An automobile parameterized texture generation method and system, and a storage medium

Abstract

The application relates to a car parameterized texture generation method and system and a storage medium, which comprises the following steps: acquiring a random noise, inputting the random noise into a pre-trained Gan model for processing to generate a corresponding random texture unit graph; acquiring a gray-scale image, dividing the gray-scale image into multiple square regions, and placing a random texture unit graph in each square region; adjusting the size of the random texture unit graph in each square region according to the average pixel brightness of each square region; and adjusting the distribution law of the random texture unit graphs in the multiple square regions to obtain a car texture image. Through the application, the technical problem of low work efficiency of current car parameterized texture design can be solved.

Description

TECHNICAL FIELD

[0001] The present application relates to the technical field of automobile design, and particularly relates to an automobile parameterized texture generation method and system and a storage medium. BACKGROUND

[0002] Automobile parameterized texture design is an important part of the automobile modeling research and development process. This texture design is reflected in the grille, lamps, horn hole and other parts of the automobile. Usually, the texture distribution has certain regularity, which is specifically manifested as the arrangement of single special elements according to certain rules. At present, automobile modeling designers generally design manually and then combine with digital modeling. A relatively advanced method is to complete the entire design according to the parameterized design concept with the help of CAD tools such as Grasshopper and Rhino. The above design methods have two problems: on the one hand, the texture unit needs to rely on the designer's inspiration, and such inspiration is limited by the designer's artistic level; on the other hand, manual modeling takes a lot of time, and the adjustment and modification of the scheme almost starts from the beginning. If CAD software is used, it also needs to be constantly adjusted and optimized. Such software is generally heavy and complex in function, and the learning cost is high, which is difficult to master, thereby resulting in low working efficiency of the current automobile parameterized texture design. SUMMARY

[0003] The purpose of the present application is to provide an automobile parameterized texture generation method and system to solve the technical problem of low working efficiency of the current automobile parameterized texture design.

[0004] To achieve the above purpose, an embodiment of the present application provides an automobile parameterized texture generation method, comprising the following steps:

[0005] A random noise is obtained, and the random noise is input into a pre-trained Gan model for processing to generate a corresponding random texture unit graph;

[0006] A gray scale image is obtained, the gray scale image is divided into a plurality of square regions, and a random texture unit graph is placed in each square region;

[0007] The size of the random texture unit graph in each square region is adjusted according to the average pixel brightness in each square region;

[0008] The distribution rule of the random texture unit graph in the plurality of square regions is adjusted to obtain an automobile texture image.

[0009] Preferably, the Gan model is trained based on a texture unit graph sample set, and the texture unit graph sample set includes single points, single line segments, point and line segment combinations, or line segment and line segment combinations.

[0010] Preferably, the Gan model is a StyleGan generator.

[0011] Preferably, the adjusting the size of the random texture cell map in each block region according to the average brightness of the pixels in each block region comprises:

[0012] The greater the average brightness of the pixels in a block region, the greater the size of the random texture cell map in the block region.

[0013] Preferably, the adjusting the size of the random texture cell map in each block region according to the average brightness of the pixels in each block region comprises:

[0014] According to the average brightness of the pixels in each block region and the initial width value of the random texture cell map generated by the Gan model, a target width value of the adjusted random texture cell map is calculated.

[0015] According to the average brightness of the pixels in each block region and the initial length value of the random texture cell map generated by the Gan model, a target length value of the adjusted random texture cell map is calculated.

[0016] Embodiments of the present application also propose a car parameterized texture generation system, comprising:

[0017] A random generation module is configured to obtain a random noise, input the random noise into a pre-trained Gan model for processing, and generate a corresponding random texture cell map.

[0018] A cell map filling module is configured to obtain a gray scale image, divide the gray scale image into a plurality of block regions, and place a random texture cell map in each block region.

[0019] A size adjustment module is configured to adjust the size of the random texture cell map in each block region according to the average brightness of the pixels in each block region.

[0020] A distribution adjustment module is configured to adjust the distribution of the random texture cell maps in the plurality of block regions to obtain a car texture image.

[0021] Preferably, the Gan model is trained based on a texture cell map sample set, and the texture cell map sample set includes a single point, a single line segment, a combination of points and line segments, or a combination of line segments and line segments.

[0022] The Gan model is a StyleGan generator.

[0023] Preferably, the greater the average brightness of the pixels in a block region, the greater the size of the random texture cell map in the block region.

[0024] Preferably, the size adjustment module is specifically used for:

[0025] calculating a target width value of the adjusted random texture unit graph according to the pixel brightness mean value in each block region and an initial width value of the random texture unit graph generated by the Gan model;

[0026] calculating a target length value of the adjusted random texture unit graph according to the pixel brightness mean value in each block region and an initial length value of the random texture unit graph generated by the Gan model.

