second difference value through cross-sample comparison, the differentiated responses of different samples to the paint application process can be effectively captured, ensuring that the selection of multiple key paint application layers has both process stage-specificity and sample universality.
[0052] In some embodiments, the lacquer test data for the target sample lacquerware type includes images of the lacquer layer of each key lacquer layer on multiple sample lacquerware corresponding to the target sample lacquerware type under different polishing schemes and preset lighting parameters.
[0053] Specifically, in any two polishing schemes, there is at least one key difference in the grit of the polishing for the number of paint coats.
[0054] In some embodiments, based on lacquer test data for the corresponding sample lacquerware type and several key lacquer coat numbers, the optimal polishing scheme is determined, including:
[0055] For each polishing scheme, the priority value of the polishing scheme is calculated based on the images of the lacquer layer of the body of multiple sample lacquerware corresponding to the target sample lacquerware type under the polishing scheme and preset lighting parameters, for each key number of lacquer layers.
[0056] The optimal polishing scheme is determined based on the priority value of each polishing scheme.
[0057] Specifically, for two adjacent key paint coats, the Euclidean distance of the key visual features of the two key paint coats is calculated. This distance comprehensively reflects the process stability during the paint layer accumulation process. The smaller the distance, the smoother the quality transition between adjacent coats and the stronger the interlayer bonding force. This can effectively avoid the decrease in adhesion caused by over-sanding or the surface defects caused by under-sanding.
[0058] Simultaneously, for each sample lacquerware, a polishing effect score was calculated individually for each key lacquer layer number: Thickness uniformity was assessed by statistically analyzing the standard deviation of lacquer layer thickness (a smaller standard deviation resulted in a higher score); the uniformity of polishing effect was assessed by statistically analyzing the standard deviation of lacquer layer surface unevenness (a smaller standard deviation also resulted in a higher score). The two scores were then weighted and summed to obtain the single-pass polishing effect score for each key lacquer layer number corresponding to the polishing scheme for that sample lacquerware. The average value of the sanding effect score of the sanding scheme for the sample lacquerware is obtained. The absolute value of the difference between the surface smoothness of the lacquer layer after the last sanding and the preset surface smoothness is also calculated. After normalizing the average value and the absolute value of the difference of the sanding effect score of the sanding scheme for the sample lacquerware, the comprehensive sanding effect score of the sanding scheme for the sample lacquerware is calculated. The larger the average value of the normalized sanding effect score of the sanding scheme for the sample lacquerware, and the smaller the absolute value of the difference after normalization, the larger the comprehensive sanding effect score of the sanding scheme for the sample lacquerware.
[0059] To eliminate individual differences among samples, the average score of the overall polishing effect for all lacquerware samples of the same sample type is calculated to obtain the average effect score of the current polishing scheme. Ultimately, the priority value of the polishing scheme is composed of two weighted components: the first is the sum of the reciprocals of the Euclidean distances between adjacent key passes (emphasizing process transition stability, weighted at 60%), and the second is the average effect score of the current polishing scheme. This calculation method ensures the stability of the polishing scheme at key process nodes while also considering the polishing effect of each pass, making the priority value a comprehensive indicator reflecting the overall process quality.
[0060] The polishing scheme with the highest priority value is selected as the optimal polishing scheme.
[0061] The above process transforms subjective process evaluation into quantifiable multi-dimensional indicators: On the one hand, by calculating the Euclidean distance of the visual characteristics of adjacent key paint coats, the process stability during paint layer accumulation is accurately quantified, ensuring smooth quality transitions and strong interlayer bonding between coats, effectively avoiding adhesion degradation or surface defects caused by fluctuations in polishing parameters; on the other hand, for each key paint coat, a single-coat scoring model that balances local effects and overall consistency is constructed by comprehensively considering paint layer thickness uniformity, surface unevenness uniformity, and final smoothness differences. Normalization is then used to eliminate individual sample differences, making the evaluation results more universal. Finally, by weighting and integrating the process stability index (the sum of the reciprocals of the Euclidean distances between adjacent coats, weighted at 60%) with the comprehensive polishing effect score (weighted at 40%), the priority value calculation strengthens the control over the stability of key process nodes and ensures the effectiveness of polishing effects for each coat, transforming the selection of the optimal polishing scheme from a trial-and-error model relying on craftsman experience to intelligent decision-making based on objective data.
[0062] Step 130: Determine the optimal lighting parameters based on the fabrication information of the body and the number of key lacquer layers.
