A packaging box modeling and rendering method and system

Abstract

The present invention provides a packaging box modeling and rendering method and system, relating to the technical field of three-dimensional modeling and rendering. The method includes: obtaining 3D model data of a packaging box to be rendered, and constructing a model geometric feature vector of the packaging box to be rendered; processing the model geometric feature vector through a material matching model to generate a plurality of candidate materials and a first matching degree of each candidate material, and querying candidate textures of each candidate material through a material texture list; determining a plurality of first rendering paths according to the candidate textures of each candidate material, and analyzing the plurality of first rendering paths through a texture matching model to generate a second matching degree of each first rendering path; determining a second rendering path from the plurality of first rendering paths according to the second matching degree, and rendering the packaging box to be rendered through the second rendering path. The present invention realizes the improvement of the intelligent level of packaging box rendering.

Description

Technical Field

[0001] The present invention relates to the technical field of three-dimensional modeling and rendering, and particularly to a method and system for modeling and rendering a packaging box. Background Art

[0002] The rendering effect of packaging box design plays an important role in product display and customer communication. Some traditional methods for modeling and rendering packaging boxes rely more on designers to manually adjust and select materials, textures, and other visual effects. Although this method can meet basic requirements, there are certain deficiencies in terms of efficiency and intelligence level. This usually requires designers to make manual operations and subjectively judge the selection of materials and textures based on experience and aesthetics. Although this way can achieve a good rendering effect, the rendering efficiency is low, and it requires designers to have excellent knowledge reserves. It has insufficient adaptability to diverse requirements and complex rendering effects, and the rendering efficiency needs to be improved. Summary of the Invention

[0003] To solve the above technical problems, the present invention proposes a method and system for modeling and rendering a packaging box, which mines the complex relationship between geometric features and materials, textures from historical data, analyzes the geometric features of the packaging box to be rendered, and intelligently selects the optimal combination among multiple candidate materials and textures, so as to realize the optimization of the automated rendering path and improve the intelligence level of packaging box rendering.

[0004] To achieve the above object, the first aspect of the present invention provides a method for modeling and rendering a packaging box, including:

[0005] Obtain the 3D model data of the packaging box to be rendered, extract geometric feature parameters from the 3D model data of the packaging box to be rendered to obtain multiple geometric feature parameters, and construct a model geometric feature vector of the packaging box to be rendered;

[0006] Process the model geometric feature vector through a material matching model to generate multiple candidate materials and the first matching degree of each candidate material, and query the candidate textures of each candidate material through a material texture list, where the material texture list is constructed by analyzing the historical rendering data of the packaging box;

[0007] Determine multiple first rendering paths according to the candidate textures of each candidate material, and analyze the multiple first rendering paths through a texture matching model to generate the second matching degree of each first rendering path;

[0008] Determine a second rendering path from the multiple first rendering paths according to the second matching degree, and render the packaging box to be rendered through the second rendering path;

[0009] Among them, analyzing the multiple first rendering paths through a texture matching model includes:

[0010] After obtaining the historical rendering data of the packaging box, according to multiple candidate materials and each candidate material, a set of geometric texture association data for each first rendering path is extracted from the historical rendering data. The third matching degree under each first rendering path is extracted through multiple sets of geometric texture association data, and the second matching degree of each first rendering path is calculated based on the first matching degree and the third matching degree.

[0011] Preferably, extracting a set of geometric texture association data for each first rendering path from the historical rendering data according to multiple candidate materials and each candidate material, and extracting the third matching degree under each first rendering path through multiple sets of geometric texture association data includes:

[0012] Classify multiple first rendering paths based on the candidate materials to obtain multiple rendering path categories, and extract the reference rendering data corresponding to each rendering path category from the historical rendering data, including extracting sub-sample data under multiple rendering operations from the historical rendering data according to the candidate materials to which each rendering path category belongs, and selecting multiple sets of target sample data from multiple sets of sub-sample data according to multiple candidate textures involved in the rendering path category, and constructing a set of geometric texture association data containing multiple sets of target sample data for each rendering path category, where multiple first rendering paths containing the same candidate material correspond to the same set of geometric texture association data;

[0013] Extract the target geometric feature vectors of each set of target sample data, cluster the multiple sets of target sample data in each set of geometric texture association data based on the target geometric feature vectors to obtain multiple geometric feature clusters, and determine the first association value of each geometric feature cluster with respect to multiple candidate textures;

[0014] Determine the matching list of the packaging box to be rendered with multiple candidate textures in each set of geometric texture association data according to the model geometric feature vectors, and determine the third matching degree under each first rendering path according to the matching list.

