Machine vision-based rice cake forming mold wear degree evaluation method and system

Abstract

The invention discloses a machine vision-based rice cake forming mold wear degree evaluation method and system, and relates to the technical field of machine vision. The method comprises the following steps: performing multi-angle acquisition on a rice cake forming grinding tool through machine vision to generate a mold image data set; multi-dimensional feature extraction is conducted on the mold image data set, a wear feature parameter set is obtained, multi-stage wear evaluation is conducted based on the wear feature parameter set, a mold surface wear distribution atlas is constructed, and the mold surface wear distribution atlas comprises mold wear levels; performing rice cake process matching according to the mold wear level, generating a mold health assessment report, and triggering a maintenance decision instruction to perform self-adaptive repair on the rice cake forming mold. The technical problems that in the prior art, rice cake forming mold wear degree evaluation depends on artificial experience, evaluation efficiency is low, and accuracy is insufficient are solved, and the technical effects of improving mold wear detection precision and achieving automatic evaluation and repair decision making are achieved.

Description

Technical Field

[0001] The present invention relates to the technical field of machine vision, and in particular to a method and system for evaluating the wear degree of a rice cake forming mold based on machine vision. Background Art

[0002] During the rice cake production process, the quality of the mold directly affects the appearance consistency and product qualification rate of the rice cake. Traditional rice cake molding molds are prone to wear during long-term use, leading to problems such as blurred molding edges, unclear patterns, and dimensional deviations. Currently, the assessment of mold wear mainly relies on manual visual inspection and empirical judgment. This is not only highly subjective and inefficient, but also difficult to achieve standardized and quantitative evaluation, making it difficult to meet the high requirements of the modern food processing industry for product consistency and automation. Especially in multi-batch continuous production, if the health status of the mold cannot be identified in a timely and accurate manner, it is very easy to cause fluctuations in the quality of batch products or even scrap them, increasing production costs and quality risks. Summary of the Invention

[0003] The present application provides a method and system for evaluating the degree of wear of rice cake forming molds based on machine vision, which solves the technical problems in the prior art that the evaluation of the degree of wear of rice cake forming molds relies on manual experience, has low evaluation efficiency and insufficient accuracy.

[0004] In a first aspect of the present application, a method for evaluating the degree of wear of a rice cake forming mold based on machine vision is provided, the method comprising:

[0005] Machine vision is used to capture multi-angle images of rice cake forming molds to generate a mold image dataset. Multi-dimensional feature extraction is performed on the mold image dataset to obtain a wear characteristic parameter group. Based on the wear characteristic parameter group, a multi-stage wear assessment is performed to construct a mold surface wear distribution map. The mold surface wear distribution map includes the mold wear level. Rice cake process matching is performed according to the mold wear level, and a mold health assessment report is generated. Maintenance decision instructions are triggered to perform adaptive repair on the rice cake forming mold.

[0006] The second aspect of the present application provides a rice cake forming mold wear assessment system based on machine vision, the system comprising:

[0007] The image acquisition module is used to perform multi-angle acquisition of the rice cake forming mold through machine vision to generate a mold image data set; the wear assessment module is used to perform multi-dimensional feature extraction on the mold image data set to obtain a wear characteristic parameter group, perform multi-stage wear assessment based on the wear characteristic parameter group, and construct a mold surface wear distribution map, wherein the mold surface wear distribution map includes the mold wear level; the report generation module is used to match the rice cake process according to the mold wear level, generate a mold health assessment report, and trigger maintenance decision instructions to adaptively repair the rice cake forming mold.

[0008] One or more technical solutions provided in this application have at least the following technical effects or advantages:

[0009] First, machine vision is used to capture images of the rice cake forming mold from multiple angles, generating a mold image dataset. Multidimensional feature extraction is then performed on this mold image dataset to obtain a set of wear characteristic parameters. Based on this set of wear characteristic parameters, a multi-stage wear assessment is performed to construct a mold surface wear distribution map, which includes the mold wear level. Finally, the rice cake process is matched according to the mold wear level, generating a mold health assessment report that triggers maintenance decision-making instructions for adaptive repair of the rice cake forming mold. This method addresses the existing technical issues of relying on manual experience, resulting in low evaluation efficiency and insufficient accuracy in rice cake forming mold wear assessment. It achieves the technical effect of improving mold wear detection accuracy and enabling automated assessment and repair decision-making. BRIEF DESCRIPTION OF THE DRAWINGS

[0010] In order to more clearly illustrate the technical solutions in the embodiments of the present invention, the following briefly introduces the drawings required for use in the description of the embodiments. Obviously, the drawings described below are only some embodiments of the present invention. For ordinary technicians in this field, other drawings can be obtained based on these drawings without creative work.

[0011] Figure 1 A flow chart of a method for evaluating the degree of wear of a rice cake forming mold based on machine vision provided in an embodiment of the present application;

[0012] Figure 2 Schematic diagram of the structure of the rice cake forming mold wear degree assessment system based on machine vision provided in an embodiment of the present application.

[0013] Description of reference numerals: image acquisition module 11 , wear assessment module 12 , report generation module 13 . DETAILED DESCRIPTION

[0014] This application solves the technical problems in the prior art that the wear degree assessment of rice cake forming molds relies on manual experience, has low assessment efficiency and insufficient accuracy by providing a method and system for assessing the wear degree of rice cake forming molds based on machine vision.