[0027] The embodiment of the present application also provides a computer readable storage medium, which stores a computer program, and the computer program is executed by a processor to realize the steps of the automobile parameterized texture generation method.

[0028] The embodiment of the present application has at least the following beneficial effects:

[0029] The embodiment of the present application can generate a large number of rich random texture units based on a Gan model, fill the generated random texture units into a gray scale image, adjust the size of the random texture units according to the pixel brightness of the gray scale image, adjust the distribution rule of the random texture units, and obtain a corresponding automobile texture image, so that the whole process consumes less time, does not need to be manually designed by a designer, can present diversified texture effects, assists the designer to make a decision faster, inspires inspiration, and improves efficiency.

[0030] Other features and advantages of the embodiment of the present application will be described in the following description. BRIEF DESCRIPTION OF DRAWINGS

[0031] In order to more clearly illustrate the technical solutions in the embodiments of the present application or the prior art, the following will briefly introduce the drawings needed to be used in the embodiment or prior art description. Obviously, the drawings in the following description are only some embodiments of the present application, and for those skilled in the art, other drawings can also be obtained without creative labor on the basis of these drawings.

[0032] Figure 1 It is a flowchart of the automobile parameterized texture generation method in the embodiment of the present application.

[0033] Figure 2 It is a structure diagram of the StyleGan model in the embodiment of the present application.

[0034] Figure 3 It is a texture unit graph diagram in the embodiment of the present application.

[0035] Figure 4 It is a gray scale image diagram in the embodiment of the present application.

[0036] Figure 5 A schematic diagram of a vehicle texture image generated in an embodiment of the present application.

[0037] Figure 6 A schematic diagram of a vehicle parameterized texture generation system in an embodiment of the present application. DETAILED DESCRIPTION

[0038] Various exemplary embodiments, features and aspects of the present disclosure will be described in detail below with reference to the accompanying drawings. In addition, in order to better illustrate the present application, numerous specific details are given in the specific embodiments below. Those skilled in the art will understand that the present application can be implemented without certain specific details. In some examples, means well known to those skilled in the art are not described in detail in order to highlight the main idea of the present application.

[0039] Referring to Figure 1 , an embodiment of the present application proposes a vehicle parameterized texture generation method, comprising the following steps:

[0040] Step S100, a random noise is obtained, and the random noise is input into a pre-trained Gan model for processing to generate a corresponding random texture unit graph;

[0041] Specifically, the Gan model mainly includes DCGAN, bigGAN, proGAN, SAGAN and the like, and in the present embodiment, a StyleGan model is preferably but not limitedly used. The structure of the StyleGan model is as shown in Figure 2 The StyleGan model is a new architecture for unsupervised automatic learning of decoupling and separation of high-level attributes of images, and the StyleGan model has excellent performance under the same batch of data sets;

[0042] Before applying the method of the present embodiment, a StyleGan model needs to be trained first, for example, 2000 texture unit graphs are used as a training sample set. Such texture unit graphs can be single points, single line segments, point and line segment combinations, or line segment and line segment combinations of geometric unit graphs. Correspondingly, the original texture unit graph and the generated random texture unit graph in the application process are also single points, single line segments, point and line segment combinations, or line segment and line segment combinations of geometric unit graphs, as shown in Figure 3 ;

[0043] The trained StyleGan model is used to generate texture unit graphs similar to the training samples. According to the characteristics of the adversarial network, any input random noise can be obtained by the StyleGan model based on the random noise to obtain a new random texture unit graph. Thus, an infinite number of random texture unit graphs can be generated without designers to conceive and provide;

[0044] Step S200, obtaining a gray scale image, dividing the gray scale image into a plurality of square regions, and placing one random texture unit graph in each square region;

[0045] Specifically, in this embodiment, the random texture unit graphs generated by the StyleGan model in step S100 are arranged based on a gray scale image, that is, a plurality of random texture unit graphs are filled in the gray scale image. The gray scale image is as shown in Figure 4 The regions can be divided according to the resolution of the gray scale image and the resolution of the generated random texture unit graph. For example, if the resolution of the gray scale image is 1000*1000 and the resolution of the generated random texture unit graph is 50*50, the gray scale image can be divided into 20*20 squares, that is, 1000 / 50=20;

[0046] Step S300, adjusting the size of the random texture unit graph in each square region according to the average pixel brightness in each square region;