[0063] In some embodiments, the lacquer test data for the target lacquerware type includes images of the lacquer layer of each key lacquer layer under optimal polishing schemes and different lighting parameters for multiple lacquerware samples corresponding to the target lacquerware type. Specifically, different lighting parameters refer to the lighting environment applied to the lacquerware samples under different conditions during the lacquer test, mainly covering combinations and variations of light source type (such as standard D65 light source simulating natural sunlight, A light source simulating warm white light, etc.), lighting angle (such as 0° vertical incidence, 45° oblique incidence to highlight surface texture), lighting intensity (in lux, controlling the brightness of light), and color temperature (in Kelvin, affecting the warm and cool tones of the image). Differences in these parameters will significantly change the presentation effect of the lacquer layer image. For example, high-angle strong light can clearly expose defects such as orange peel texture on the lacquer surface, while low-angle weak light is more conducive to observing the uniformity of lacquer layer thickness. By collecting image data of key lacquer layers under different lighting parameters, the influence of lighting on the visual feature detection of the lacquer layer can be comprehensively analyzed. Then, based on the body manufacturing information, the optimal combination of lighting parameters that can most realistically reflect the quality of the lacquer layer and is most conducive to defect identification and process evaluation can be selected.
[0064] In some embodiments, step 130 specifically includes:
[0065] For each set of lighting parameters, based on the images of the lacquer layer of the body of multiple sample lacquerware corresponding to the target sample lacquerware type under the optimal polishing scheme and lighting parameters, the effective value of the lighting parameters is calculated.
[0066] Based on the effective values of each set of illumination parameters, the optimal illumination parameters are determined.
[0067] Specifically, for each set of illumination parameters, its effective value is calculated from two dimensions: First, the discernibility of key visual features. Image processing algorithms extract core features such as paint layer thickness uniformity, surface unevenness, and orange peel texture density, and the contrast-to-noise ratio (SNR) between these features and the background is calculated. A higher SNR indicates clearer features under illumination conditions, effectively reducing detection errors caused by light interference. Second, defect detection sensitivity. Based on a pre-set paint layer defect library (such as pinholes, runs, wrinkles, etc.), the number of identifiable defects and classification accuracy in images under different illumination parameters are statistically analyzed. A higher defect detection rate and a lower false alarm rate result in a higher score for the illumination parameter. To eliminate individual sample differences, the average of the above two indicators is taken for all samples of the same sample type, yielding sub-scores for the set of illumination parameters in the dimensions of discernibility and defect detection sensitivity. These sub-scores are then weighted and summed to obtain the effective value of the illumination parameter.
[0068] The illumination parameter with the largest effective value is taken as the optimal illumination parameter.
[0069] Step 140: For each key number of lacquer passes, based on the optimal lighting parameters, acquire the body lacquer layer image corresponding to the key number of lacquer passes, identify the polishing defect information corresponding to the key number of lacquer passes, and generate the polishing control strategy for the next number of lacquer passes based on the polishing defect information corresponding to the key number of lacquer passes.
[0070] In some embodiments, based on the paint layer image corresponding to the key paint layer number, the polishing defect information corresponding to the key paint layer number is identified, including:
[0071] The images of the lacquer layer corresponding to the key lacquer layer numbers are divided into multiple region images;
[0072] For each region image, an adaptive segmentation threshold is determined. Based on the adaptive segmentation threshold, a binary image corresponding to the region image is generated, and the surface smoothness corresponding to the region image is identified. Among them, the key polishing defect information corresponding to the number of lacquer layers includes the surface smoothness corresponding to each region image.
[0073] In some embodiments, determining the adaptive segmentation threshold corresponding to the region image includes:
[0074] Determine the optimal segmentation threshold for multiple sample region images;
[0075] Based on the optimal segmentation threshold corresponding to multiple sample region images, multiple threshold influencing factors are determined;
[0076] Based on multiple threshold influence factors and the optimal segmentation threshold corresponding to multiple sample region images, the adaptive segmentation threshold corresponding to the region image is determined.
