Textile texture and color synthesis method based on instance image
Through the textile texture and color synthesis method based on example images, using texture normalization, adaptive color conversion and high-frequency removal technologies, the problem that the existing technology cannot achieve texture and color synthesis is solved, and fast and interactive fabric appearance simulation and Design effect.
Patent Information
- Application Number
- CN202411183156.8
- Authority / Receiving Office
- CN · China
- Patent Type
- Applications(China)
- Current Assignee / Owner
- Filing Date
- 2024-08-27
- Publication Date
- 2024-12-06
AI Technical Summary
The prior art is difficult to use the relationship between texture and color appearance to achieve the synthesis of color patterns and fabric textures, and cannot provide intuitive simulation of the appearance effect of fabrics.
Using the textile texture and color synthesis method based on example images, through texture normalization, adaptive color conversion and high-frequency removal steps, YIQ color space and Gaussian filtering technology are used to realize texture and color synthesis, allowing users to control the fluctuations of texture appearance, Thickness and roughness to reduce unnatural appearance.
It realizes fast, interactive and intuitive texture and color synthesis, can reproduce real and natural output images, and supports real-time simulation and design of images.
Smart Images

Figure CN119090736A_ABST
Abstract
Description
Technical Field
[0001] The invention belongs to the field of textile appearance texture simulation, and in particular relates to a textile texture and color synthesis method based on an example image. Background Art
[0002] In recent years, digital manufacturing technology has developed rapidly. Fabric texture simulation has a very positive role in promoting the development of the textile industry, and texture synthesis technology is a very important component of the digital design system of textiles. The appearance and perception simulation of fabric texture has been studied in the fields of computer vision and graphics. For example, the use of Fourier transform to analyze the appearance of plain fabrics, and the method of obtaining and reconstructing the appearance of materials. In addition, some scholars also focus on studying the relationship between texture appearance and image features. For example, by analyzing the influence of color on texture perception, a simulation model of fabric texture perception is constructed; based on evaluation experiments, a method of controlling the touch of materials by manipulating image intensity; by analyzing the relationship between fabric appearance and image features, a color appearance and perception model is established. As mentioned above, many researchers focus on color and texture appearance. In addition, some studies focus on the perceptual dependence between color, texture, material, and viewing conditions. In these traditional studies, the relationship between color, lighting, and surface texture properties (such as roughness and gloss) is the main focus. Then, it is found that both lighting conditions and observer knowledge have an impact on color and texture perception. Therefore, traditional research mainly focuses on the interpretation of the surface appearance or material perception of three-dimensional objects.
[0003] Some effective algorithms and tools for color and texture manipulation and design have been developed. In texture synthesis tools, users are required to input texture instance images to achieve texture transmission and design. In color processing tools, color transfer, coloring, and recoloring have been proposed for color processing and design. However, there is currently no method that utilizes the relationship between texture and color appearance to achieve the synthesis of color patterns and fabric textures and provide an intuitive simulation of fabric appearance effects. Summary of the invention
[0004] The purpose of the present invention is to provide a method for synthesizing textile texture and color based on example images, which can synthesize the pattern and texture of the example image in real time, and is based on the visual appearance of the image, rather than the physical characteristics of the fabric, so that the synthesis effect can be presented quickly, interactively and intuitively. The method allows flexible user control to achieve real-time simulation of fabric appearance based on fabric example images and color images.
[0005] In order to solve the above technical problems, the following technical solutions are adopted:
[0006] A method for synthesizing textile texture and color based on example images, characterized by comprising the following steps:
[0007] (1) Texture normalization: First, the fabric image I f , transform to YIQ color space, and extract fabric image I f The Y component of the image is used to characterize the undulation, thickness and roughness of the fabric texture appearance. The specific frequency band of the fabric image is related to the specific fabric texture appearance. The abnormal band information is eliminated according to the band interval of the specific frequency band to realize the normalization of the fabric texture.
[0008] (2) Adaptive color conversion: After texture normalization, the color information is converted from the color image I c Transfer to fabric image I f The YIQ value of the output color conversion image is determined by the color image I c The IQ component and Y′ of the color image are composed of f The Y component deviation is added to the color image I c In the Y′ component of
[0009] (3) High-frequency removal: After color transfer, in order to reduce the unnaturalness in the color transfer image, Gaussian filtering is applied to the input color image to suppress texture and color. After Gaussian filtering, the color information is transferred to the fabric texture image in the YIQ color space to achieve the synthesis of texture and color, and obtain the synthesized image I N .
[0010] After optimization, in the step (1), the undulation of the fabric corresponds to a low-frequency band, and the low-frequency band corresponds to 0.57-2.29 cycles / degree.