[0015] The following will be combined with the accompanying drawings in the embodiments of this application to clearly and completely describe the technical solutions in the embodiments of this application. Obviously, the embodiments described are only some of the embodiments of this application, not all of them. Based on the embodiments in this application, all other embodiments obtained by ordinary technicians in this field without making creative efforts are within the scope of protection of this application.

[0016] It should be noted that the terms "including" and "having" are intended to cover non-exclusive inclusions. For example, a process, method, system, product or server that includes a series of steps or units is not necessarily limited to those steps or units clearly listed, but may include other steps or modules that are not clearly listed or are inherent to these processes, methods, products or devices.

[0017] Example 1, as Figure 1 As shown, the present application provides a method for evaluating the wear degree of rice cake forming molds based on machine vision, wherein the method includes:

[0018] The rice cake forming mold is captured from multiple angles through machine vision to generate a mold image dataset.

[0019] A multi-angle industrial camera array is used to collect three-dimensional image data of the surface and edges of rice cake forming molds to generate a mold image dataset.

[0020] Furthermore, machine vision is used to capture the rice cake forming mold from multiple angles to generate a mold image dataset, including:

[0021] A ring-shaped LED light source array is configured and initial lighting parameters are set. The ring-shaped LED light source array is activated according to the initial lighting parameters to control a multi-angle industrial camera array to perform surface acquisition on a rice cake forming mold to obtain multiple surface images. The multiple surface images are arranged according to an acquisition time sequence to determine multiple two-dimensional image sequences. Three-dimensional point cloud reconstruction is performed on the multiple two-dimensional image sequences to generate three-dimensional point cloud data of the mold. The three-dimensional point cloud data of the mold is integrated to generate the mold image dataset.

[0022] A circular LED light array is arranged on the image acquisition platform of the rice cake forming mold. This array is symmetrically arranged around the mold forming area and features adjustable illumination angle, brightness, and frequency. Initial illumination parameters, including intensity, incident angle, and uniformity, are preset based on the mold material (such as stainless steel or alloy steel) and the reflective properties of the surface. Once these initial illumination parameters are set, the control module activates the light array, enabling it to work in conjunction with the industrial camera.

[0023] Multiple high-resolution industrial cameras are arranged at predetermined angles, for example, at 12 equally spaced angles (every 30°) around the mold, or a single industrial camera can be combined with a robotic arm for rotational acquisition. Each camera uses a synchronized trigger control module to capture images of the mold surface under stable lighting conditions, obtaining multiple surface images from different angles. The acquisition system records the camera number and capture timestamp for each image. Based on this information, the images are organized and categorized by time sequence and capture angle, constructing multiple 2D image sequences.

[0024] A 3D reconstruction algorithm (such as structured light reconstruction, stereo vision, photometric stereo reconstruction, or SfM+MVS) is used to process multiple 2D image sequences, identify the spatial correspondence and depth information of the mold surface, and generate point cloud data containing the 3D geometric features of the mold surface. The mold 3D point cloud data contains the 3D coordinates (X, Y, Z) of each surface point and optional texture color or grayscale value. The 3D point cloud data generated from different perspectives is subjected to coordinate alignment, posture registration, and redundancy elimination. High-precision 3D reconstruction of the mold surface is achieved through ICP (Iterative Closest Point) or feature point-based point cloud fusion algorithms. Finally, the processed point cloud data is unified and integrated to generate a mold image dataset with a standard spatial reference system.

[0025] Multidimensional feature extraction is performed on the mold image dataset to obtain a wear feature parameter group, a multi-stage wear assessment is performed based on the wear feature parameter group, and a mold surface wear distribution map is constructed, wherein the mold surface wear distribution map includes a mold wear grade.

[0026] In an embodiment of the present application, geometric deformation features of the mold image are extracted through an edge detection algorithm to obtain a wear feature parameter group, which includes geometric deformation features, surface texture features and edge sharpness features; based on the wear feature parameter group, the mold wear status is graded and evaluated through a training model to determine whether the mold is in a light, moderate or heavy wear stage, and the evaluation results are mapped to the corresponding positions of the mold three-dimensional point cloud model to generate a color-coded mold surface wear distribution map to intuitively reflect the degree of wear in each area.

[0027] Furthermore, multi-dimensional feature extraction is performed on the mold image dataset to obtain a wear feature parameter group, including:

[0028] Edge detection is performed on the mold image dataset to obtain geometric deformation features; surface texture analysis is performed on the mold image dataset through a grayscale co-occurrence matrix to obtain surface texture features; mold discreteness is calculated based on the mold three-dimensional point cloud data, and edge sharpness features are obtained based on the mold discreteness combined with the edge gradient change rate; the geometric deformation features, the surface texture features and the edge sharpness features are normalized to construct a multidimensional feature vector as the wear feature parameter group.

[0029] By performing edge detection on the mold image dataset, the edge contour of the mold surface is extracted, and the rate of change of the contour curvature is calculated to obtain geometric deformation features. Specifically, the mold image is first processed using an edge detection algorithm (such as the Canny operator or the Sobel operator) to extract the mold edge contour. Next, the curvature of the edge is calculated using the derivative information of the curve to evaluate the degree of curvature of the contour. The rate of change of curvature between adjacent points is further calculated to reflect the geometric changes in the mold surface caused by wear or deformation. Finally, the rate of change of contour curvature is used as a geometric deformation feature to provide a basis for subsequent wear assessment.