[0015] Preferably, determining the matching list of the packaging box to be rendered with multiple candidate textures in each set of geometric texture association data according to the model geometric feature vectors, and determining the third matching degree under each first rendering path according to the matching list includes:

[0016] Determine the reference distances between the model geometric feature vectors and the multiple geometric feature clusters corresponding to each set of geometric texture association data respectively, and correct the first association value of each geometric feature cluster with respect to multiple candidate textures based on the reference distances to obtain the second association value of the packaging box to be rendered with respect to each candidate texture in each geometric feature cluster;

[0017] Perform feature fusion on multiple second correlation values to generate third correlation values of the packaging box to be rendered for each candidate texture in each geometric texture correlation data set, construct a matching list of the packaging box to be rendered with multiple candidate textures in each geometric texture correlation data set, and determine the third correlation value under each candidate texture from the matching list according to the candidate texture corresponding to the first rendering path, and record it as the third matching degree under the first rendering path.

[0018] Preferably, determine the second rendering path from multiple first rendering paths according to the second matching degree, and render the packaging box to be rendered through the second rendering path, including:

[0019] Select the first rendering path with the largest second matching degree and record it as the second rendering path, and render the packaging box to be rendered according to the candidate material and candidate texture in the second rendering path;

[0020] Among them, for the second matching degree, the first matching degree is weighted and corrected by the third matching degree to obtain the second matching degree of each first rendering path.

[0021] Preferably, for the material texture list, it also includes:

[0022] Statistically analyze the historical rendering data of the packaging box to determine multiple material texture combinations, count the frequency parameters of each material texture combination, and construct a geometric structure data set of each material texture combination based on the historical rendering data;

[0023] Determine multiple target geometric feature vectors in each geometric structure data set, and calculate the entropy value of each geometric structure data set based on the target geometric feature vectors;

[0024] Optimize the frequency parameters of each material texture combination according to the entropy value of the geometric structure data set to obtain the matching parameters of each material texture combination, determine multiple target combinations from multiple material texture combinations associated with each material, and construct a material texture list for each material according to the target combinations.

[0025] Preferably, for the material matching model, it also includes:

[0026] Construct a training set based on the historical rendering data of the packaging box, and train the material matching model through the training set;

[0027] Among them, for the construction of the training set, extract multiple groups of geometric structure data from the historical rendering data, and the corresponding rendering material data of each group of geometric structure data;

[0028] Using multiple sets of geometric structure data as the input of the material matching model, and using the rendering material data corresponding to each set of geometric structure data as the training target of the material matching model, the material matching model is obtained through training with a training set, where the material matching model is a multi-layer perceptron model.

[0029] The second aspect of the present invention provides a packaging box modeling and rendering system for implementing the above-mentioned packaging box modeling and rendering method, including:

[0030] A model data preprocessing module, configured to obtain 3D model data of the packaging box to be rendered, extract multiple geometric feature parameters from the 3D model data of the packaging box to be rendered, and construct a model geometric feature vector of the packaging box to be rendered;

[0031] A rendering texture analysis module, configured to process the model geometric feature vector through the material matching model to generate multiple candidate materials and the first matching degree of each candidate material, and query the candidate texture of each candidate material through the material texture list, where the material texture list is constructed by analyzing the historical rendering data of the packaging box;

[0032] A rendering path analysis module, configured to determine multiple first rendering paths according to the candidate textures of each candidate material, and analyze the multiple first rendering paths through the texture matching model to generate the second matching degree of each first rendering path;

[0033] A packaging box modeling and rendering module, configured to determine a second rendering path from the multiple first rendering paths according to the second matching degree, and render the packaging box to be rendered through the second rendering path.