[0077] Specifically, multiple sample region images cover scenarios with varying texture complexity, illumination reflectivity, and defect types. An iterative threshold optimization algorithm (such as Otsu's method combined with local contrast enhancement) is used to determine the optimal segmentation threshold for each sample image. Based on this dataset, key image feature factors are extracted, including texture complexity (quantified by the entropy value of the gray-level co-occurrence matrix), illumination reflectivity (calculated based on the mean of the V channels in the HSV space), and edge density (statistically calculated by detecting edge pixels using the Canny operator). For each key image feature factor, the optimal segmentation threshold and the value of the key image feature factor corresponding to the multiple sample region images are used as two variables. The absolute value of the Pearson correlation coefficient between the key image feature factor and the optimal segmentation threshold is calculated. Key image feature factors whose absolute Pearson correlation coefficient is greater than the absolute value threshold (e.g., 0.5) are used as threshold influencing factors.
[0078] When processing the lacquer layer images corresponding to the key lacquer layer numbers, in order to accurately capture local defect features, the original image is first divided into multiple non-overlapping unit images (e.g., a 512×512 pixel image is divided into 16 64×64 pixel units). Each unit serves as the smallest analysis unit, retaining independent texture, lighting, and defect information. Subsequently, predefined threshold influence factors (such as texture complexity, light reflectivity, edge density, etc.) are extracted for each unit image, and the statistical values of each factor within the unit are quantified (e.g., texture complexity is calculated using the entropy value of the gray-level co-occurrence matrix, and light reflectivity is obtained based on the mean of the V channels in the HSV space). Furthermore, the Euclidean distance between any two unit images on all threshold influence factors is calculated. An improved DBSCAN clustering algorithm is used to group the unit images based on the Euclidean distance between any two unit images on all threshold influence factors: units with an Euclidean distance less than the Euclidean distance threshold (e.g., 0.2) are grouped into the same cluster to form regional images with similar characteristics, while isolated units or small clusters are merged into neighboring large clusters or treated as independent regions, thereby dividing the lacquer layer image corresponding to the key lacquer path number into multiple regional images.
[0079] When generating binary images, an adaptive segmentation threshold is used as a benchmark. Pixels with grayscale values higher than the adaptive segmentation threshold are marked as foreground (defects or highly reflective areas), while those lower are marked as background (normal paint layer), resulting in a clear black-and-white binary image. Morphological processing (such as opening operations to remove isolated noise and closing operations to fill internal holes in defects) is then used to optimize the defect contour, improving the accuracy of subsequent analysis. Surface smoothness identification is based on the morphological characteristics of defects in the binary image: calculating the fill rate (the percentage of defect pixels) and contour complexity (calculated by chain code encoding, the ratio of contour perimeter to area). A high fill rate and irregular contour (such as jagged edges or high fractal dimension) indicate significant surface undulations (such as scratches or orange peel texture); conversely, a low fill rate and smooth contour (approaching a circle or rectangle) indicate good smoothness.
[0080] Based on a threshold influence factor constructed from a sample library, adaptive segmentation thresholds can be dynamically generated to suit different regional characteristics, effectively solving the segmentation failure problem of traditional fixed threshold methods in complex texture scenes, and improving defect recognition accuracy by 37%. Secondly, intelligent segmentation of regional images is achieved through an improved DBSCAN clustering algorithm, making defect boundaries highly consistent with actual distribution and reducing the risk of cross-regional false detection. Furthermore, based on morphological analysis and defect morphology quantification (fill rate, contour complexity) of binary images, micro-defects at the 0.1mm level can be accurately identified and surface flatness quantified, providing data support for differentiated polishing.
[0081] In some embodiments, a polishing control strategy for the next paint layer is generated based on polishing defect information corresponding to key paint layer numbers, including:
[0082] Based on the surface flatness corresponding to each region image, the number of polishing times for the corresponding polishing region is determined. The polishing control strategy for the next paint pass includes the number of polishing times for the corresponding polishing region of each region image.
[0083] Specifically, when generating the next polishing control strategy based on the polishing defect information corresponding to the key paint coat number, the number of polishing passes is determined by the difference between the surface smoothness of each area and the overall mean, ensuring accuracy and scientific rigor. Surface smoothness directly reflects the flatness of the paint layer surface. Different areas exhibit variations in smoothness due to factors such as polishing processes and paint characteristics. Calculating the difference between the surface smoothness of a polished area and the mean surface smoothness of all polished areas effectively measures the degree of flatness deviation of that area relative to the overall surface.
[0084] A large difference indicates that the area deviates significantly from the overall average smoothness, potentially indicating prominent defects such as deep scratches or noticeable orange peel texture, resulting in significant surface undulations. To eliminate these defects and make the paint surface in this area more consistent with other areas, it is necessary to increase the number of sanding passes. Through multiple sanding passes, raised areas can be gradually smoothed out, and recessed areas can be filled, effectively improving surface smoothness.
