Three-dimensional teaching resource integration system for environmental art design

Abstract

The invention discloses a three-dimensional teaching resource integration system for environmental art design, and the system comprises a data receiving module which collects line features, color features and material features in an environmental art design picture; the style disassembling module is disassembled into a straight line gene unit, a curve gene unit, a dominant hue gene unit, a saturation gene unit and a luminosity gene unit; the style matching module is used for calculating the style identification entropy value of each gene unit and matching a preset style matrix; the learning path recommendation module is used for generating a curriculum schedule according to the matched preset style matrix; and the visualization module is used for visually displaying the preset style label corresponding to the matched preset style matrix and the curriculum schedule. Therefore, the accuracy of environmental art design style recognition can be improved, the curriculum schedule is recommended based on the style matching result, so that the learning path better conforms to the learning requirements of students for specific design styles, and the students can conveniently and visually obtain information through the visual association style label and the curriculum schedule.

Description

Technical Field

[0001] This invention relates to the field of information technology, specifically to a three-dimensional teaching resource integration system for environmental art design. Background Technology

[0002] Currently, in the teaching of environmental art design, teachers often guide students to analyze stylistic characteristics by uploading classic or cutting-edge design images. Students also need to rely on image examples to understand the relationship between line usage, color matching and style expression, and then carry out learning in conjunction with theoretical courses and practical training projects. Therefore, three-dimensional platforms that integrate multi-dimensional teaching resources are gradually becoming popular. Although such platforms can realize basic functions such as design image storage and course resource display, they still have some shortcomings in supporting the core teaching process of case style analysis, targeted course recommendations, or intuitive association of learning information.

[0003] After teachers upload design images, the platform's processing of visual features often remains at a basic extraction level, only able to identify macro categories such as lines and colors, but unable to further break down the details crucial for style judgment. This makes it difficult for the platform to accurately pinpoint the design style corresponding to the image. Furthermore, the ambiguity in style recognition means that subsequent course recommendations can only be based on general course categories, failing to match targeted content with the specific styles that students are interested in. Students find it difficult to quickly find course resources directly related to their target style. At the same time, the platform often presents the style information of design images separately from the course schedule. When analyzing a style case, students need to conduct additional searches to find the corresponding learning courses, making it impossible to intuitively establish the connection between style characteristics and learning content, which to some extent affects learning efficiency and the practicality of resource integration. Summary of the Invention

[0004] The present invention aims to at least partially solve the technical problems in the above-mentioned technologies.

[0005] Therefore, this invention discloses a three-dimensional teaching resource integration system for environmental art design, comprising:

[0006] The data receiving module receives uploaded environmental art design images and collects the line features, color features, and material features from the environmental art design images.

[0007] The style decomposition module decomposes the line features into straight line gene units and curve gene units, and the color features into main color gene units, saturation gene units and brightness gene units.

[0008] The style matching module calculates the style identification entropy value for each gene unit using the Shannon entropy model, and processes each style identification entropy value according to the following steps:

[0009] A random matrix and random fixed values ​​are generated, and the random fixed values ​​are added to each element of the random matrix to obtain an correlation matrix;

[0010] The style recognition entropy value is multiplied by each element in the correlation matrix to obtain the convolution kernel;

[0011] Each style matrix in the preset style matrix set is divided element-wise with the correlation matrix to obtain the transformation matrix;

[0012] Each of the transformation matrices is convolved using the convolution kernel, and a single-value index is obtained by solving the global L2 norm after convolution.

[0013] Based on the largest single-value index, the corresponding style recognition entropy value is matched with the preset style matrix in the preset style matrix set;

[0014] The learning path recommendation module, based on the matched preset style matrix, traverses the preset course tags and generates a course schedule;

[0015] The visualization module will visually display the preset style tags corresponding to the matched preset style matrix and the course schedule.

[0016] The three-dimensional teaching resource integration system for environmental art design disclosed in this invention can improve the accuracy of environmental art design style identification, recommend course schedules based on style matching results to make the learning path more in line with students' learning needs for specific design styles, and visualize the association between style tags and course schedules to facilitate students' intuitive access to information.

[0017] In addition, the three-dimensional teaching resource integration system for environmental art design disclosed in this invention may also have the following additional technical features:

[0018] In one embodiment of the present invention, the value of each element in the random matrix satisfies a truncated normal distribution (0.5, 0.1). 2 The value of the random fixed value is one order of magnitude smaller than the smallest element value in the random matrix.

[0019] In one embodiment of the present invention, in the data receiving module, line features in the environmental art design image are extracted based on the Canny algorithm, and color features in the environmental art design image are extracted based on the K-Means algorithm.

