CN114285955B - Color gamut mapping method based on dynamic deviation map neural network - Google Patents

Abstract

The color gamut mapping method based on the dynamic deviation map neural network comprises the following steps: 1) Collecting a color sample of a printing and dyeing printer; 2) Building and training a neural network from XYZ to CMKY color space deviation mapping chart; 3) Performing coarse-grained local mapping reinforcement based on a mask; 4) Adjusting the optimization for a machine-specific map neural mapping network; 5) Mask-based machine-specific local mapping enforcement; 6) Mapping from CMYK to XYZ color space; 7) Color gamut mapping based on local range matching. According to the color gamut mapping method based on the dynamic deviation graph neural network established in the steps, the dynamic deviation is introduced into the neural network, and the accurate mapping of colors from the color space of the printing and dyeing printer to the standard color space is realized while the color mapping from the color space of the printing and dyeing printer to the standard color space is learned from a large number of collected printing and dyeing samples.

Description

Color Gamut Mapping Method Based on Dynamic Deviation Graph Neural Network

technical field

The invention belongs to the field of color mapping. Aiming at the problems of uneven color space and inaccurate color mapping between printing and dyeing printers and the standard color space CIE XYZ, a color gamut mapping method based on a dynamic deviation graph neural network is proposed. The ability to learn nonlinear conversion from a large number of samples, by collecting a large number of printer color space color samples, printing through different printers, and measuring with color measuring instruments, on the basis of obtaining a large number of color mapping sample pairs, combining the differences of different printing and dyeing printers and different The difference between printing and dyeing printers at the same time, constructing a dynamic deviation map neural network, and completing the accurate mapping between the color of the standard color space and the color of different printing and dyeing printers.

Background technique

With the rapid development of deep learning technology, deep models have made a series of breakthroughs in computer and interdisciplinary fields. In the field of deep learning, graph neural networks have made some progress in spatial mapping: such as literature 1 (Xin Gao, Zhenjiang Liu, Zunlei Feng, Chengji Shen, Kairi Ou, Haihong Tang, MingliSong, Shape Controllable Virtual Try-on for Underwear Models , ACM Multimedia2021) uses a graph neural network to realize the mapping of clothing key points between tiled clothing and human body clothing; Document 2 (Wang N, Zhang Y, Li Z, et al. Pixel2Mesh: Generating 3D Mesh Models from SingleRGB Images[C]//European Conference on Computer Vision. Springer, Cham, 2018) solves the mapping between images and 3D mesh vertices through the graph neural network, and realizes the reconstruction of a single image to a three-dimensional mesh. The current graph neural network has been verified in many fields to realize the mapping of different spaces is effective and feasible, and the color gamut mapping based on the graph neural network is less, the main reason is that different printing and dyeing printers and the printing and dyeing color rendering of the same printer in different time periods have different characteristics. Bias.

In terms of color gamut mapping, existing color gamut mapping methods mainly include two categories: image-related dynamic color gamut mapping methods and fixed color gamut mapping methods. The image-related dynamic color gamut mapping method mainly considers the characteristics of the color values in the image to minimize the error in the color conversion of the image. This type of method needs to combine the characteristics of the image color and dynamically adjust the mapping relationship between color points in the color space. A better color rendering effect is obtained, but the processing time is slower and it is not suitable for large-scale promotion. Fixed color gamut mapping methods mainly include three categories: physical conversion model, numerical quantization conversion model, and 3D LUT method. The physical conversion model generally assumes that the color channels are independent of each other, the chromaticity is constant, and the color space is uniform. However, when the human eye observes the real color, the color difference is not uniform; the numerical quantization conversion model learns the device-related color space and device through the numerical model. Regardless of the conversion relationship of the color space, the numerical models mainly include polynomial regression method, neural network method, continuous linear interpolation method, radial basis function method, etc., for example, Document 3 (Song Mingli, Sheng Nan, Feng Zunlei, etc. Color Consistency Mapping Method for Inkjet Printing and Dyeing of Textiles: CN110418030A[P].2019) The RBF neural network is used to realize the mapping between the CMYK color space of the printing and dyeing machine and the RGB color space of the display, but the limited learning ability of RBF limits the color mapping At the same time, this work did not consider the color deviation of different machines and different times. The 3D LUT method establishes the conversion relationship between the device-independent color space and the device-dependent color space in a table. The accuracy of the 3D LUT method mainly depends on the number of colors measured and the interpolation method selected during color conversion. In addition, some European companies have achieved color mapping between multiple points in the color space and achieved better accuracy. However, due to commercial privacy, the technology used has not been made public, and the technical route is unknown.