[0030] Surface texture feature analysis is performed on mold images. Using a gray-level co-occurrence matrix method, texture-related statistics are extracted from a local window of the mold image. These metrics include energy, contrast, entropy, and correlation. These metrics are used to measure changes in mold surface roughness, corrosion spots, fine cracks, or texture degradation caused by machining wear, forming surface texture features with regional resolution capabilities.

[0031] Spatial structure analysis is performed based on the 3D point cloud data reconstructed from the mold image sequence. Specifically, the local density distribution and normal vector changes of the mold point cloud model are calculated to obtain the degree of spatial distribution dispersion (i.e., mold discreteness). Combined with the grayscale gradient change rate of the edge area in the mold's 2D image, the degree of boundary blur and blunting trend are evaluated, and an edge sharpness feature is constructed to reflect changes in boundary clarity.

[0032] Geometric deformation, surface texture, and edge sharpness features are normalized separately, and each dimension is mapped to a unified numerical range using Z-score or Min-Max normalization methods to eliminate dimensional differences and ensure consistency and robustness in subsequent calculations. The normalized multidimensional feature vectors are aggregated and integrated by region to ultimately construct a representative set of wear characteristic parameters, providing a data foundation for subsequent wear grade determination and atlas construction.

[0033] Furthermore, a multi-stage wear assessment is performed based on the wear characteristic parameter group to construct a mold surface wear distribution map, including:

[0034] The mold production batch data is retrieved for wear analysis, and a wear threshold curve is constructed, wherein the wear threshold curve corresponds to the mold production batch data; the rice cake forming mold is continuously inspected to generate multiple local wear values, and when any of the multiple local wear values exceeds the wear threshold curve, a sampling instruction is generated; secondary image sampling is performed on the rice cake forming mold through the sampling instruction to obtain mold deformation data; a multi-stage wear assessment is performed on the rice cake forming mold based on the mold deformation data and the wear characteristic parameter group, and a wear distribution map of the mold surface is constructed.

[0035] Preferably, mold production batch data, including relevant data such as production date, production process, usage time, and load conditions, is obtained from a database. A wear threshold curve is constructed by analyzing historical production data. The wear threshold curve reflects the expected variation in mold wear under different usage conditions, and the wear threshold curve corresponds to the mold production batch data, facilitating wear prediction based on the usage of different batches of molds. A real-time monitoring system is used to regularly collect image data of the rice cake forming mold surface and perform image processing to obtain multiple local wear values. The local wear values represent the wear conditions of the mold surface in that area. The local wear values are obtained by calculating the degree of wear in each local area, using a weighted calculation of the curvature change of the surface profile, crack length, and surface roughness. When a local wear value exceeds a preset wear threshold curve, indicating that the wear in that area has exceeded a predetermined tolerance range, further testing is required. A sampling instruction is then generated to perform secondary image sampling on that mold area to obtain mold deformation data.

[0036] By combining the acquired mold deformation data with the previously extracted wear characteristic parameter set, a multi-stage wear assessment is performed. During this process, mold deformation data (such as shape change, curvature change, surface cracks, etc.) and wear characteristics (such as geometric deformation characteristics, surface texture characteristics, edge sharpness characteristics, etc.) are combined to evaluate the wear of the mold at each stage of use and determine whether the mold has entered a high-wear area, a potential failure area, or an area requiring repair. Based on the results of the multi-stage wear assessment, a mold surface wear distribution map is constructed. The mold surface wear distribution map shows the degree of wear in each area of the mold surface. The color, brightness, or pattern of the area can reflect the severity of the wear.

[0037] Furthermore, the process of constructing the mold wear level includes:

[0038] The three-dimensional point cloud data of the mold is spatially registered to obtain deformation point cloud data, deformation calculation is performed based on the deformation point cloud data, and the local deformation deviation value is determined; the surface change curvature of the rice cake forming mold is constructed based on the geometric deformation characteristics, and the wear gradient area is divided according to the surface change curvature; the contrast attenuation rate of the surface texture feature and the mold discreteness of the edge sharpness feature are extracted to construct a wear level quantization matrix; the wear gradient area is screened according to the local deformation deviation value, and the wear level quantization matrix is mapped to the three-dimensional point cloud data of the mold according to the screening result to obtain the mold wear level, and the mold wear level is added to the mold surface wear distribution map.

[0039] Preferably, the three-dimensional point cloud data of the mold is aligned with the standard CAD point cloud, and the abnormal point cloud data is extracted to obtain the deformed point cloud data; the deviation is calculated based on the deformed point cloud data and the standard point cloud data to obtain the local deformation deviation value, which represents the morphological changes in the area caused by wear or use.

[0040] The surface curvature of the mold surface is constructed based on the geometric deformation characteristics, and different wear gradient areas are divided based on this. Specifically, the edge contour curve is extracted from the mold 3D point cloud data, and the contour curvature k of each point is calculated by the curvature analysis method. The curvature change rate index Δk = |k-k0| / k0 is used to evaluate the degree of deviation of the current mold from the standard curvature k0 in the initial unworn state, where k0 is the curvature value of the corresponding position in the standard mold model. The greater the curvature change rate, the more significant the geometric deformation of the area, indicating that the area has undergone more severe wear. According to the numerical range of the curvature change rate, multiple thresholds are set to divide different wear gradient areas. For example, the area with a change rate below 5% is defined as a light wear area, 5% to 15% is defined as a moderate wear area, and the area above 15% is marked as a heavy wear area.