[0034] Preferably, for the rendering path analysis module, analyzing the multiple first rendering paths through the texture matching model includes:

[0035] After obtaining the historical rendering data of the packaging box, according to the multiple candidate materials and each candidate material, extract the geometric texture association data set of each first rendering path from the historical rendering data, extract the third matching degree of each first rendering path through the multiple geometric texture association data sets, and calculate the second matching degree of each first rendering path according to the first matching degree and the third matching degree.

[0036] The present invention has the following beneficial effects:

[0037] By analyzing historical rendering data, the present invention determines the associations between geometric structures, rendering materials, rendering textures, etc. during the rendering process of the packaging box, designs an automated packaging box rendering process, and can automatically select the most suitable material-texture combination according to the geometric structure information of the packaging box to be rendered, reducing the time for designers to repeatedly select and adjust during the rendering process, saving the time and cost of manual operations, providing sufficient flexibility for designers, allowing designers to make personalized improvements on the basis of automated rendering, and enhancing the efficiency and intelligence level of packaging box rendering. Brief Description of the Drawings

[0038] Figure 1 It is a schematic flow chart of a packaging box modeling and rendering method provided by an embodiment of the present invention.

[0039] Figure 2 It is a schematic structural diagram of a packaging box modeling and rendering system provided by an embodiment of the present invention. Detailed Embodiments

[0040] In order to enable those skilled in the art to better understand the technical solutions in the present invention, the technical solutions in the embodiments of the present invention will be clearly and completely described below in conjunction with the accompanying drawings in the embodiments of the present invention. It should be understood that the specific embodiments described herein are only used to explain the present invention and are not used to limit the present invention.

[0041] An embodiment of the present invention provides a packaging box modeling and rendering method. Please refer to Figure 1 , and the method specifically includes the following steps:

[0042] Step S1: Obtain the 3D model data of the packaging box to be rendered, extract geometric features from the 3D model data of the packaging box to be rendered to obtain a plurality of geometric feature parameters, and construct a model geometric feature vector of the packaging box to be rendered.

[0043] In this embodiment, for the 3D model data of the packaging box to be rendered, it can be from various 3D design software such as Rhino, Blender, etc., which contains the basic geometric structure information of the packaging box model with a designed structure. Geometric features are extracted from the 3D model data of the packaging box to obtain a plurality of geometric feature parameters. The geometric feature parameters include but are not limited to features such as the external dimensions, edge angles, surface curvatures, crease positions, concave-convex shapes, cutting lines, etc. of the model. According to these extracted geometric feature parameters, a model geometric feature vector of the packaging box to be rendered is constructed to represent the shape, surface attributes, geometric structure and other features of the packaging box model.

[0044] Step S2: Process the model geometric feature vector through a material matching model to generate a plurality of candidate materials and the first matching degree of each candidate material, and query the candidate textures of each candidate material through a material texture list.

[0045] In this embodiment, for the extracted model geometric feature vectors, they are processed by a pre-trained material matching model to generate multiple candidate materials. Among them, the material matching model is trained based on a training set constructed from historical rendering data and can be used to identify multiple materials that are more suitable for rendering the packaging box under different geometric structures. At the same time, the first matching degree of each candidate material is also output by the model, which is used to indicate the matching degree between the packaging box to be rendered under the current structure and the multiple candidate materials. Then, multiple textures that are more matched with each material are determined according to the pre-analyzed material texture list. It should be noted that the material texture list is constructed by analyzing and sorting the historical rendering data of the packaging box. The historical rendering data contains rendering schemes that match well with packaging boxes of different structures, such as information on the rendered materials, textures, etc. By analyzing these data, multiple texture schemes suitable for each material can be determined as reference data during the matching process.

[0046] Step S3: Determine multiple first rendering paths according to the candidate textures of each candidate material, and analyze the multiple first rendering paths through a texture matching model to generate the second matching degree of each first rendering path.

[0047] In this embodiment, after determining the multiple candidate textures corresponding to each candidate material, each candidate material is combined with its corresponding multiple candidate textures respectively to obtain multiple first rendering paths for the packaging box to be rendered, which are used to indicate multiple optional schemes for rendering the packaging box in terms of materials and textures. On this basis, the multiple first rendering paths are further analyzed through a texture matching model, with the second matching degree of each first rendering path. The second matching degree is used to indicate the overall adaptability of the texture and material of this rendering path under the current packaging box structure.