Contents of the invention

The present invention solves the problem of printing and dyeing color deviation between different printers and the same printer at different times, and realizes accurate mapping of colors from the color space of the printing and dyeing printer to the standard color space.

Textile printing and dyeing printers have always faced the challenge of unstable color printing and dyeing and the large difference between the original color space. The main problem is that the color space is not uniformly distributed and the color rendering of the printer is biased. The present invention proposes a color gamut mapping method based on a dynamic deviation map neural network. By introducing the dynamic deviation into the neural network, the color mapping between the printing and dyeing printer color space and the standard color space is learned from a large number of collected printing and dyeing samples. Realize the accurate mapping of colors from printing and dyeing printer color space to standard color space.

A color gamut mapping method based on a dynamic deviation map neural network, comprising the following steps:

1) Collection of color samples for printing and dyeing printers;

墨量为中心点，以(-d％,d％)为扰动采样区间，四个通道通过组合获得T⁴个颜色样本点；对于P个打印机，通过每个打印采集Q次，共获得P*Q*T⁴个颜色样本点；对于印染的样本，通过颜色测量仪i1Pro2测得所有样本点CIE XYZ颜色空间对应颜色点；The collection of color samples of printing and dyeing printers mainly includes the collection of color samples of different printers and the collection of different colors of the same printer. The present invention collects samples of P printers, and each printing and dyeing printer collects Q times; for each printing and dyeing printer C, M, Y , K four color channels, each channel collects T color points at a time, respectively with

Ink volume as the center point, with (-d%, d%) as the disturbance sampling interval, the four channels are combined to obtain T ⁴ color sample points; for P printers, each printing is collected Q times, and a total of P* Q*T ⁴ color sample points; for printing and dyeing samples, the corresponding color points of all sample points in CIE XYZ color space are measured by the color measuring instrument i1Pro2;

2) Neural network construction and training from XYZ to CMYK color space deviation map;

For the Q-time color point sampling of each printing and dyeing machine, according to the order of C, M, Y, K, according to the size of the ink volume in each channel, the color value matrix D of T ⁴ * 4 is obtained, and the corresponding CIE XYZ The color value matrix of the color space is S, and according to the adjacency relationship in the C, M, Y, K four-color spaces, the adjacency matrix A of T ⁴ feature points is constructed; using the graph convolutional network H ^l+1 = σ(AH ^l W ^l ) Perform a three-layer convolution operation on the input color value matrix S to obtain the predicted input O=H ³ , where H ⁰ =S, σ is the ReLU activation function; use the following deviation mapping loss function L ₁ to obtain the prediction input for all samples There is a biased mapping:

Where e is the threshold of color deviation, O _i and D _i are the color values of the i-th row of samples; for all samples, a preliminary color mapping network is obtained through R iterations;

3) Mask-based coarse-grained local mapping enhancement;

For the constructed graph neural network with T ⁴ nodes, 10%, 20%, 30%, 40%, 50%, 60%, 70%, 80%, 90%, 95 % sample points, use the graph convolution network H ^l+1 = σ(A'H ^l W ^l ) to perform three-layer convolution operation on the input color value matrix S after the M mask to obtain the predicted input O' = H ³ , where A' is the adjacency matrix after the mask, H ⁰ =mS, m is the random mask that increases with the number of iterations, the gradual increase ratio of the random mask is evenly distributed in the R times of iterations, and for obtaining the mask input O ', trained with a coarse _- grained local map reinforcement loss function L2 after the mask:

Where |m| is the color sample point retained after the mask;

4) Adjust and optimize the graph neural mapping network for specific machines;

For the preliminary color mapping network obtained in step 3), for a specific machine, use the collected Q samples to increase the anti-noise ability of the network by randomly adding [-u,u] micro-perturbation to the input color value, using the following Specific precise mapping loss function L ₃ :

In the first R' iterative optimization, the network is trained using color samples with added noise disturbance, and then in the subsequent R' iterative optimization, the network is trained using color samples without noise perturbation to obtain XYZ for a specific machine ->CMYK accurate color mapping network;

5) Mask-based machine-specific local mapping enhancement;