[0041] A wear level quantification matrix is constructed. Specifically, a grayscale co-occurrence matrix analysis is performed on the mold image dataset to extract the contrast features of the surface texture. The texture contrast of the current image is compared with the standard contrast in the initial unworn state, and the contrast decay rate of the surface texture features is calculated to quantify the degree of texture blur. At the same time, based on the mold three-dimensional point cloud data and edge detection results, the spatial distribution variance of the mold edge area point cloud is calculated, and the mold discreteness is extracted as a quantitative indicator of the edge sharpness feature. The contrast decay rate and mold discreteness are normalized separately to construct a two-dimensional wear level quantification matrix composed of multiple regional feature values.

[0042] Based on the local deformation deviation values calculated from the deformation point cloud data, various areas of the mold surface are screened, and areas with deviation values exceeding a preset threshold are identified and preliminarily determined to be suspected wear areas. Within these areas, a membership function is constructed by combining characteristic information such as geometric deformation, surface texture, and edge sharpness contained in the wear feature parameter set to quantify the degree to which each feature is attributed to different wear levels (mild, moderate, and severe).

[0043] First-order (slight) wear membership function: Among them, x represents the geometric deformation characteristic value of the area, a1 is the minimum threshold of slight wear, indicating the starting point from no deformation to slight deformation, b1 is the maximum threshold of slight wear, indicating the end point of slight wear, that is, if it exceeds this value, it cannot be considered as slight wear, μ1(x) is the membership value of the geometric deformation characteristic x to the slight wear level, a value of 1 indicates that it completely belongs to slight wear, and a value of 0 indicates that it does not belong to slight wear at all.

[0044] Secondary (moderate) wear membership function: Among them, x represents the geometric deformation characteristic value of the area, a2 is the lowest threshold of moderate wear, indicating the starting point of moderate wear, b2 is the middle threshold of moderate wear, indicating that within this range, deformation begins to be obvious and wear gradually increases, c2 is the maximum threshold of moderate wear, indicating the end point of moderate wear. Exceeding this value means entering the severe wear stage, and μ2(x) is the membership value of the geometric deformation characteristic x to the moderate wear level.

[0045] Membership function of level three (severe) wear: Among them, x represents the geometric deformation characteristic value of the area, a3 is the minimum threshold of severe wear, indicating the starting point of severe wear, b3 is the maximum threshold of severe wear, indicating the boundary of severe wear. If it exceeds this value, it is considered to be severely worn, and μ3(x) is the membership value of the geometric deformation characteristic x to the severe wear level.

[0046] Surface texture features like contrast decay and edge sharpness dispersion are also evaluated using membership functions to quantify their contributions to different wear levels. Each feature value in each region is substituted into these functions to determine its membership to each level, which is then used in subsequent weighted evaluation and comprehensive assessment.

[0047] The analytic hierarchy process (AHP) was used to assign weights to each feature and determine their relative importance in the comprehensive evaluation. A fuzzy comprehensive evaluation method was then used to comprehensively judge the selected areas and, based on the principle of maximum membership, determine the specific wear level to which they belonged. Finally, the resulting wear level labels were mapped onto the mold's 3D point cloud data, creating a wear distribution map with level information. This serves as a visual representation of the mold's health status and a basis for subsequent maintenance decisions.

[0048] Furthermore, a multi-stage wear assessment is performed on the rice cake forming mold based on the mold deformation data in combination with the wear characteristic parameter group to construct a wear distribution map of the mold surface, including:

[0049] A multi-stage wear assessment is performed on the rice cake forming mold based on the mold deformation data in combination with the wear characteristic parameter group: S1: Preliminary wear area positioning is performed on the rice cake forming mold based on the geometric deformation feature in combination with the mold deformation data, and multiple wear candidate areas are determined; S2: Sub-pixel level analysis and identification is performed in combination with the surface texture feature in the multiple wear candidate areas to identify multiple wear defect data; S3: Finite element simulation is used to verify the spatial correlation between the multiple wear candidate areas and the stress concentration area based on the edge sharpness feature, and multiple wear deformation data are obtained; S4: Wear prediction is performed on the multiple wear candidate areas based on the multiple wear defect data and the multiple wear deformation data, and a wear prediction development trend graph is generated; wear assessment is performed according to the wear prediction development trend graph in combination with the mold wear level, and the mold surface wear distribution graph is constructed.

[0050] By combining mold deformation data and a set of wear characteristic parameters, a multi-stage wear assessment of the rice cake forming mold is conducted, ultimately constructing a mold surface wear distribution map. Specifically, the wear areas of the rice cake forming mold are initially located using geometric deformation features and mold deformation data. Specifically, the mold surface deformation data is used to screen candidate areas with deformation deviation values exceeding 0.1mm, and multiple candidate wear areas are identified. Within the multiple candidate wear areas, these areas are analyzed and identified at a more precise sub-pixel level in combination with the surface texture features. Detailed analysis of surface texture features (such as surface roughness and contrast attenuation) further identifies specific wear defect data. Using edge sharpness features, finite element simulation methods are used to verify the spatial correlation between multiple candidate wear areas and possible stress concentration areas on the mold surface. Through simulation analysis, multiple wear deformation data are obtained. Based on the multiple wear defect data and multiple wear deformation data, wear predictions are performed on the multiple candidate wear areas. Machine learning, regression analysis, and other methods are used to predict the wear evolution process and generate corresponding wear prediction development trend maps. Combining the wear prediction trend chart with the mold wear grade, a comprehensive wear assessment is performed on each wear area, and a mold surface wear distribution map is constructed. This map shows the wear status of each area on the mold surface, including areas with light, moderate, and severe wear, as well as the future wear trends of each area, providing a scientific basis for the maintenance and management of rice cake molds.