[0085] Conversely, if the difference is small, it indicates that the surface smoothness of that area is close to the overall average level, with fewer and less severe defects. In this case, excessive sanding will not only waste time and sanding materials but may also damage the original texture of the lacquer layer, or even cause the lacquer layer to thin, affecting the quality and durability of the lacquerware. Therefore, for such areas, the number of sanding operations should be appropriately reduced to preserve the original state of the lacquer layer as much as possible while ensuring surface quality.
[0086] This method of determining the number of polishing cycles based on the difference enables differentiated and refined control of the polishing process. It allows for the development of reasonable polishing strategies for different areas, effectively improving the smoothness and uniformity of the entire lacquerware surface and providing a good foundation for subsequent lacquering processes.
[0087] Figure 3 These are schematic diagrams of a machine vision-based lacquerware varnishing control system, as shown in some embodiments of this specification. Figure 3 As shown, a machine vision-based lacquerware varnishing control system may include an information acquisition module, a scheme optimization module, a parameter optimization module, and a lacquer varnishing control module.
[0088] The information acquisition module is used to acquire information about the fabrication of the embryo.
[0089] The scheme optimization module is used to determine the number of key lacquer layers and the optimal polishing scheme based on the production information of the lacquerware body and the total number of lacquer layers. The optimal polishing scheme includes the polishing grit number corresponding to each key lacquer layer.
[0090] The parameter optimization module is used to determine the optimal lighting parameters based on the manufacturing information of the body and the number of several key lacquer layers.
[0091] The varnish control module is used to acquire images of the paint layer corresponding to each key varnish layer number based on optimal lighting parameters, identify polishing defect information corresponding to the key varnish layer number based on the paint layer images corresponding to the key varnish layer number, and generate polishing control strategies for the next varnish layer number based on the polishing defect information corresponding to the key varnish layer number.
[0092] The lacquerware varnishing control system based on machine vision can apply the aforementioned machine vision-based lacquerware varnishing control method, which will not be elaborated further here.
[0093] Finally, it should be understood that the embodiments described in this specification are merely illustrative of the principles of the embodiments described herein. Other variations may also fall within the scope of this specification. Therefore, alternative configurations of the embodiments described herein are intended to be illustrative rather than limiting, and should be considered consistent with the teachings of this specification. Accordingly, the embodiments described herein are not limited to those explicitly introduced and described herein.
Claims
1. A machine vision-based method for controlling the lacquer application process on lacquerware, characterized in that, include: Obtain information on the fabrication of the embryo; Based on the production information of the lacquerware body and the total number of lacquer coats, several key lacquer coat numbers and optimal polishing schemes are determined. The optimal polishing scheme includes the polishing grit number corresponding to each key lacquer coat number. Based on the manufacturing information of the body and the number of key lacquer layers, the optimal lighting parameters are determined; For each key lacquer pass, based on the optimal lighting parameters, the image of the paint layer corresponding to the key lacquer pass is acquired. Based on the image of the paint layer corresponding to the key lacquer pass, the polishing defect information corresponding to the key lacquer pass is identified. Based on the polishing defect information corresponding to the key lacquer pass, the polishing control strategy for the next lacquer pass is generated. Based on the manufacturing information of the lacquerware body and the total number of lacquer coats, several key lacquer coat numbers and optimal polishing schemes were determined, including: Based on the manufacturing information of the body and the total number of lacquer coats, the target sample lacquerware type is determined from multiple sample lacquerware types; Based on the lacquer test data of the target sample lacquerware type, several key lacquer layer numbers are determined. The lacquer test data of the target sample lacquerware type includes images of the lacquer layer of the body of multiple sample lacquerware corresponding to the target sample lacquerware type under preset polishing schemes and preset lighting parameters for each lacquer layer number. Specifically, for each lacquer layer number, based on the images of the body lacquer layer of multiple sample lacquerware under preset polishing schemes and preset lighting parameters, key values of the lacquer layer number are determined, and based on the key values of each lacquer layer number, several key lacquer layer numbers are determined. Based on the lacquer test data of the target sample lacquerware type and the number of lacquer coats in several key areas, the optimal polishing scheme was determined. Based on the lacquer layer images corresponding to the key lacquer layer numbers, the polishing defect information corresponding to the key lacquer layer numbers is identified, including: The images of the lacquer layer corresponding to the key lacquer layer numbers are divided into multiple region images; For each region image, an adaptive segmentation threshold is determined. Based on the adaptive segmentation threshold, a binary image corresponding to the region image is generated, and the surface smoothness corresponding to the region image is identified. Among them, the key polishing defect information corresponding to the number of lacquer layers includes the surface smoothness corresponding to each region image.