[0048] Step S4: Determine a second rendering path from the multiple first rendering paths according to the second matching degree, and render the packaging box to be rendered through the second rendering path.

[0049] In this embodiment, after the second matching degree of each first rendering path, a second rendering path is selected from the multiple rendering paths. Specifically, the higher the second matching degree, the higher the degree of adaptation of the scheme to the structure of the packaging box to be rendered. Therefore, the first rendering path with the largest second matching degree is selected and recorded as the second rendering path, and the packaging box to be rendered is rendered according to the candidate material and candidate texture in the second rendering path, realizing intelligent packaging box rendering.

[0050] It should be noted that the traditional rendering of packaging boxes relies heavily on the experience and subjective judgment of designers. Designers need to manually select materials and textures according to different geometric features and design requirements. This process is not only cumbersome but also easily affected by human biases, and the rendering efficiency is relatively low. Through the above automated steps, the present invention completes the rendering process of the packaging box, which can automatically select the most suitable combination of material textures, reducing the time for designers to repeatedly select and adjust during the rendering process, saving the time and cost of manual operations. Moreover, based on the automatic recommendation of the most suitable materials and texture combinations according to historical data and geometric features in the present invention, sufficient flexibility is also provided for designers, allowing designers to make personalized improvements on the basis of automated rendering. For example, designers can further adjust certain details, optimize the visual effects or add specific design elements based on the recommended rendering path and texture combination to achieve the final effect that better meets the creative requirements. This enables designers not only to reduce cumbersome operations but also to achieve personalized artistic expression under a more efficient workflow, improving the rendering efficiency of packaging box design while flexibly meeting the needs of different customers and markets.

[0051] In an optional implementation process, for the material texture list in step S2, it is constructed in the following manner:

[0052] Statistically analyze the historical rendering data of the packaging box to determine multiple material texture combinations, count the frequency parameters of each material texture combination, and construct a geometric structure data set for each material texture combination based on the historical rendering data, that is, extract the historical rendering data related to the materials and textures in the material texture combination together to obtain the geometric structure data set for each material texture combination.

[0053] Then determine multiple target geometric feature vectors in each geometric structure data set. Since the historical rendering data includes the geometric structure data of the rendered packaging box under multiple rendering operations, as well as data such as textures and materials involved in the rendering scheme, by analyzing the geometric structure data of the rendered packaging box under each rendering operation, extract the geometric structure features therein to construct the target geometric feature vectors, and calculate the entropy value of each geometric structure data set based on the target geometric feature vectors. The higher the entropy value, the more dispersed the geometric structures corresponding to the multiple packaging boxes in the geometric structure data set and the higher the degree of dispersion.

[0054] Optimize the frequency parameter of each material texture combination according to the entropy value of the geometric structure dataset. Specifically, during the optimization process, the smaller the entropy value of the geometric structure dataset, the larger the frequency parameter. The weight parameter used to optimize the frequency parameter of each material texture combination can be determined by the entropy value of the geometric structure dataset. Optimize the frequency parameter of each material texture combination through the weight parameter to obtain the matching parameter of each material texture combination. The larger the matching parameter, the more matching the materials and textures involved in the material texture combination. Through a preset matching threshold, multiple target combinations can be determined from multiple material texture combinations associated with each material. Construct the material texture list of each material according to the target combination, which is used to indicate multiple textures that are more suitable under each material.

[0055] In an alternative implementation process, for the material matching model in step S2, it is constructed in the following manner:

[0056] Construct a training set based on the historical rendering data of the packaging box, and train the material matching model through the training set;

[0057] Specifically, extract multiple groups of geometric structure data from the historical rendering data, as well as the rendering material data corresponding to each group of geometric structure data, and associate each group of geometric structure data with the corresponding rendering material data to construct a training set. During the model training process, use multiple groups of geometric structure data as the input of the material matching model, and use the rendering material data corresponding to each group of geometric structure data as the training target of the material matching model. The material matching model learns the association between the geometric structure of the packaging box and different rendering materials through the training set. Finally, after inputting the geometric structure data to be analyzed, it can generate the matching values between the geometric structure data to be analyzed and multiple rendering materials, which is used to indicate the matching degree of different rendering materials under this geometric structure. Among them, the material matching model can specifically be a multi-layer perceptron model. The multi-layer perceptron can be used to handle prediction tasks with multiple target variables and is widely used in various regression tasks. The training process of the multi-layer perceptron is a well-known technical means for those skilled in the art and will not be elaborated here.