[0051] Furthermore, a wear assessment is performed according to the wear prediction development trend diagram in combination with the mold wear grade to construct a mold surface wear distribution diagram, including:

[0052] The three-dimensional point cloud data of the mold is converted into a triangular facet mesh, and the triangular facet mesh is color-coded and rendered according to the mold wear level to generate a rendered mesh; the multiple wear candidate areas are traversed for superposition and magnification analysis to generate a surface morphology view; according to the wear prediction development trend chart, the rendered mesh and the surface morphology view are comprehensively analyzed to construct the mold surface wear distribution map.

[0053] Preferably, the three-dimensional point cloud data of the mold is converted into a triangular facet mesh model, and these triangular facet meshes are color-coded and rendered according to the wear level of the mold. By setting different colors to represent different wear levels, lightly worn areas may be displayed in green or yellow, and severely worn areas may be displayed in red or dark tones, thereby clearly showing the wear distribution on the mold surface. After generating the rendered mesh, the multiple candidate wear areas previously identified are traversed, and these areas are superimposed and magnified for analysis. By magnifying the local area, the details of the wear can be accurately identified, revealing possible fatigue points and surface damage such as cracks, dents or deformations. Based on the wear prediction development trend map, combined with the mold wear level, the rendered mesh and the surface morphology view are comprehensively analyzed. The wear prediction development trend map reflects the evolution of mold wear, and taking into account the actual wear status and possible future changes, the system evaluates the current wear condition of the mold and predicts future wear development trends. On this basis, the rendered mesh and the surface morphology view are superimposed and analyzed to construct a comprehensive and accurate mold surface wear distribution map.

[0054] The rice cake process is matched according to the mold wear level, a mold health assessment report is generated, and a maintenance decision instruction is triggered to adaptively repair the rice cake forming mold.

[0055] Based on the mold wear level, the system matches the rice cake production process to the desired one. By analyzing the relationship between mold wear and production processes, the system optimizes process parameters during the rice cake forming process. This analysis then generates a mold health assessment report detailing the mold's current health, wear level, and potential future trends. Based on this assessment report, the system automatically triggers maintenance decision instructions. Taking into account wear conditions and production needs, the system guides adaptive repair of the rice cake forming mold, ensuring the mold continues to operate efficiently and optimally, extending its service life and minimizing downtime.

[0056] Furthermore, the rice cake process is matched according to the mold wear level, a mold health assessment report is generated, and a maintenance decision instruction is triggered to adaptively repair the rice cake forming mold, including:

[0057] Retrieve the rice cake pressing process parameter database, match the mold wear level according to the mold surface wear distribution map with the rice cake pressing process parameter database, and formulate a process compensation plan based on the matching result; perform real-time monitoring on the rice cake forming mold to generate real-time monitoring data, perform life assessment on the rice cake forming mold based on historical production data and the real-time monitoring data, and generate a mold health assessment report, wherein the mold health assessment report includes a remaining life prediction value of the rice cake forming mold; trigger a maintenance decision instruction based on the remaining life prediction value to perform adaptive repair on the rice cake forming mold.

[0058] Specifically, the system retrieves a rice cake pressing process parameter database, which stores the optimal production process parameters for different wear levels. The mold wear level is compared and matched with the parameters in the pressing process parameter database to determine the most suitable rice cake production process for the current wear state. Based on the matching results, the system develops a corresponding process compensation plan to adjust various process parameters during the rice cake production process, such as pressing pressure, molding temperature, and pressing time, to compensate for the impact of mold wear and ensure production stability and product quality. Next, the rice cake forming mold is monitored in real time. Real-time monitoring data is collected and compared with historical production data to generate a mold health assessment report. The mold health assessment report includes a predicted remaining lifespan for the rice cake forming mold, which is used to assess the mold's future usage and predict potential failures or wear issues. Based on the predicted remaining lifespan, the system automatically triggers maintenance decision instructions, guiding adaptive repair of the rice cake forming mold. By adjusting the repair strategy and rationally scheduling and prioritizing repairs, the system ensures that the mold continues to operate in optimal condition, extending its service life and reducing production downtime.

[0059] Furthermore, triggering a maintenance decision instruction based on the remaining life prediction value to adaptively repair the rice cake forming mold includes:

[0060] Based on the mold health assessment report, the maintenance decision instruction is triggered to perform distribution analysis of the mold wear level and determine multiple areas to be repaired; repair analysis is performed on the multiple areas to be repaired according to the remaining life prediction value, and a repair priority sequence is set; maintenance analysis is performed based on the repair priority sequence, and a three-dimensional repair path is generated. According to the three-dimensional repair path, the rice cake forming mold is adaptively repaired.