For the refined color mapping network obtained in step 4), by randomly masking out 90% and 95% of the sample points in the first R'/2 iterations and the last R'/2 iterations respectively, using the graph convolutional network H ^l+1 = σ(A'H ^l W ^l ) Perform three-layer convolution operation on the input color value matrix S after the M mask to obtain the predicted input O"=H ³ , where A' is the adjacency matrix after the mask, H ⁰ = mS, m is a random mask that increases with the number of iterations. For the mask input O', use the mask-specific machine local fine mapping to strengthen the loss function L ₄ for training:

Where |m| is the color sample point retained after the mask;

6) Mapping from CMYK to XYZ color space;

Through steps 2), 3), 4), and 5), the mapping from XYZ to CMYK color space is realized. By using CMYK as the color input sample point and XYZ as the output color sample point, construct a color space deviation map neural network from CMYK to XYZ mapping. Network, through step 2) similar bias coarse-grained mapping, step 3) mask-based coarse-grained local map reinforcement, step 4) machine-specific graph neural map network tuning optimization, step 5) mask-based machine-specific local Mapping enhancement and other four steps to realize the mapping from CMYK to XYZ color space;

7) Color gamut mapping based on local range matching;

墨量以内的图节点，对于输入的XYZ颜色值，需要匹配到T⁴个图节点中

以内色差值的图节点，对于不需要映射的颜色值点，设置掩码为0，通过以上方式实现CMYK到XYZ的颜色双向映射。In practical applications, the realization of color value mapping requires coarse-grained matching of the input color domain. For the input CMYK color value, it needs to be matched to T ⁴ graph nodes

For the graph nodes within the ink volume, for the input XYZ color value, it needs to be matched to T ⁴ graph nodes

For the graph nodes of the inner color difference value, for the color value points that do not need to be mapped, set the mask to 0, and realize the two-way color mapping from CMYK to XYZ through the above method.

Preferably, the threshold e of color deviation in step 2) is set to 3.

Preferably, the number of printers P in step 1) is 100, the number of printing acquisitions Q is 100, and the number of color points T is 10.

The method of the invention is a color gamut mapping method based on a dynamic deviation graph neural network, and is used for the mapping conversion of color values between the CMYK color space of a textile printing and dyeing printer and the standard CIE XYZ color space.

The color gamut mapping method based on the dynamic deviation map neural network established through the above steps, by introducing the dynamic deviation into the neural network, learns the color mapping between the printing and dyeing printer color space and the standard color space from a large number of collected printing and dyeing samples, and realizes Accurate mapping of colors from printing and dyeing printer color space to standard color space.

The beneficial effects of the present invention are: based on the powerful learning ability of the graph neural network, on the basis of collecting a large number of samples, by modeling the color deviation of different printing and dyeing printers and the same printer at different times into the graph neural network color mapping model, realizing Dynamic and accurate mapping of colors.

Description of drawings

Figure 1 is a flow chart of the method of the present invention.

detailed description

The technical solution of the present invention will be further described below in conjunction with the accompanying drawings.

The present invention is based on the color gamut mapping method of dynamic deviation graph neural network, comprises the following steps:

1) Collection of color samples for printing and dyeing printers;

The collection of color samples of printing and dyeing printers mainly includes the collection of color samples of different printers and the collection of different colors of the same printer. The present invention collects samples of 100 printers, and each printing and dyeing printer collects 100 times; for each printing and dyeing printer's C, M, Y , K four color channels, each channel takes {5%, 15%, 25%, 35%, 45%, 55%, 65%, 75%, 85%, 95%} ink volume as the center point, with (-5%, 5%) is the disturbance sampling interval, each channel collects 10 color points at a time, and the four channels are combined to obtain 10,000 color sample points; for 100 printers, each print is collected 100 times, a total of Obtain 100*100*10000 color sample points; for printing and dyeing samples, measure the corresponding color points of all sample points in CIE XYZ color space through the color measuring instrument i1Pro2;

2) Neural network construction and training from XYZ to CMYK color space deviation map;

For 100 color point samples of each printing and dyeing machine, according to the order of C, M, Y, K, according to the size of the ink volume in each channel, a 10000*4 color value matrix D is obtained, and the corresponding CIE XYZ color The color value matrix of the space is S, and the adjacency matrix A of 10,000 feature points is constructed according to the adjacency relationship in the C, M, Y, and K four-color spaces; using the graph convolutional network H ^l+1 = σ(AH ^l W ^l ) Perform a three-layer convolution operation on the input color value matrix S to obtain the predicted input O=H ³ , where H ⁰ =S, σ is the ReLU activation function; use the following bias mapping loss function L ₁ to obtain biased values for all samples Mapping relations:

Wherein e is the threshold value of color deviation (set to 3 in the present invention), and O _i and D _i are i-th row sample color values; Obtain preliminary color mapping network by 50 iterations for all samples;

3) Mask-based coarse-grained local mapping enhancement;

For the constructed graph neural network of 10000 nodes, by gradually random masking out 10%, 20%, 30%, 40%, 50%, 60%, 70%, 80%, 90%, 95% in 50 iterations For the sample points of , use the graph convolutional network H ^l+1 = σ(A'H ^l W ^l ) to perform three-layer convolution operation on the input color value matrix S after the M mask to obtain the predicted input O'=H ³ , where A 'is the adjacency matrix after the mask, H ⁰ =mS, m is the random mask that increases with the number of iterations, the gradual increase ratio of the random mask is distributed on average in 50 iterations, and for obtaining the mask input O' , using the coarse _- grained local mapping reinforcement loss function L2 after the mask for training:

Where |m| is the color sample point retained after the mask;

4) Adjust and optimize the graph neural mapping network for specific machines;

For the preliminary color mapping network obtained in step 3), for a specific machine, use 100 samples collected to increase the anti-noise ability of the network by randomly adding [-0.1,0.1] micro-perturbation to the input color value, using the following Specific precise mapping loss function L ₃ :

In the first 10 iterative optimizations, the network is trained using color samples with added noise perturbation, and then in the next 10 iterative optimizations, the network is trained using color samples without noise perturbation to obtain XYZ-> for a specific machine CMYK accurate color mapping network;

5) Mask-based machine-specific local mapping enhancement;

For the refined color mapping network obtained in step 4), by randomly masking out 90% and 95% of the sample points in the first 5 iterations and the last 5 iterations respectively, using the graph convolutional network H ^l+1 =σ( A'H ^l W ^l ) Perform three-layer convolution operation on the input color value matrix S after the M mask to obtain the predicted input O"=H ³ , where A' is the adjacency matrix after the mask, H ⁰ =MS, and M is With random masks increasing in number of iterations, for obtaining masked input O', train with mask-specific machine-local fine _- map augmentation loss function L4:

Where |M| is the color sample point retained after the mask;

6) Mapping from CMYK to XYZ color space;

Through steps 2), 3), 4), and 5), the mapping from XYZ to CMYK color space is realized. By using CMYK as the color input sample point and XYZ as the output color sample point, construct a color space deviation map neural network from CMYK to XYZ mapping. Network, through step 2) similar bias coarse-grained mapping, step 3) mask-based coarse-grained local map reinforcement, step 4) machine-specific graph neural map network tuning optimization, step 5) mask-based machine-specific local Mapping enhancement and other four steps to realize the mapping from CMYK to XYZ color space;

7) Color gamut mapping based on local range matching;

In practical applications, the realization of color value mapping requires coarse-grained matching of the input color domain. For the input CMYK color value, it needs to be matched to the graph nodes within 5% of the 10,000 graph nodes. For the input XYZ The color value needs to match the graph nodes with a color difference value within 20 among the 10,000 graph nodes. For the color value points that do not need to be mapped, set the mask to 0, and realize the two-way color mapping from CMYK to XYZ through the above method.

The method of the invention is a color gamut mapping method based on a dynamic deviation graph neural network, and is used for the mapping conversion of color values between the CMYK color space of a textile printing and dyeing printer and the standard CIE XYZ color space.

The beneficial effects of the present invention are: based on the powerful learning ability of the graph neural network, on the basis of collecting a large number of samples, by modeling the color deviation of different printing and dyeing printers and the same printer at different times into the graph neural network color mapping model, realizing Dynamic and accurate mapping of colors.

The content described in the embodiments of this specification is only an enumeration of the implementation forms of the inventive concept. The protection scope of the present invention should not be regarded as limited to the specific forms stated in the embodiments. The protection scope of the present invention also extends to the field Equivalent technical means that the skilled person can think of based on the concept of the present invention.