[0020] In one embodiment of the present invention, the line features are processed in the style decomposition module according to the following steps:

[0021] Calculate the slope between the start and end pixels of the line to obtain the baseline slope;

[0022] Calculate the local slope between the starting pixel of the line and the ending pixels of the 1 / 2, 1 / 4, and 1 / 8 segments respectively;

[0023] Calculate the local slope between the endpoint pixel of the line and the starting pixel of the 1 / 2 segment, 1 / 4 segment, and 1 / 8 segment, respectively;

[0024] If the difference between any local slope and the baseline slope is less than ±2%, the line feature corresponding to that line is decomposed into a straight line gene unit; otherwise, it is decomposed into a curve gene unit.

[0025] In one embodiment of the present invention, the color features are processed according to the following steps in the style decomposition module:

[0026] Based on each obtained color cluster, calculate the color value of the center pixel of all pixels in each color cluster, extract the hue parameter corresponding to the color value of the center pixel, match the hue parameter of each color cluster with the preset hue interval library, each successfully matched hue interval corresponds to a main color category, and use the main color category and the pixel proportion of the corresponding color cluster as the core data of the main color gene unit to decompose and obtain the main color gene unit.

[0027] Traverse all pixels within a cluster, extract the saturation parameter of each pixel, calculate the mean and standard deviation of the saturation parameter within each color cluster, use the mean to reflect the overall saturation level of the color cluster, and use the standard deviation to reflect the degree of fluctuation of saturation within the cluster. Use the mean, standard deviation and corresponding color cluster identifier as the core data of the saturation gene unit to decompose and obtain the saturation gene unit.

[0028] The brightness parameters of all pixels in each color cluster are extracted, and the maximum, minimum and mean values ​​of the brightness parameters in each color cluster are determined. The brightness distribution range is calculated by using the maximum and minimum values. The mean value reflects the overall brightness trend of the color cluster, and the brightness distribution range reflects the range of brightness difference. The mean value, brightness distribution range and corresponding color cluster identifier are used as the core data of the brightness gene unit to decompose and obtain the brightness gene unit.

[0029] In one embodiment of the present invention, in the style matching module, when the Shannon entropy model calculates the style recognition entropy value of each gene unit, the feature distribution frequency of the gene unit in the environmental art design image is used as the calculation input. The feature distribution frequency is the proportion of the number of straight line gene units and curve gene units in the overall lines of the image, the pixel proportion of the color clusters corresponding to each main color category in the image, the value distribution frequency of the saturation parameter of each color cluster, and the value distribution frequency of the brightness parameter of each color cluster.

[0030] In one embodiment of the present invention, in the learning path recommendation module, when traversing the preset course tags, firstly, based on the matching preset style matrix, a correlation evaluation index between the preset style matrix and each preset course tag is established, and then the preset course tags are filtered and sorted in order from high to low according to the correlation evaluation index, and a course schedule containing course priorities is generated based on the filtered and sorted preset course tags.

[0031] In one embodiment of the present invention, the correlation evaluation index is the convolution value of a preset style matrix using a weighted fusion of line features, color features, and material features as the convolution kernel.

[0032] Additional features and advantages of this invention will be set forth in the description which follows, or may be learned by practicing the invention. Attached Figure Description

[0033] The technical solution and beneficial effects of the present invention will become apparent and readily understood from the following description in conjunction with the accompanying drawings, wherein:

[0034] Figure 1 This is a schematic diagram of the integrated teaching resource system for environmental art design according to the present invention.

[0035] Figure 2 This is another schematic diagram of the three-dimensional teaching resource integration system for environmental art design according to the present invention;

[0036] Figure 3 This is another schematic diagram of the three-dimensional teaching resource integration system for environmental art design according to the present invention;

[0037] Figure 4 This is another schematic diagram of the three-dimensional teaching resource integration system for environmental art design according to the present invention;

[0038] Figure 5 This is another schematic diagram of the three-dimensional teaching resource integration system for environmental art design according to the present invention;

[0039] Figure 6 This is another schematic diagram of the three-dimensional teaching resource integration system for environmental art design according to the present invention. Detailed Implementation

[0040] The technical solutions of the present invention will be clearly and completely described below with reference to the accompanying drawings of the embodiments of the present invention. Obviously, the described embodiments are only some embodiments of the present invention, and not all embodiments.

[0041] The integrated teaching resource system for environmental art design disclosed in this invention will now be described with reference to the accompanying drawings.