2. The lacquer application control method based on machine vision according to claim 1, characterized in that, The manufacturing information of the substrate includes the type of substrate and the composition of the ash used in the puttying process.
3. The lacquer application control method based on machine vision according to claim 1, characterized in that, Based on images of the lacquer layer on the body of multiple sample lacquerware pieces under preset polishing schemes and lighting parameters, the key values for the number of lacquer layers were determined, including: For each sample lacquerware, based on the images of the body lacquer layer with the number of lacquer layers and the body lacquer layer images of adjacent lacquer layers, the performance difference value corresponding to the number of lacquer layers is determined. Based on the performance difference value of each sample lacquerware corresponding to the number of lacquer layers, calculate the first difference value corresponding to the number of lacquer layers; Based on the lacquer layer image of the body of each sample lacquerware under the preset polishing scheme and preset lighting parameters, the second difference value corresponding to the number of lacquer layers is calculated. Based on the first and second difference values, the key value for the number of lacquer layers is calculated.
4. The lacquer application control method based on machine vision according to claim 2, characterized in that, The lacquer test data for the target sample lacquerware type includes images of the lacquer layer of the body for each key lacquer layer under different polishing schemes and preset lighting parameters for multiple sample lacquerware corresponding to the target sample lacquerware type. Based on the lacquer test data of the corresponding sample lacquerware type and several key lacquer layers, the optimal polishing scheme was determined, including: For each polishing scheme, based on the images of the lacquer layer of the base body of multiple sample lacquerware corresponding to the target sample lacquerware type under the polishing scheme and preset lighting parameters, the priority value of the polishing scheme is calculated. The optimal polishing scheme is determined based on the priority value of each polishing scheme.
5. The lacquer application control method based on machine vision according to claim 2, characterized in that, The lacquer test data for the target sample lacquerware type includes images of the lacquer layer of the body for each key lacquer layer under the optimal polishing scheme and different lighting parameters for multiple sample lacquerware corresponding to the target sample lacquerware type. Based on the fabrication information of the base material and several key lacquer layers, the optimal lighting parameters were determined, including: For each set of lighting parameters, based on the images of the lacquer layer of the body of multiple sample lacquerware corresponding to the target sample lacquerware type under the optimal polishing scheme and lighting parameters, the effective value of the lighting parameters is calculated. Based on the effective values of each set of illumination parameters, the optimal illumination parameters are determined.
6. The lacquer application control method based on machine vision according to claim 1, characterized in that, Determine the adaptive segmentation threshold corresponding to the region image, including: Determine the optimal segmentation threshold for multiple sample region images; Based on the optimal segmentation threshold corresponding to multiple sample region images, multiple threshold influencing factors are determined; Based on multiple threshold influence factors and the optimal segmentation threshold corresponding to multiple sample region images, the adaptive segmentation threshold corresponding to the region image is determined.
7. The lacquer application control method based on machine vision according to claim 1, characterized in that, Based on the sanding defect information corresponding to the key paint application layers, a sanding control strategy for the next paint application layer is generated, including: Based on the surface flatness corresponding to each region image, the number of polishing times for the corresponding polishing region is determined. The polishing control strategy for the next paint pass includes the number of polishing times for the corresponding polishing region of each region image.
8. A lacquerware varnishing control system based on machine vision, characterized in that, The lacquer application control method based on machine vision as described in claim 1 includes: The information acquisition module is used to acquire information about the fabrication of the embryo. The scheme optimization module is used to determine the number of key lacquer layers and the optimal polishing scheme based on the production information of the lacquerware body and the total number of lacquer layers. The optimal polishing scheme includes the polishing grit number corresponding to each key lacquer layer. The parameter optimization module is used to determine the optimal lighting parameters based on the manufacturing information of the body and the number of several key lacquer layers. The varnish control module is used to acquire images of the paint layer corresponding to each key varnish layer number based on optimal lighting parameters, identify polishing defect information corresponding to the key varnish layer number based on the paint layer images corresponding to the key varnish layer number, and generate polishing control strategies for the next varnish layer number based on the polishing defect information corresponding to the key varnish layer number.