[0058] In an alternative implementation process, for step S3, analyze multiple first rendering paths through a texture matching model, specifically including:

[0059] After obtaining the historical rendering data of the packaging box, according to multiple candidate materials and each candidate material, a set of geometric texture association data for each first rendering path is extracted from the historical rendering data. Among them, the geometric texture association data set is used to record the feature data under the combination of different geometric features and textures, and is specifically selected based on each candidate material and the basis, and is used to analyze the texture scheme that is more suitable for different geometric structures of the packaging box under each candidate material. Then, the third matching degree under each first rendering path is extracted through multiple geometric texture association data sets. Finally, the second matching degree of each first rendering path is calculated according to the first matching degree and the third matching degree, so as to reflect the comprehensive adaptation effect of the geometric features of the packaging box to be rendered with multiple candidate materials and texture combinations.

[0060] In an optional implementation process, according to multiple candidate materials and each candidate material, a set of geometric texture association data for each first rendering path is extracted from the historical rendering data, and the third matching degree under each first rendering path is extracted through multiple geometric texture association data sets, which specifically includes:

[0061] Classify multiple first rendering paths based on the candidate materials to obtain multiple rendering path categories, extract the reference rendering data corresponding to each rendering path category from the historical rendering data, and construct a set of geometric texture association data including multiple groups of target sample data for each rendering path category.

[0062] It should be noted that although the geometric structure characteristics of the packaging box cannot directly determine the suitable materials, but other personalized artistic design requirements also need to be considered. However, by analyzing and mining a large number of historical rendering data with good design effects, since it contains the materials and textures with good matching effects of the packaging box under different geometric structures, some statistical laws can be found by analyzing these data, that is, specific geometric features such as smoothness, curved surfaces, creases, etc. usually match better with certain types of materials and textures. For example, surfaces with more folds and edges may be more suitable for paper or cardboard materials; smooth and shiny surfaces may require plastic or metal materials with stronger reflectivity. These laws can be used as a reference basis in the intelligent rendering process, so that the most suitable materials and textures can be recommended during automated rendering. And under different materials, the suitable textures will also be different. For example, for different materials such as wood, metal, and plastic, for smooth materials such as plastic or metal, fine textures, which may be shiny texture, are suitable. For rough materials such as wood or paper packaging, natural and rough textures such as wood grain and paper texture can be selected. And when a material is suitable for multiple textures, the difference in geometric features will affect the specific selection and application effect of the texture under a certain material. Therefore, the present invention takes the material as the core and further studies in more detail the influence of the difference in geometric features on the multiple textures suitable for the same material.

[0063] In this embodiment, multiple first rendering paths are classified with the candidate material as the core to determine the rendering path category corresponding to each candidate material, and then the reference rendering data corresponding to each rendering path category is extracted to construct a geometric texture association data set corresponding to each rendering path category. Specifically, according to the candidate material to which each rendering path category belongs, sub-sample data under multiple rendering operations is extracted from the historical rendering data. The same material is involved in the packaging box rendering operations related to multiple groups of sub-sample data. For multiple groups of sub-sample data under each rendering path category, multiple groups of target sample data are selected from the multiple groups of sub-sample data according to multiple candidate textures involved in the rendering path category to remove some of the noise data. In this way, a geometric texture association data set containing multiple groups of target sample data for each rendering path category is constructed. Each geometric texture association data set contains multiple first rendering paths with the same candidate material.

[0064] The target geometric feature vectors of each group of target sample data are extracted, and based on the target geometric feature vectors, the multiple groups of target sample data in each geometric texture association data set are clustered to obtain multiple geometric feature clusters, and the first association value of each geometric feature cluster with respect to multiple candidate textures is determined.