[0011] Preferably, in step (1), the thickness of the fabric corresponds to 2.29-4.28 cycles / degree.
[0012] After optimization, in the step (1), the roughness of the fabric corresponds to a medium frequency band, and the medium frequency band corresponds to 6.57-15.14 cycles / degree.
[0013] After optimization, the step (2) converts the fabric image I f The Y component deviation is added to the color image I c The fabric image I is defined by subtracting the average value of the Y component from the Y component of each pixel. f The deviation of the Y component in f The AC component of the input color image I in frequency space c All components of are added together; the average Y component of the output color conversion image is the same as the Y′ component of the original color image; then the Y component of the output color conversion image is denoted as Y″:
[0014]
[0015] Among them, (x, y) represents the image coordinates, Y f is the Y component of the input fabric image, is the average value of the Y component of the input fabric image, Y c is the Y′ component of the input color image.
[0016] After optimization, in step (3), the image is processed by Gaussian filtering in the RGB color space, and a convolution operation is performed on the image using a Gaussian kernel generated by a Gaussian function, that is:
[0017] I σ =I=G σ
[0018]
[0019] Where I is the fabric texture image, * represents the convolution operation, G σ represents a two-dimensional Gaussian kernel with a standard deviation of σ. In this method, the interval of σ is [0.5, 5.0]. Here, the user can select different σ values in this interval to achieve real-time control of the fabric improvement effect. After Gaussian filtering, the color information is passed to the fabric texture image in the YIQ color space to achieve the synthesis of texture and color.
[0020] The above technical solution has the following beneficial effects:
[0021] The present invention can effectively synthesize the texture and color of an image. First, the texture pattern of the input fabric image is modulated in the frequency domain. The user can control the fluctuation, thickness and roughness of the texture appearance. Furthermore, by synthesizing two different input images, a real and natural output image can be reproduced. The color pattern of the input color image is transferred to the modulated texture pattern in the YIQ color space. The deviation of the Y component of the input fabric image is added to the Y' component of the filtered color image to synthesize the Y component of the color transfer image. Thereby, real-time synthesis of fabric finishing is realized. Finally, a Gaussian filter is used to remove the high-frequency components in the original color image, effectively reducing the unnatural appearance of the output color transfer image. The method can quickly realize the synthesis of fabric texture images and color patterns, so as to support image interactive simulation and design of fabric texture and color. BRIEF DESCRIPTION OF THE DRAWINGS
[0022] The present invention will be further described below in conjunction with the accompanying drawings:
[0023] Figure 1 This is a schematic diagram of the synthesis of Example 1.
[0024] Figure 2This is a schematic diagram of the synthesis of Example 2.
[0025] Figure 3 This is a schematic diagram of the synthesis of Example 3. DETAILED DESCRIPTION
[0026] The present invention aims to provide a method for synthesizing textile texture and color based on example images, which can synthesize the pattern and texture of example images in real time. The method is based on the visual appearance of the image, rather than the physical characteristics of the fabric, so it can present the synthesis effect quickly, interactively and intuitively. The method allows flexible user control to achieve real-time simulation of fabric appearance based on fabric example images and color images.
[0027] The technical solution of the present invention is described in detail below in conjunction with specific embodiments:
[0028] (1) Texture normalization: First, the fabric image I f , transform to YIQ color space, and extract fabric image I f The Y component of the image is used to characterize the undulation, thickness and roughness of the fabric texture appearance; the specific frequency band of the fabric image is related to the specific fabric texture appearance, that is, the specific frequency band of the fabric image is related to the fabric undulation, thickness and roughness. The undulation of the fabric corresponds to a low-frequency band, which corresponds to 0.57-2.29 cycles / degree; a frequency band corresponds to the thickness of the fabric or knitted fabric, that is, it corresponds to 2.29-4.28 cycles / degree; the roughness of the fabric corresponds to a medium-frequency band, which corresponds to 6.57-15.14 cycles / degree. According to the band interval of the specific frequency band, the abnormal band information is eliminated to achieve the normalization of the fabric texture;
[0029] (2) Adaptive color conversion: After texture normalization, the color information is converted from the color image I c Transfer to fabric image I f The YIQ value of the output color conversion image is determined by the color image I c The IQ component and Y′ of the color image are composed of f The Y component deviation is added to the color image I c The fabric image I is defined by subtracting the average value of the Y component from the Y component of each pixel. f The deviation of the Y component in f The AC component of the input color image I in frequency space c All components are added together; the average Y component of the output color conversion image is the same as the Y′ component of the original color image; then the Y component of the output color conversion image is recorded as Y′′:
[0030]
[0031] Among them, (x, y) represents the image coordinates, Y f is the Y component of the input fabric image, is the average value of the Y component of the input fabric image, Y c is the Y′ component of the input color image.