[0061] Based on the predicted RLU, the system first triggers a maintenance decision and then conducts a detailed analysis of the mold's wear distribution based on the mold health assessment report. By analyzing the mold surface wear distribution, the system identifies multiple areas requiring repair. These areas exhibit high levels of wear and may impact the performance and production efficiency of the rice cake mold. Repair analysis is then performed on these areas based on the predicted RLU. By comprehensively considering the mold's remaining life, the importance of the wear area, and the cost of repair, the system sets a repair priority sequence for each area. This sequence helps maintenance personnel prioritize repair work, prioritizing areas with the greatest impact on production and the shortest remaining life, thereby maximizing repair benefits and extending the mold's service life. After determining the repair priority, the system further conducts a maintenance analysis and generates a corresponding 3D repair path based on the repair priority sequence. This path considers spatial constraints, repair sequence, repair methods, and process requirements during the repair process, ensuring efficient and orderly repair. Finally, the system adaptively repairs the rice cake mold according to the 3D repair path.

[0062] In summary, the embodiments of the present application have at least the following technical effects:

[0063] First, machine vision is used to capture images of the rice cake forming mold from multiple angles, generating a mold image dataset. Multidimensional feature extraction is then performed on this mold image dataset to obtain a set of wear characteristic parameters. Based on this set of wear characteristic parameters, a multi-stage wear assessment is performed to construct a mold surface wear distribution map, which includes the mold wear level. Finally, the rice cake process is matched according to the mold wear level, generating a mold health assessment report that triggers maintenance decision-making instructions for adaptive repair of the rice cake forming mold. This method addresses the existing technical issues of relying on manual experience, resulting in low evaluation efficiency and insufficient accuracy in rice cake forming mold wear assessment. It achieves the technical effect of improving mold wear detection accuracy and enabling automated assessment and repair decision-making.

[0064] Example 2, based on the same inventive concept as the method for evaluating the degree of wear of rice cake forming molds based on machine vision in the above embodiment, Figure 2 As shown, the present application provides a rice cake forming mold wear assessment system based on machine vision, wherein the system includes:

[0065] The image acquisition module 11 is used to perform multi-angle acquisition of the rice cake forming mold through machine vision to generate a mold image data set; the wear assessment module 12 is used to perform multi-dimensional feature extraction on the mold image data set to obtain a wear feature parameter group, perform multi-stage wear assessment based on the wear feature parameter group, and construct a mold surface wear distribution map, wherein the mold surface wear distribution map includes the mold wear level; the report generation module 13 is used to match the rice cake process according to the mold wear level, generate a mold health assessment report, and trigger a maintenance decision instruction to adaptively repair the rice cake forming mold.

[0066] Furthermore, the image acquisition module 11 is used to perform the following method:

[0067] A ring-shaped LED light source array is configured and initial lighting parameters are set. The ring-shaped LED light source array is activated according to the initial lighting parameters to control a multi-angle industrial camera array to perform surface acquisition on a rice cake forming mold to obtain multiple surface images. The multiple surface images are arranged according to an acquisition time sequence to determine multiple two-dimensional image sequences. Three-dimensional point cloud reconstruction is performed on the multiple two-dimensional image sequences to generate three-dimensional point cloud data of the mold. The three-dimensional point cloud data of the mold is integrated to generate the mold image dataset.

[0068] Furthermore, the wear assessment module 12 is configured to perform the following method:

[0069] Edge detection is performed on the mold image dataset to obtain geometric deformation features; surface texture analysis is performed on the mold image dataset through a grayscale co-occurrence matrix to obtain surface texture features; mold discreteness is calculated based on the mold three-dimensional point cloud data, and edge sharpness features are obtained based on the mold discreteness combined with the edge gradient change rate; the geometric deformation features, the surface texture features and the edge sharpness features are normalized to construct a multidimensional feature vector as the wear feature parameter group.

[0070] Furthermore, the wear assessment module 12 is configured to perform the following method:

[0071] The mold production batch data is retrieved for wear analysis, and a wear threshold curve is constructed, wherein the wear threshold curve corresponds to the mold production batch data; the rice cake forming mold is continuously inspected to generate multiple local wear values, and when any of the multiple local wear values exceeds the wear threshold curve, a sampling instruction is generated; secondary image sampling is performed on the rice cake forming mold through the sampling instruction to obtain mold deformation data; a multi-stage wear assessment is performed on the rice cake forming mold based on the mold deformation data and the wear characteristic parameter group, and a wear distribution map of the mold surface is constructed.

[0072] Furthermore, the wear assessment module 12 is configured to perform the following method:

[0073] The three-dimensional point cloud data of the mold is spatially registered to obtain deformation point cloud data, deformation calculation is performed based on the deformation point cloud data, and the local deformation deviation value is determined; the surface change curvature of the rice cake forming mold is constructed based on the geometric deformation characteristics, and the wear gradient area is divided according to the surface change curvature; the contrast attenuation rate of the surface texture feature and the mold discreteness of the edge sharpness feature are extracted to construct a wear level quantization matrix; the wear gradient area is screened according to the local deformation deviation value, and the wear level quantization matrix is mapped to the three-dimensional point cloud data of the mold according to the screening result to obtain the mold wear level, and the mold wear level is added to the mold surface wear distribution map.