[0065] In this embodiment, after obtaining the geometric texture association data set corresponding to each rendering path category, the target geometric feature vectors are extracted from each group of target sample data. These geometric feature vectors represent the characteristics of different geometric structures under specific materials and textures. By analyzing the target geometric feature vectors, the differences in texture adaptation under different geometric structures can be identified, and then clustering analysis is performed on the multiple groups of target sample data in each geometric texture association data set. The purpose of clustering is to classify similar geometric features into one category to better analyze the texture adaptation under these geometric features. Through clustering analysis, the texture adaptation pattern of geometric features under the same material can be identified. Each geometric feature cluster represents a specific group of geometric structures, and they show similar regularities in texture adaptation. By analyzing the multiple groups of target sample data in each geometric feature cluster, the statistical features regarding the candidate textures are extracted, such as counting the number of times each candidate texture is used in the geometric feature cluster and performing normalization processing, and the obtained multiple parameters are used as the first association values corresponding to the multiple candidate textures respectively. The first association value reflects the matching degree of different geometric features to different candidate textures among the multiple candidate textures adapted under a specific material.

[0066] Finally, a matching list of the packaging box to be rendered with multiple candidate textures in each geometric texture association data set is determined according to the model geometric feature vector, and a third matching degree under each first rendering path is determined according to the matching list.

[0067] Specifically, the reference distances corresponding to multiple geometric feature clusters in each geometric texture association data set are determined for the model geometric feature vector, and the first association values of each geometric feature cluster with respect to multiple candidate textures are corrected based on the reference distances to obtain the second association values of the packaging box to be rendered with respect to each candidate texture in each geometric feature cluster.

[0068] In this embodiment, the center of each geometric feature cluster is first determined. For example, the mean value of multiple target geometric feature vectors in each geometric feature cluster is calculated as the cluster center, and the reference distances between the model geometric feature vector and multiple cluster centers are calculated respectively based on a distance calculation formula such as the Euclidean distance, and then the first association values of multiple candidate textures corresponding to each geometric feature cluster are corrected. Since the similarity between the geometric feature vector of the packaging box to be rendered and the geometric feature cluster is not a perfect match, the correction method can better measure the matching degree of the packaging box to be rendered with multiple candidate textures under different candidate materials. The correction of the first association value can specifically be to weight the first association value, that is, the corresponding weights are determined according to multiple reference distances. Each geometric feature cluster corresponds to the same weight, and the closer the distance to the geometric feature cluster, the greater the weight. In this way, the first association value of each candidate texture in each cluster can be corrected to obtain the second association value of the packaging box to be rendered with respect to each candidate texture in each geometric feature cluster.

[0069] Finally, feature fusion is performed on multiple second association values. For example, the mean value of multiple second association values of each candidate texture is calculated, and the calculated mean value is used as the third association value of the packaging box to be rendered with respect to each candidate texture in each geometric texture association data set, which is used to characterize the matching degree of the packaging box to be rendered with multiple candidate textures under the same material, so as to effectively evaluate the adaptability of different textures.

[0070] A matching list of the packaging box to be rendered with multiple candidate textures in each geometric texture association data set is constructed through multiple third association values. Then, according to the candidate texture corresponding to the first rendering path, the third association value under each candidate texture is determined from the matching list and recorded as the third matching degree under the first rendering path. In this case, the magnitude of the third matching degree is used to indicate the matching degree between the geometric features of the packaging box to be rendered and different candidate textures under a certain candidate material.

[0071] For the second matching degree of each first rendering path, the first matching degree is weighted and corrected by the third matching degree to obtain the second matching degree of each first rendering path. Specifically, for the first matching degree and the third matching degree of the packaging box to be rendered, the first matching degree is used to indicate the matching degree between the geometric features of the packaging box to be rendered and multiple candidate materials, while the third matching degree is the matching degree between the geometric structure and texture of the packaging box to be rendered under a specific material based on the known candidate materials. The finally calculated second matching degree comprehensively represents the interactive influence among the geometric structure, material, and texture while reducing noise interference, and can provide accurate data reference for selecting an appropriate rendering path.