[0032] (3) High frequency removal: After color transfer, due to the high edge intensity and higher spatial frequency of the converted image than the fabric image, the color converted image gives the impression that the texture and color are not merged in the same layer, which leads to the visual effect of the transferred image being unnatural. In order to reduce the unnaturalness in the color converted image, a Gaussian filter is applied to the input color image to suppress these two features.
[0033] In the RGB color space, the image is processed by Gaussian filtering, and the image is convolved using the Gaussian kernel generated by the Gaussian function, that is:
[0034] I σ =I×G σ
[0035]
[0036] Where I is the fabric texture image, * represents the convolution operation, G σ represents a two-dimensional Gaussian kernel with a standard deviation of σ. In this method, the interval of σ is [0.5, 5.0]. Here, the user can select different σ values in this interval to achieve real-time control of the fabric improvement effect. After Gaussian filtering, the color information is transferred to the fabric texture image in the YIQ color space to achieve the synthesis of texture and color, and the synthetic image I is obtained. N .
[0037] Example 1
[0038] like Figure 1 As shown, the texture image 1 and the color image 1 are synthesized into a fabric image 1 by using the above method;
[0039] Example 2
[0040] like Figure 2 As shown, the texture image 2 and the color image 2 are synthesized into a fabric image 2 by using the above method;
[0041] Example 3
[0042] like Figure 3 As shown, the texture image 3 and the color image 3 are synthesized into the fabric image 3 by using the above method.
[0043] The above are only specific embodiments of the present invention, but the technical features of the present invention are not limited thereto. Any simple changes, equivalent replacements or modifications made based on the present invention to solve basically the same technical problems and achieve basically the same technical effects are all included in the protection scope of the present invention.
Claims
1. A method for synthesizing textile texture and color based on example images, characterized in that The steps include: (1) Texture normalization: First, the fabric image I f , transform to YIQ color space, and extract fabric image I f The Y component of the image is used to characterize the undulation, thickness and roughness of the fabric texture appearance. The specific frequency band of the fabric image is related to the specific fabric texture appearance. The abnormal band information is eliminated according to the band interval of the specific frequency band to realize the normalization of the fabric texture. (2) Adaptive color conversion: After texture normalization, the color information is converted from the color image I c Transfer to fabric image I f The YIQ value of the output color conversion image is determined by the color image I c The IQ component and Y′ of the color image are composed of f The Y component deviation is added to the color image I c In the Y′ component of (3) High-frequency removal: After color transfer, in order to reduce the unnaturalness in the color transfer image, Gaussian filtering is applied to the input color image to suppress texture and color. After Gaussian filtering, the color information is transferred to the fabric texture image in the YIQ color space to achieve the synthesis of texture and color, and obtain the synthesized image I N .
2. The method for synthesizing textile texture and color based on example images according to claim 1, characterized in that: In the step (1), the undulation of the fabric corresponds to a low-frequency band, and the low-frequency band corresponds to 0.57-2.29 cycles / degree.
3. The method for synthesizing textile texture and color based on example images according to claim 1, characterized in that: In the step (1), the thickness of the fabric corresponds to 2.29-4.28 cycles / degree.
4. The method for synthesizing textile texture and color based on example images according to claim 1, characterized in that: In the step (1), the roughness of the fabric corresponds to a medium frequency band, and the medium frequency band corresponds to 6.57-15.14 cycles / degree.
5. The method for synthesizing textile texture and color based on example images according to claim 1, characterized in that: In step (2), the fabric image I f The Y component deviation is added to the color image I c The fabric image I is defined by subtracting the average value of the Y component from the Y component of each pixel. f The deviation of the Y component in f The AC component of the input color image I in frequency space c All components of are added together; the average Y component of the output color conversion image is the same as the Y′ component of the original color image; then the Y component of the output color conversion image is denoted as Y″: Among them, (x, y) represents the image coordinates, Y f is the Y component of the input fabric image, is the average value of the Y component of the input fabric image, Y c is the Y′ component of the input color image.
6. The method for synthesizing textile texture and color based on example images according to claim 1, characterized in that: In step (3), the image is processed by Gaussian filtering in the RGB color space, and a convolution operation is performed on the image using a Gaussian kernel generated by a Gaussian function, that is: I σ =I×G σ Where I is the fabric texture image, * represents the convolution operation, G σ represents a two-dimensional Gaussian kernel with a standard deviation of σ. In this method, the interval of σ is [0.5, 5.0]. Here, the user can select different σ values in this interval to achieve real-time control of the fabric improvement effect. After Gaussian filtering, the color information is passed to the fabric texture image in the YIQ color space to achieve the synthesis of texture and color.