[0074] Furthermore, the wear assessment module 12 is configured to perform the following method:

[0075] A multi-stage wear assessment is performed on the rice cake forming mold based on the mold deformation data in combination with the wear characteristic parameter group: S1: Preliminary wear area positioning is performed on the rice cake forming mold based on the geometric deformation feature in combination with the mold deformation data, and multiple wear candidate areas are determined; S2: Sub-pixel level analysis and identification is performed in combination with the surface texture feature in the multiple wear candidate areas to identify multiple wear defect data; S3: Finite element simulation is used to verify the spatial correlation between the multiple wear candidate areas and the stress concentration area based on the edge sharpness feature, and multiple wear deformation data are obtained; S4: Wear prediction is performed on the multiple wear candidate areas based on the multiple wear defect data and the multiple wear deformation data, and a wear prediction development trend graph is generated; wear assessment is performed according to the wear prediction development trend graph in combination with the mold wear level, and the mold surface wear distribution graph is constructed.

[0076] Furthermore, the wear assessment module 12 is configured to perform the following method:

[0077] The three-dimensional point cloud data of the mold is converted into a triangular facet mesh, and the triangular facet mesh is color-coded and rendered according to the mold wear level to generate a rendered mesh; the multiple wear candidate areas are traversed for superposition and magnification analysis to generate a surface morphology view; according to the wear prediction development trend chart, the rendered mesh and the surface morphology view are comprehensively analyzed to construct the mold surface wear distribution map.

[0078] Furthermore, the report generation module 13 is used to perform the following method:

[0079] Retrieve the rice cake pressing process parameter database, match the mold wear level according to the mold surface wear distribution map with the rice cake pressing process parameter database, and formulate a process compensation plan based on the matching result; perform real-time monitoring on the rice cake forming mold to generate real-time monitoring data, perform life assessment on the rice cake forming mold based on historical production data and the real-time monitoring data, and generate a mold health assessment report, wherein the mold health assessment report includes a remaining life prediction value of the rice cake forming mold; trigger a maintenance decision instruction based on the remaining life prediction value to perform adaptive repair on the rice cake forming mold.

[0080] Furthermore, the report generation module 13 is used to perform the following method:

[0081] Based on the mold health assessment report, the maintenance decision instruction is triggered to perform distribution analysis of the mold wear level and determine multiple areas to be repaired; repair analysis is performed on the multiple areas to be repaired according to the remaining life prediction value, and a repair priority sequence is set; maintenance analysis is performed based on the repair priority sequence, and a three-dimensional repair path is generated. According to the three-dimensional repair path, the rice cake forming mold is adaptively repaired.

[0082] It should be noted that the order in which the embodiments of the present application are presented is for illustrative purposes only and does not necessarily represent the superiority or inferiority of the embodiments. Furthermore, the foregoing descriptions of specific embodiments of this specification are provided. The processes depicted in the accompanying drawings do not necessarily require the specific order or sequential order shown to achieve the desired results. In certain embodiments, multitasking and parallel processing are also possible or may be advantageous.

[0083] The above description is only a preferred embodiment of the present application and is not intended to limit the present application. Any modifications, equivalent replacements, improvements, etc. made within the spirit and principles of the present application shall be included in the scope of protection of the present application.

[0084] This specification and drawings are merely illustrative of the present application and are intended to cover any and all modifications, variations, combinations, or equivalents within the scope of this application. Obviously, those skilled in the art may make various modifications and variations to this application without departing from the scope of this application. Thus, this application is intended to include such modifications and variations as fall within the scope of this application and its equivalents.

Claims

1. A method for evaluating the wear degree of rice cake forming molds based on machine vision, characterized in that: The method comprises: Use machine vision to capture images of rice cake forming molds from multiple angles and generate mold image datasets. performing multidimensional feature extraction on the mold image dataset to obtain a wear characteristic parameter group, performing a multi-stage wear assessment based on the wear characteristic parameter group, and constructing a mold surface wear distribution map, wherein the mold surface wear distribution map includes a mold wear grade; The rice cake process is matched according to the mold wear level, a mold health assessment report is generated, and a maintenance decision instruction is triggered to adaptively repair the rice cake forming mold.

2. The method for evaluating the degree of wear of a rice cake forming mold based on machine vision according to claim 1, wherein: The rice cake forming mold is captured from multiple angles using machine vision to generate a mold image dataset. The method includes: configuring an annular LED light source array and setting initial lighting parameters, activating the annular LED light source array according to the initial lighting parameters to control a multi-angle industrial camera array to perform surface acquisition on a rice cake forming mold to obtain multiple surface images; Arranging the plurality of surface images according to an acquisition time sequence to determine a plurality of two-dimensional image sequences, and reconstructing three-dimensional point clouds on the plurality of two-dimensional image sequences to generate three-dimensional point cloud data of the mold; The mold three-dimensional point cloud data is integrated to generate the mold image data set.

3. The method for evaluating the degree of wear of rice cake forming molds based on machine vision as claimed in claim 2, wherein: Performing multidimensional feature extraction on the mold image data set to obtain a wear feature parameter group, the method comprising: Performing edge detection based on the mold image data set to obtain geometric deformation features; Performing surface texture analysis on the mold image dataset using a gray level co-occurrence matrix to obtain surface texture features; Calculating mold discreteness based on the mold three-dimensional point cloud data, and obtaining edge sharpness features according to the mold discreteness combined with the edge gradient change rate; The geometric deformation feature, the surface texture feature and the edge sharpness feature are normalized to construct a multi-dimensional feature vector as the wear feature parameter group.