[0072] The embodiment of the present invention also provides a packaging box modeling and rendering system, which is specifically used to implement the above-mentioned packaging box modeling and rendering method. Please refer to Figure 2 , including:

[0073] The model data preprocessing module is used to obtain the 3D model data of the packaging box to be rendered, extract multiple geometric feature parameters from the 3D model data of the packaging box to be rendered, and construct the model geometric feature vector of the packaging box to be rendered;

[0074] The rendering texture analysis module is used to process the model geometric feature vector through the material matching model to generate multiple candidate materials and the first matching degree of each candidate material, and query the candidate texture of each candidate material through the material texture list, where the material texture list is constructed by analyzing the historical rendering data of the packaging box;

[0075] The rendering path analysis module is used to determine multiple first rendering paths according to the candidate textures of each candidate material, analyze the multiple first rendering paths through the texture matching model, and generate the second matching degree of each first rendering path;

[0076] Analyzing the multiple first rendering paths through the texture matching model includes:

[0077] After obtaining the historical rendering data of the packaging box, according to multiple candidate materials and each candidate material, extract the geometric texture association data set of each first rendering path from the historical rendering data, extract the third matching degree of each first rendering path through the multiple geometric texture association data sets, and calculate the second matching degree of each first rendering path according to the first matching degree and the third matching degree.

[0078] The packaging box modeling and rendering module is used to determine the second rendering path from the multiple first rendering paths according to the second matching degree, and render the packaging box to be rendered through the second rendering path.

[0079] Specifically, select the first rendering path with the second largest matching degree as the second rendering path, and render the packaging box to be rendered according to the candidate materials and candidate textures in the second rendering path;

[0080] Among them, for the second matching degree, the first matching degree is weighted and corrected through the third matching degree to obtain the second matching degree of each first rendering path.

[0081] The above are only specific implementation manners of the present invention, enabling those skilled in the art to understand or implement the present invention. Various modifications to these embodiments will be obvious to those skilled in the art. The parts not described in detail in this specification belong to the prior art well-known to those skilled in the art.

Claims

1. A packaging box modeling and rendering method, characterized in that: include: Acquire 3D model data of the packaging box to be rendered, perform geometric feature extraction on the 3D model data of the packaging box to be rendered to obtain multiple geometric feature parameters, and construct a model geometric feature vector of the packaging box to be rendered; The model geometric feature vector is processed by a material matching model to generate a plurality of candidate materials and a first matching degree of each candidate material, and the candidate texture of each candidate material is queried by a material texture list, wherein the material texture list is constructed by analyzing historical rendering data of the packaging box; Determine a plurality of first rendering paths according to the candidate texture of each candidate material, analyze the plurality of first rendering paths by a texture matching model, and generate a second matching degree of each first rendering path; Determine a second rendering path from the plurality of first rendering paths according to the second matching degree, and render the packaging box to be rendered through the second rendering path; The analyzing the plurality of first rendering paths by using the texture matching model includes: After acquiring historical rendering data of the packaging box, extracting a geometric texture associated data set of each first rendering path from the historical rendering data according to multiple candidate materials and each candidate material, extracting a third matching degree under each first rendering path through the multiple geometric texture associated data sets, and calculating a second matching degree of each first rendering path according to the first matching degree and the third matching degree; Extracting a geometric texture associated data set of each first rendering path from the historical rendering data according to the multiple candidate materials and each candidate material, and extracting a third matching degree under each first rendering path through the multiple geometric texture associated data sets, including: Classifying a plurality of first rendering paths based on candidate materials to obtain a plurality of rendering path categories, extracting reference rendering data corresponding to each rendering path category from the historical rendering data, including extracting sub-sample data under a plurality of rendering operations from the historical rendering data according to the candidate materials to which each rendering path category belongs, selecting a plurality of groups of target sample data from a plurality of groups of sub-sample data according to a plurality of candidate textures involved in the rendering path category, and constructing a geometric texture associated data set containing a plurality of groups of target sample data for each rendering path category, wherein a plurality of first rendering paths containing the same candidate material correspond to the same geometric texture associated data set; Extracting a target geometric feature vector from each group of target sample data, clustering multiple groups of target sample data in each geometric texture association data set based on the target geometric feature vector to obtain multiple geometric feature clusters, and determining a first association value of each geometric feature cluster with respect to multiple candidate textures; Determine a matching list of a packaging box to be rendered and a plurality of candidate textures in each geometric texture associated data set according to the model geometric feature vector, and determine a third matching degree under each first rendering path according to the matching list; For material texture lists, also include: Performing statistical analysis on historical rendering data of the packaging box to determine multiple material texture combinations, counting frequency parameters of each material texture combination, and constructing a geometric structure data set of each material texture combination based on the historical rendering data; Determine a plurality of target geometric feature vectors in each geometric structure data set, and calculate an entropy value of each geometric structure data set based on the target geometric feature vectors; The frequency parameters of each material-texture combination are optimized according to the entropy value of the geometric structure data set to obtain the matching parameters of each material-texture combination, multiple target combinations are determined from the multiple material-texture combinations associated with each material, and a material texture list for each material is constructed according to the target combinations.