Claims (3)

1. The color gamut mapping method based on the dynamic deviation map neural network comprises the following steps:

1) Collecting a color sample of a printing and dyeing printer;

collecting color samples of the printing and dyeing printers comprises collecting color samples of different printing and dyeing printers and collecting different colors of the same printing and dyeing printer, collecting the samples of P printing and dyeing printers, and collecting Q times of each printing and dyeing printer; for the four color channels of C, M, Y and K of each printing and dyeing printer, each channel collects T color points once, and the color points are respectively processed by

The ink amount is taken as a central point, (-d%, d%) is taken as a disturbance sampling interval, and T is obtained by combining four channels ⁴ A color sample point; for P printing printers, Q times are collected by each printing printer to obtain P × Q × T ⁴ A color sample point; for the printed and dyed sample, measuring color points corresponding to all sample points CIE XYZ color space through a color measuring instrument i1Pro 2;

2) Building and training a neural network from XYZ to CMYK color space deviation mapping;

and (3) sampling Q times of color points of each printing and dyeing printer, and arranging the color points according to the sequence of C, M, Y and K and the size of the ink in each channel to obtain T ⁴ *4, the color value matrix D of the CIE XYZ color space corresponding to the color value matrix D is S, and T is constructed according to the adjacent relation in the four color spaces of C, M, Y and K ⁴ An adjacency matrix A of feature points; using graph convolution networks H ^l+1 ＝σ(AH ^l W ^l ) Performing three-layer convolution operation on an input color value matrix S to obtain a predicted input O = H ³ In which H is ⁰ = S, σ is the ReLU activation function; mapping the loss function L with the following bias ₁ Obtaining a deviation mapping relation for all samples:

where e is the threshold value for the color deviation, O _i And D _i Is the sample color value of the ith row; obtaining a preliminary color mapping network by performing R iterations on all samples;

3) Strengthening coarse-grained local mapping based on a mask;

for constructed T ⁴ The graph neural network of nodes utilizes graph convolution network H by progressively randomly masking out 10%,20%,30%,40%,50%,60%,70%,80%,90%,95% of the sample points in R iterations ^l+1 ＝σ(A′H ^l W ^l ) Performing three-layer convolution operation on the M-masked input color value matrix S to obtain a predicted input O' = H ³ Where A' is the masked adjacency matrix, H ⁰ Where m is a random mask with increasing iteration number, the gradual increase proportion of the random mask is evenly and gradually distributed in R iteration numbers, and for obtaining a mask input O', a loss function L is strengthened by using masked coarse-grained local mapping ₂ Training is carried out:

wherein | m | is a color sample point reserved after the mask;

4) Adjusting the optimization for a machine-specific map neural mapping network;

for the preliminary color mapping network obtained in step 2), for a particular machine, the input color values are incremented by [ -u, u ] by random using the Q samples collected]To increase the noise immunity of the network, using a specific accurate mapping loss function L as follows ₃ ：

In the previous R 'iteration optimization, the color sample added with noise disturbance is used for training the network, and then in the subsequent R' iteration optimization, the color sample without noise disturbance is used for training the network, so that an XYZ- > CMYK accurate color mapping network for a specific machine is obtained;

5) Mask-based specific machine local mapping enforcement;

with respect to XYZ->The CMYK accurate color mapping network utilizes a graph convolution network H by masking out 90% and 95% of sample points randomly in the first R '/2 and the last R'/2 iterations, respectively ^l+1 ＝σ(A′H ^l W ^l ) Performing three-layer convolution operation on the M-masked input color value matrix S to obtain a prediction input O = H ³ Where A' is the masked adjacency matrix, H ⁰ = mS, m being a random mask with increasing number of iterations, for obtaining a mask input O', a penalty function L is enforced using mask-specific machine local fine mapping ₄ Training is carried out:

wherein | m | is a color sample point reserved after the mask;

6) Mapping from CMYK to XYZ color space;

constructing a color space deviation mapping graph neural network for mapping CMYK to XYZ by using CMYK as a color input sample point and XYZ as an output color sample point, so as to realize the mapping from CMYK to XYZ color space;

7) Color gamut mapping based on local range matching;

in practical applications, coarse-grained matching of input color fields is required to realize mapping of color values, and matching of input CMYK color values to T is required ⁴ In a node of a graph

The graph nodes within the ink volume need to be matched to T for the input XYZ color values ⁴ An imageIn the node

The color bidirectional mapping of CMYK to XYZ is realized by setting the mask to 0 for a color value point that does not need to be mapped with a map node of an internal color difference value in the above manner.

2. The dynamic deviant graph neural network-based color gamut mapping method of claim 1, wherein: the threshold e for the color deviation of step 2) is set to 3.

3. The dynamic deviant map neural network-based color gamut mapping method of claim 1, wherein: the number P of the printing and dyeing printers in the step 1) is 100, the collection times Q of each printing and dyeing printer is 100, and the number T of the color points is 10.

Images