4. The method for evaluating the degree of wear of rice cake forming molds based on machine vision according to claim 3, wherein: Performing a multi-stage wear assessment based on the wear characteristic parameter group and constructing a mold surface wear distribution map includes: Retrieving mold production batch data for wear analysis and constructing a wear threshold curve, wherein the wear threshold curve corresponds to the mold production batch data; Continuously detecting the rice cake forming mold to generate multiple local wear values, and generating a sampling instruction when any value of the multiple local wear values exceeds the wear threshold curve; Perform secondary image sampling on the rice cake forming mold through the sampling instruction to obtain mold deformation data; Based on the mold deformation data and the wear characteristic parameter group, a multi-stage wear evaluation is performed on the rice cake forming mold to construct the mold surface wear distribution map.

5. The method for evaluating the degree of wear of rice cake forming molds based on machine vision according to claim 3, wherein: The process and method for constructing the mold wear level include: Performing spatial registration on the three-dimensional point cloud data of the mold to obtain deformation point cloud data, performing deformation calculation based on the deformation point cloud data, and determining a local deformation deviation value; Constructing a surface curvature change of the rice cake forming mold based on the geometric deformation characteristics, and dividing the wear gradient area according to the surface curvature change; Extracting the contrast attenuation rate of the surface texture feature and the mold discreteness of the edge sharpness feature to construct a wear level quantization matrix; The wear gradient area is screened according to the local deformation deviation value, and the wear level quantization matrix is mapped to the mold three-dimensional point cloud data according to the screening result to obtain the mold wear level, and the mold wear level is added to the mold surface wear distribution map.

6. The method for evaluating the degree of wear of rice cake forming molds based on machine vision according to claim 4, wherein: Based on the mold deformation data and the wear characteristic parameter group, a multi-stage wear evaluation is performed on the rice cake forming mold to construct the mold surface wear distribution map, the method comprising: Based on the mold deformation data and the wear characteristic parameter group, a multi-stage wear evaluation is performed on the rice cake forming mold: S1: Preliminarily locating the wear area of the rice cake forming mold based on the geometric deformation features and the mold deformation data, and determining multiple candidate wear areas; S2: performing sub-pixel level analysis and identification in the plurality of wear candidate areas in combination with the surface texture features to identify a plurality of wear defect data; S3: Verifying the spatial correlation between the plurality of wear candidate regions and the stress concentration region using finite element simulation based on the edge sharpness feature to obtain a plurality of wear deformation data; S4: performing wear prediction on the multiple wear candidate areas according to the multiple wear defect data and the multiple wear deformation data, and generating a wear prediction development trend graph; Wear assessment is performed according to the wear prediction development trend diagram in combination with the mold wear grade to construct the mold surface wear distribution diagram.

7. The method for evaluating the degree of wear of rice cake forming molds based on machine vision according to claim 6, wherein: Wear assessment is performed according to the wear prediction development trend diagram in combination with the mold wear grade to construct a mold surface wear distribution diagram, the method comprising: Converting the mold three-dimensional point cloud data into a triangular facet mesh, and color-coding and rendering the triangular facet mesh according to the mold wear level to generate a rendering mesh; Traversing the plurality of candidate wear areas for superposition and magnification analysis to generate a surface topography view; According to the wear prediction development trend diagram, the rendering grid and the surface topography view are comprehensively analyzed to construct the mold surface wear distribution map.

8. The method for evaluating the degree of wear of rice cake forming molds based on machine vision according to claim 1, wherein: Matching the rice cake process according to the mold wear level, generating a mold health assessment report, triggering a maintenance decision instruction to adaptively repair the rice cake forming mold, the method comprising: Retrieving a rice cake pressing process parameter database, matching the mold wear level according to the mold surface wear distribution map with the rice cake pressing process parameter database, and formulating a process compensation plan based on the matching result; Performing real-time monitoring on a rice cake forming mold to generate real-time monitoring data, performing a life assessment on the rice cake forming mold based on historical production data and the real-time monitoring data, and generating a mold health assessment report, wherein the mold health assessment report includes a predicted value of the remaining life of the rice cake forming mold; A maintenance decision instruction is triggered based on the remaining life prediction value to perform adaptive repair on the rice cake forming mold.

9. The method for evaluating the degree of wear of rice cake forming molds based on machine vision according to claim 8, wherein: Triggering a maintenance decision instruction based on the remaining life prediction value to adaptively repair the rice cake forming mold includes: Triggering the maintenance decision instruction based on the mold health assessment report to perform distribution analysis of the mold wear level and determine multiple areas to be repaired; Performing repair analysis on the plurality of areas to be repaired according to the remaining life prediction values, and setting a repair priority sequence; A maintenance analysis is performed based on the repair priority sequence to generate a three-dimensional repair path, and the rice cake forming mold is adaptively repaired according to the three-dimensional repair path.

10. A rice cake forming mold wear assessment system based on machine vision, characterized in that: A system for implementing the method for evaluating the degree of wear of a rice cake forming mold based on machine vision according to any one of claims 1 to 9, the system comprising: The image acquisition module is used to capture the rice cake forming mold from multiple angles through machine vision to generate a mold image dataset; a wear assessment module configured to perform multidimensional feature extraction on the mold image dataset to obtain a wear feature parameter group, perform multi-stage wear assessment based on the wear feature parameter group, and construct a mold surface wear distribution map, wherein the mold surface wear distribution map includes a mold wear grade; The report generation module is used to match the rice cake process according to the mold wear level, generate a mold health assessment report, and trigger maintenance decision instructions to adaptively repair the rice cake forming mold.

Images