[0042] A comprehensive teaching resource integration system for environmental art design, including:

[0043] The data receiving module receives uploaded environmental art design images and collects line features, color features, and material features from the environmental art design images;

[0044] The style decomposition module breaks down line features into straight line gene units and curve gene units, and color features into main color gene units, saturation gene units, and brightness gene units.

[0045] The style matching module calculates the style identification entropy value for each gene unit using the Shannon entropy model, and processes each style identification entropy value according to the following steps:

[0046] like Figure 1 As shown, a random matrix is ​​generated for the upper part of the figure and a random fixed value is generated for the middle part of the figure. The random fixed value is then added to each element of the random matrix to obtain the correlation matrix for the lower part of the figure.

[0047] It should be noted that in this embodiment, when processing the style recognition entropy value of each gene unit, a different random matrix and a random fixed value of 0.001 are generated each time, and the correlation matrix is ​​obtained after the operation.

[0048] By using randomly generated random matrices and random fixed values, the computational bias of fixed matrices can be avoided, and as a correlation medium, it can effectively control the stability of subsequent calculation results.

[0049] like Figure 2 As shown, the style recognition entropy value of the middle part of the figure is multiplied by each element of the correlation matrix of the upper part of the figure to obtain the convolution kernel of the lower part of the figure.

[0050] It should be noted that in this embodiment, injecting the style recognition entropy value of 0.33 into the convolution kernel can ensure that the correlation of the correlation matrix is ​​not lost.

[0051] It should also be noted that in this embodiment, the random fixed value breaks the linear correlation between the correlation matrix and the convolution matrix, which can avoid convolution degradation caused by subsequent convolution calculations. In the subsequent convolution operation, the correlation features between the convolved matrix and the random matrix can be preserved, so as to realize the style recognition entropy value through the correlation between the random matrix and the convolved matrix.

[0052] like Figure 3 As shown, each style matrix in the preset style matrix set in the upper left part of the figure is divided element-wise with the correlation matrix in the upper right part of the figure to obtain the transformation matrix in the lower part of the figure.

[0053] It should be noted that in this embodiment, the preset style matrix set pre-sets four style matrices, such as... Figure 4 As shown, the style matrix of modernism is in the upper left part of the figure, the style matrix of traditional classical style is in the upper right part of the figure, the style matrix of natural ecological style is in the lower left part of the figure, and the style matrix of modern industrial style is in the lower right part of the figure.

[0054] It should also be noted that, in this embodiment, this part can be used to reverse-weight each style matrix in the preset style matrix set, which can make the originally inconspicuous local differences of the style matrix more prominent, making subsequent convolution easier to distinguish, and avoiding the possibility of being masked by other elements in the convolution, resulting in insensitive matching degree judgment.

[0055] Each transformation matrix is ​​convolved using a convolution kernel;

[0056] like Figure 5 As shown, a single-valued index is obtained by solving the global L2 norm after convolution;

[0057] It should be noted that, in this embodiment, the global L2 norm of the style matrix for the modernist style is 4.6626, the global L2 norm of the style matrix for the traditional classical style is 5.1969, the global L2 norm of the style matrix for the natural ecological style is 4.9404, and the global L2 norm of the style matrix for the modern industrial style is 4.4556.

[0058] Based on the largest single-value indicator, the corresponding style recognition entropy value is matched with the preset style matrix in the preset style matrix set;

[0059] It should be noted that, in this embodiment, the global L2 norm of the style matrix of the traditional classical style is obviously solved to obtain a single-value index of 5.1969, which is the largest value. Therefore, the corresponding style recognition entropy value matches the preset style matrix of the traditional classical style in the preset style matrix set.

[0060] The learning path recommendation module generates a course schedule by traversing preset course tags based on the matched preset style matrix.

[0061] The visualization module visually displays the preset style tags and course schedules corresponding to the matched preset style matrix;

[0062] It should be noted that, in this embodiment, WebGL-driven 3D scene rendering is used to transform preset style tags into iconic symbols in three-dimensional space and form a spatial binding with the timeline of the course schedule. Course nodes are distributed along the timeline, and style symbols are associated with relevant course nodes through dynamic connections. The thickness of the lines corresponds to the matching degree between the style and the course.

[0063] Each element in the random matrix takes values ​​that follow a truncated normal distribution (0.5, 0.1). 2 The value of the random fixed value is one order of magnitude smaller than the smallest element value in the random matrix;

[0064] It should be noted that, in the embodiments, the range of random matrix elements is strictly limited to [0.24, 0.76]. Values ​​below 0.24 will result in excessively low style feature weights, making it impossible to effectively distinguish differences, while values ​​above 0.76 are prone to overfitting, misjudging secondary features as core features.