[0047] Specifically, the larger the average pixel brightness in a square region, the larger the size of the random texture unit graph in the square region, and vice versa. The smaller the average pixel brightness in a square region, the smaller the size of the random texture unit graph in the square region;

[0048] Step S400, adjusting the distribution of the random texture unit graphs in the plurality of square regions to obtain a car texture image;

[0049] The above steps S100-S300 have completed the basic texture image. In addition, parameters can be adjusted as needed to arrange different texture unit graphs in the image, such as diamond distribution, Chirplet distribution, etc. The texture unit is rotated, and multiple images are superimposed to set parameters according to certain rules to finally obtain a car texture image. For example, Figure 5 The effect of the uniform arrangement of the texture unit in Figure 3 The effect of the uniform arrangement of the texture unit in Figure 3 The effect of the uniform arrangement of the texture unit in Figure 5 The effect of the uniform arrangement of the texture unit in Figure 5 The effect of the uniform arrangement of the texture unit in

[0050] It should be noted that the embodiment of the present application is different from the manual design and the mode of using CAD tools. Many CAD tools are used for architectural design, and the tools are relatively complex and difficult to master. Many functions are not needed for automobile texture design. Human inspiration is always unstable and easy to dry up. Therefore, the embodiment of the present application gives a solution to the situation, which can realize infinite generation of texture units, and the light and targeted implementation method based on image gray scale.

[0051] Exemplarily, the step S300 can include:

[0052] Step S301, calculating a target width value of the adjusted random texture unit graph according to the pixel brightness mean value in each block region and an initial width value of the random texture unit graph generated by the Gan model;

[0053] Specifically, the target width value of the random texture unit graph in the block region is calculated according to the following formula:

[0054]

[0055] Wherein, represents the pixel brightness sum of the block region divided from the gray scale image, represents the brightness value of the (i, j) pixel in the block, i represents the horizontal coordinate of the pixel point, and j represents the vertical coordinate of the pixel point, represents the width of the block region, that is, the number of pixels of the block region in the width direction, represents the height of the block region, that is, the number of pixels of the block region in the height direction, represents the total number of pixels of the block region, and the pixel value of the gray scale image is 0-255, represents the initial width of the random texture unit graph.

[0056] Step S302, calculating a target length value of the adjusted random texture unit graph according to the pixel brightness mean value in each block region and an initial length value of the random texture unit graph generated by the Gan model;

[0057] Specifically, the target length value of the random texture unit graph in the block region is calculated according to the following formula: the height of the unit graph in the region is obtained, wherein, represents the initial height of the random texture unit graph.

[0058] Another embodiment of the present application also provides a car parameterized texture generation system. Each functional module of the system of the embodiment can be used to execute the corresponding steps of the method described in the above embodiment. For details, refer to Figure 6 The system of the embodiment includes:

[0059] The random generation module 1 is configured to obtain a random noise, input the random noise into a pre-trained Gan model for processing, and generate a corresponding random texture unit graph;

[0060] The unit graph filling module 2 is configured to obtain a gray-scale image, divide the gray-scale image into a plurality of square regions, and place a random texture unit graph in each square region.

[0061] The size adjustment module 3 is configured to adjust the size of the random texture unit graph in each square region according to the average pixel brightness in each square region.

[0062] The distribution adjustment module 4 is configured to adjust the distribution of the random texture unit graphs in the plurality of square regions to obtain an automobile texture image.

[0063] Further, the Gan model is trained based on a texture unit graph sample set, and the texture unit graph sample set includes a single point, a single line segment, a point and a line segment combination, or a line segment and a line segment combination.

[0064] The Gan model is a StyleGan generator.

[0065] Further, the greater the average pixel brightness in a square region, the greater the size of the random texture unit graph in the square region.

[0066] Further, the size adjustment module 3 is specifically configured to:

[0067] calculate a target width value of the adjusted random texture unit graph according to the average pixel brightness in each square region and an initial width value of the random texture unit graph generated by the Gan model;

[0068] calculate a target length value of the adjusted random texture unit graph according to the average pixel brightness in each square region and an initial length value of the random texture unit graph generated by the Gan model.

[0069] The system of the above-described embodiments is only schematic, wherein the units illustrated as separate components can or can not be physically separate, and the components illustrated as units can or can not be physical units, i.e., can be located in one place, or can be distributed on a plurality of network units. According to actual needs, part or all of the modules can be selected to achieve the purpose of the system solution of the embodiments.

[0070] It should be noted that the system of the above-mentioned embodiments corresponds to the method of the above-mentioned embodiments, and therefore the parts not described in detail of the system of the above-mentioned embodiments can be obtained by referring to the content of the method of the above-mentioned embodiments, that is, the specific step content recorded in the method of the above-mentioned embodiments can be understood as the function that the system of the above-mentioned embodiments can achieve, which will not be described here.