2. A packaging box modeling and rendering method according to claim 1, characterized in that: Determining a matching list of a packaging box to be rendered and a plurality of candidate textures in each geometric texture associated data set according to the model geometric feature vector, and determining a third matching degree under each first rendering path according to the matching list, including: Determine the reference distances respectively corresponding to the model geometric feature vector and multiple geometric feature clusters in each geometric texture association data set, and based on the reference distances, correct the first association value of each geometric feature cluster with respect to multiple candidate textures to obtain the second association value of the to-be-rendered packaging box in each geometric feature cluster with respect to each candidate texture; Feature fusion is performed on multiple second association values ​​to generate a third association value for each candidate texture of the packaging box to be rendered in each geometric texture association data set, and a matching list of the packaging box to be rendered and multiple candidate textures in each geometric texture association data set is constructed. According to the candidate texture corresponding to the first rendering path, the third association value under each candidate texture is determined from the matching list and recorded as the third matching degree under the first rendering path.

3. A packaging box modeling and rendering method according to claim 1, characterized in that: Determining a second rendering path from the plurality of first rendering paths according to the second matching degree, and rendering the packaging box to be rendered through the second rendering path, including: Selecting the first rendering path with the largest second matching degree as the second rendering path, and rendering the packaging box to be rendered according to the candidate material and the candidate texture in the second rendering path; Among them, for the second matching degree, the first matching degree is weightedly corrected by the third matching degree to obtain the second matching degree of each first rendering path.

4. A packaging box modeling and rendering method according to claim 1, characterized in that: For material matching models, also include: A training set is constructed based on the historical rendering data of the packaging box, and the material matching model is trained through the training set; Among them, for the construction of the training set, multiple groups of geometric structure data and rendering material data corresponding to each group of geometric structure data are extracted from the historical rendering data; Taking multiple groups of geometric structure data as input of the material matching model, taking the rendering material data corresponding to each group of geometric structure data as the training target of the material matching model, the material matching model is obtained by training with the training set, wherein the material matching model is a multi-layer perceptron model.

5. A packaging box modeling and rendering system, characterized in that: The system is used to implement a packaging box modeling and rendering method according to any one of claims 1 to 4, comprising: A model data preprocessing module is used to obtain 3D model data of the packaging box to be rendered, extract geometric features of the 3D model data of the packaging box to be rendered to obtain multiple geometric feature parameters, and construct a model geometric feature vector of the packaging box to be rendered; A rendering texture analysis module is used to process the model geometric feature vector through a material matching model, generate multiple candidate materials and a first matching degree of each candidate material, and query the candidate texture of each candidate material through a material texture list, wherein the material texture list is constructed by analyzing the historical rendering data of the packaging box; A rendering path analysis module, used to determine a plurality of first rendering paths according to a candidate texture of each candidate material, analyze the plurality of first rendering paths through a texture matching model, and generate a second matching degree for each first rendering path; A packaging box modeling and rendering module, used to determine a second rendering path from a plurality of first rendering paths according to a second matching degree, and render a packaging box to be rendered through the second rendering path; For the rendering path analysis module, analyzing the plurality of first rendering paths by using the texture matching model includes: After obtaining the historical rendering data of the packaging box, a geometric texture associated data set of each first rendering path is extracted from the historical rendering data based on multiple candidate materials and each candidate material, and the third matching degree under each first rendering path is extracted through the multiple geometric texture associated data sets, and the second matching degree of each first rendering path is calculated based on the first matching degree and the third matching degree.

Images