[0065] In the data receiving module, line features in environmental art design images are extracted based on the Canny algorithm, and color features are extracted based on the K-Means algorithm.

[0066] like Figure 6 As shown, in the style decomposition module, line features are processed according to the following steps:

[0067] Calculate the slope between the start and end pixels of the line to obtain the baseline slope;

[0068] Calculate the local slope between the starting pixel of the line and the ending pixels of the 1 / 2, 1 / 4, and 1 / 8 segments respectively;

[0069] Calculate the local slope between the endpoint pixel of the line and the starting pixel of the 1 / 2 segment, 1 / 4 segment, and 1 / 8 segment, respectively;

[0070] If the difference between any local slope and the baseline slope is less than ±2%, the line feature corresponding to that line is decomposed into a straight line gene unit; otherwise, it is decomposed into a curve gene unit.

[0071] It should be noted that, in the embodiments, according to Figure 6 If the local slope between the starting pixel of the line and the ending pixels of the 1 / 4 segment and the 1 / 8 segment, and the local slope between the ending pixel of the line and the starting pixels of the 1 / 4 segment and the 1 / 8 segment differ from the baseline slope by more than ±2%, then the line feature corresponding to the line is decomposed into a curve gene unit.

[0072] In the style decomposition module, color features are processed according to the following steps:

[0073] Based on each obtained color cluster, calculate the color value of the center pixel of all pixels in each color cluster, extract the hue parameter corresponding to the color value of the center pixel, match the hue parameter of each color cluster with the preset hue interval library, each successfully matched hue interval corresponds to a main color category, and use the main color category and the pixel proportion of the corresponding color cluster as the core data of the main color gene unit to decompose and obtain the main color gene unit.

[0074] Traverse all pixels within a cluster, extract the saturation parameter of each pixel, calculate the mean and standard deviation of the saturation parameter within each color cluster, use the mean to reflect the overall saturation level of the color cluster, and use the standard deviation to reflect the degree of fluctuation of saturation within the cluster. Use the mean, standard deviation and corresponding color cluster identifier as the core data of the saturation gene unit to decompose and obtain the saturation gene unit.

[0075] The brightness parameters of all pixels in each color cluster are extracted, and the maximum, minimum and mean values ​​of the brightness parameters in each color cluster are determined. The brightness distribution range is calculated by using the maximum and minimum values. The mean value reflects the overall brightness trend of the color cluster, and the brightness distribution range reflects the range of brightness difference. The mean value, brightness distribution range and corresponding color cluster identifier are used as the core data of the brightness gene unit to decompose and obtain the brightness gene unit.

[0076] In the style matching module, when the Shannon entropy model calculates the style recognition entropy value of each gene unit, it uses the feature distribution frequency of the gene unit in the environmental art design image as the calculation input. The feature distribution frequency includes the proportion of straight line gene units and curve gene units in the overall lines of the image, the pixel proportion of color clusters corresponding to each main color category in the image, the value distribution frequency of the saturation parameter of each color cluster, and the value distribution frequency of the brightness parameter of each color cluster.

[0077] In the learning path recommendation module, when iterating through the preset course tags, the system first establishes a correlation evaluation index between the preset style matrix and each preset course tag based on the matching preset style matrix. Then, the preset course tags are filtered and sorted in descending order of correlation evaluation index. Based on the filtered and sorted preset course tags, a course schedule containing course priorities is generated.

[0078] The correlation evaluation index is the convolution value of the preset style matrix by using the weighted fusion of line features, color features and material features as the convolution kernel.

[0079] In summary, the three-dimensional teaching resource integration system for environmental art design disclosed in this invention can improve the accuracy of environmental art design style identification, recommend course schedules based on style matching results to make learning paths more suitable for students' learning needs for specific design styles, and visualize the association between style tags and course schedules to facilitate students' intuitive access to information.

[0080] Although embodiments of the present invention have been shown and described above, it is understood that the above embodiments are exemplary and should not be construed as limiting the present invention. Those skilled in the art can make changes, modifications, substitutions and variations to the above embodiments within the scope of the present invention.