[0071] In addition, when the automobile parameterized texture generation system of the above-mentioned embodiments is realized in the form of a software function unit and sold or used as an independent product, it can be stored in a computer-readable storage medium.

[0072] Another embodiment of the present application provides a computer-readable storage medium having a computer program stored thereon, wherein the computer program is executed by a processor to implement the steps of the automobile parameterized texture generation method of the above-mentioned embodiments.

[0073] Specifically, the computer-readable storage medium can include any entity or device, recording medium, U disk, mobile hard disk, magnetic disk, optical disk, computer memory, read-only memory (ROM), random access memory (RAM), electrical carrier signal, telecommunication signal, and software distribution medium, etc. that can carry the computer program instructions.

[0074] The embodiments of the present application have the following advantages:

[0075] (1) The embodiments of the present application use the idea of artificial intelligence deep learning to generate simple geometric texture units infinitely, and the method is advanced, which can quickly realize a variety of texture units.

[0076] (2) The embodiments of the present application use the idea of image gray scale control texture unit arrangement, which avoids the divergence of images generated by deep learning, and can control the overall automobile texture effect according to the gray scale base map provided by the designer.

[0077] (3) The embodiments of the present application are simple, relatively lightweight and targeted compared with CAD, which saves the time of modeling designers and improves work efficiency.

[0078] The above has described the embodiments of the present application, and the above description is exemplary, not exhaustive, and is not limited to the disclosed embodiments. Many modifications and changes are obvious to those skilled in the art without departing from the scope and spirit of the described embodiments. The selection of terms used herein is intended to best explain the principles, practical applications, or technical improvements in the market of the embodiments, or to enable other ordinary skilled persons in the art to understand the embodiments disclosed herein.

Claims

1. A method for generating parametric textures for automobiles, characterized in that, Includes the following steps: A random noise is obtained and input into a pre-trained Gan model for processing to generate a corresponding random texture unit map. A grayscale image is acquired, the grayscale image is divided into multiple square regions, and a random texture unit image is placed in each square region; The size of the random texture unit map in each square area is adjusted according to the average pixel brightness in each square area; if the average pixel brightness in a square area is larger, the size of the random texture unit map in that square area is larger, and vice versa. The distribution pattern of random texture unit maps in the multiple square regions is adjusted to obtain a car texture image.

2. The method according to claim 1, characterized in that, The GA model is trained based on a texture unit graph sample set, which includes a single point, a single line segment, a combination of points and line segments, or a combination of line segments and line segments.

3. The method according to claim 1, characterized in that, The Gan model is StyleGan.

4. The method according to claim 1, characterized in that, The step of adjusting the size of the random texture unit map in each block region based on the average pixel brightness in each block region includes: The target width of the adjusted random texture unit map is calculated based on the average pixel brightness in each block region and the initial width value of the random texture unit map generated by the Gan model. The target length of the adjusted random texture unit map is calculated based on the average pixel brightness in each block region and the initial length of the random texture unit map generated by the Gan model.

5. A parametric texture generation system for automobiles, characterized in that, include: The random graph generation module is used to acquire random noise, input the random noise into a pre-trained GA model for processing, and generate a corresponding random texture unit graph. The unit map filling module is used to acquire a grayscale image, divide the grayscale image into multiple square regions, and place one of the random texture unit maps in each square region. The size adjustment module is used to adjust the size of the random texture unit map in each square area according to the average pixel brightness in each square area. If the average pixel brightness in a square area is larger, the size of the random texture unit map in that square area is larger; conversely, if the average pixel brightness in a square area is smaller, the size of the random texture unit map in that square area is smaller. The distribution adjustment module is used to adjust the distribution pattern of random texture unit maps in the multiple square regions to obtain a car texture image.

6. The system according to claim 5, characterized in that, The GA model is trained based on a texture unit graph sample set, which includes a single point, a single line segment, a combination of points and line segments, or a combination of line segments and line segments. The Gan model is StyleGan.

7. The system according to claim 5, characterized in that, The size adjustment module is specifically used for: The target width of the adjusted random texture unit map is calculated based on the average pixel brightness in each block region and the initial width value of the random texture unit map generated by the Gan model. The target length of the adjusted random texture unit map is calculated based on the average pixel brightness in each block region and the initial length of the random texture unit map generated by the Gan model.

8. A computer-readable storage medium having a computer program stored thereon, which, when executed by a processor, implements the steps of the automotive parametric texture generation method according to any one of claims 1-4.

Images