Claims

1. A three-dimensional teaching resource integration system for environmental art design, characterized in that, include: The data receiving module receives uploaded environmental art design images and collects the line features, color features, and material features from the environmental art design images. The style decomposition module decomposes the line features into straight line gene units and curve gene units, and the color features into main color gene units, saturation gene units and brightness gene units. The style matching module calculates the style identification entropy value for each gene unit using the Shannon entropy model, and processes each style identification entropy value according to the following steps: A random matrix and random fixed values ​​are generated, and the random fixed values ​​are added to each element of the random matrix to obtain an correlation matrix; The style recognition entropy value is multiplied by each element in the correlation matrix to obtain the convolution kernel; Each style matrix in the preset style matrix set is divided element-wise with the correlation matrix to obtain the transformation matrix; Each of the transformation matrices is convolved using the convolution kernel, and a single-value index is obtained by solving the global L2 norm after convolution. Based on the largest single-value index, the corresponding style recognition entropy value is matched with the preset style matrix in the preset style matrix set; The learning path recommendation module, based on the matched preset style matrix, traverses the preset course tags and generates a course schedule; The visualization module will visually display the preset style tags corresponding to the matched preset style matrix and the course schedule.

2. The integrated teaching resource system for environmental art design as described in claim 1, characterized in that, Each element in the random matrix takes values ​​that follow a truncated normal distribution (0.5, 0.1). 2 The value of the random fixed value is one order of magnitude smaller than the smallest element value in the random matrix.

3. The integrated teaching resource system for environmental art design as described in claim 1, characterized in that, In the data receiving module, line features in environmental art design images are extracted based on the Canny algorithm, and color features are extracted based on the K-Means algorithm.

4. The integrated teaching resource system for environmental art design as described in claim 3, characterized in that, In the style decomposition module, line features are processed according to the following steps: Calculate the slope between the start and end pixels of the line to obtain the baseline slope; Calculate the local slope between the starting pixel of the line and the ending pixels of the 1 / 2, 1 / 4, and 1 / 8 segments respectively; Calculate the local slope between the endpoint pixel of the line and the starting pixel of the 1 / 2 segment, 1 / 4 segment, and 1 / 8 segment, respectively; If the difference between any local slope and the baseline slope is less than ±2%, the line feature corresponding to that line is decomposed into a straight line gene unit; otherwise, it is decomposed into a curve gene unit.

5. The integrated teaching resource system for environmental art design as described in claim 3, characterized in that, In the style decomposition module, color features are processed according to the following steps: Based on each obtained color cluster, calculate the color value of the center pixel of all pixels in each color cluster, extract the hue parameter corresponding to the color value of the center pixel, match the hue parameter of each color cluster with the preset hue interval library, each successfully matched hue interval corresponds to a main color category, and use the main color category and the pixel proportion of the corresponding color cluster as the core data of the main color gene unit to decompose and obtain the main color gene unit. Traverse all pixels within a cluster, extract the saturation parameter of each pixel, calculate the mean and standard deviation of the saturation parameter within each color cluster, use the mean to reflect the overall saturation level of the color cluster, and use the standard deviation to reflect the degree of fluctuation of saturation within the cluster. Use the mean, standard deviation and corresponding color cluster identifier as the core data of the saturation gene unit to decompose and obtain the saturation gene unit. The brightness parameters of all pixels in each color cluster are extracted, and the maximum, minimum and mean values ​​of the brightness parameters in each color cluster are determined. The brightness distribution range is calculated by using the maximum and minimum values. The mean value reflects the overall brightness trend of the color cluster, and the brightness distribution range reflects the range of brightness difference. The mean value, brightness distribution range and corresponding color cluster identifier are used as the core data of the brightness gene unit to decompose and obtain the brightness gene unit.

6. The integrated teaching resource system for environmental art design as described in claim 1, characterized in that, In the style matching module, when the Shannon entropy model calculates the style recognition entropy value of each gene unit, it uses the feature distribution frequency of the gene unit in the environmental art design image as the calculation input. The feature distribution frequency includes the proportion of straight line gene units and curve gene units in the overall lines of the image, the pixel proportion of color clusters corresponding to each main color category in the image, the value distribution frequency of the saturation parameter of each color cluster, and the value distribution frequency of the brightness parameter of each color cluster.

7. The integrated teaching resource system for environmental art design as described in claim 1, characterized in that, In the learning path recommendation module, when iterating through the preset course tags, the system first establishes a correlation evaluation index between the preset style matrix and each preset course tag based on the matching preset style matrix. Then, the preset course tags are filtered and sorted in descending order of correlation evaluation index. Based on the filtered and sorted preset course tags, a course schedule containing course priorities is generated.

8. The integrated teaching resource system for environmental art design as described in claim 7, characterized in that, The correlation evaluation index is the convolution value of the preset style matrix by using the weighted fusion of line features, color features and material features as the convolution kernel.

Images