CN111080739B - BI color matching method and system based on BP neural network - Google Patents
BI color matching method and system based on BP neural network Download PDFInfo
- Publication number
- CN111080739B CN111080739B CN201911365727.9A CN201911365727A CN111080739B CN 111080739 B CN111080739 B CN 111080739B CN 201911365727 A CN201911365727 A CN 201911365727A CN 111080739 B CN111080739 B CN 111080739B
- Authority
- CN
- China
- Prior art keywords
- color matching
- matching model
- neural network
- effect
- module
- Prior art date
- Legal status (The legal status is an assumption and is not a legal conclusion. Google has not performed a legal analysis and makes no representation as to the accuracy of the status listed.)
- Active
Links
Images
Classifications
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06T—IMAGE DATA PROCESSING OR GENERATION, IN GENERAL
- G06T11/00—2D [Two Dimensional] image generation
- G06T11/001—Texturing; Colouring; Generation of texture or colour
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06N—COMPUTING ARRANGEMENTS BASED ON SPECIFIC COMPUTATIONAL MODELS
- G06N3/00—Computing arrangements based on biological models
- G06N3/02—Neural networks
- G06N3/04—Architecture, e.g. interconnection topology
- G06N3/044—Recurrent networks, e.g. Hopfield networks
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06N—COMPUTING ARRANGEMENTS BASED ON SPECIFIC COMPUTATIONAL MODELS
- G06N3/00—Computing arrangements based on biological models
- G06N3/02—Neural networks
- G06N3/08—Learning methods
- G06N3/084—Backpropagation, e.g. using gradient descent
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06T—IMAGE DATA PROCESSING OR GENERATION, IN GENERAL
- G06T11/00—2D [Two Dimensional] image generation
- G06T11/40—Filling a planar surface by adding surface attributes, e.g. colour or texture
Abstract
The invention discloses a BI color matching method based on a BP neural network, which relates to the technical field of BI color matching and comprises the following steps: collecting an existing effect graph, and adding labels according to the client requirements corresponding to the effect graph; collecting color scheme data, extracting RGB values of main colors in an effect graph, and taking the RGB values of the main colors of the effect graph as training samples; establishing a BI color matching model based on a BP neural network, taking a label of an effect diagram as input data of the BI color matching model, taking RGB values of main color matching in the effect diagram as output data of the BI color matching model, and training the BI color matching model; and verifying whether the output result of the BI color matching model has practicability, and optimizing the BI color matching model when the output result of the BI color matching model does not have practicability. According to the method, the color matching scheme can be automatically derived according to the label through the optimized BI color matching model, and a basic color matching basis is provided for a UI designer. The invention also discloses a BI color matching system based on the BP neural network, which can automatically derive a color matching scheme to help a UI designer to improve the working efficiency.
Description
Technical Field
The invention relates to the technical field of BI color matching, in particular to a BI color matching method and system based on a BP neural network.
Background
With the rapid development of information technology, the large data age has come. More and more governments and enterprises need to conduct work by means of multidimensional data analysis, and therefore, the use of the data visualization tool BI is also becoming more and more widespread. The excellent color scheme of the BI not only can meet the aesthetic requirements of customers, but also can highlight the information value of the data. The BI color scheme system is researched, a basic color scheme can be provided for a designer, the designer is assisted to quickly output an effect diagram, and the development progress of a BI tool is improved.
Disclosure of Invention
Aiming at the needs and the shortcomings of the prior art development, the invention provides a BI color matching method and system based on a BP neural network.
Firstly, the invention provides a BI color matching method based on a BP neural network, which solves the technical problems and adopts the following technical scheme:
a BI color matching method based on BP neural network, the realization of the method includes:
step 2, collecting color scheme data, extracting RGB values of main colors in an effect graph, and taking the RGB values of the main colors of the effect graph as training samples;
step 3, building a BI color matching model based on a BP neural network, then using the label of the effect diagram in the step 1 as input data of the BI color matching model, and using the RGB value of main color matching in the effect diagram in the step 2 as output data of the BI color matching model to train the BI color matching model;
step 4, verifying whether the output result of the BI color matching model has practicability,
a) The output result has practicability, and the BI color matching model is qualified;
b) If the output result has no practicability, the BI color matching model is unqualified, and the step 5 is continuously executed;
and 5, optimizing the BI color matching model, inputting labels required by customers into the final optimized BI color matching model, and outputting RGB values of main color matching of the effect graph corresponding to the input labels by the final optimized BI color matching model.
In step 1, an existing effect diagram is collected, and a label is added according to a client requirement corresponding to the effect diagram, and the specific operation comprises the following steps:
step 1.1, collecting an existing effect diagram, and analyzing client attributes corresponding to the effect diagram;
and 1.2, adding labels to the effect graph according to the analysis result, wherein the labels comprise customer attribute labels, style labels and main tone labels, and the main tone labels comprise at least three colors.
In step 2, color scheme data is collected, and RGB values of main colors in the effect graph are extracted, wherein the specific operations include:
and analyzing the effect graph by using Photoshop or other drawing tools, and sequentially extracting RGB values of at least three colors with the duty ratio exceeding a set threshold value in the effect graph.
In step 3, the specific process of building and training the BI color matching model includes:
step 3.1, building a BI color matching model based on a three-layer structure of an input layer, an hidden layer and an output layer of the BP neural network;
step 3.2, introducing a training algorithm into the BI color matching model established in the step 3.1, inputting the label of the effect diagram in the step 1 into the BI color matching model, and outputting RGB values of main color matching in the effect diagram by the BI color matching model;
step 3.3, comparing the RGB value of the main color matching in the extraction effect diagram of the step 2 with the output result of the BI color matching model of the step 3.2, and adjusting the BI color matching model according to the comparison result;
and 3.4, setting the comparison similarity between the extracted RGB values and the output results, and circularly executing the steps 3.2-3.3 based on different effect graphs until the comparison results exceed the set comparison similarity, so as to complete training of the BI color matching model.
In the process of building and training the BI color matching model, the hidden layer of the BP neural network is a single hidden layer, the number of hidden layer nodes is set to 7, and the introduced training algorithm is set to be a Bayesian regularization algorithm.
In step 5, optimizing the BI color matching model, wherein the specific optimization includes:
step 5.1, expanding a training sample, and optimizing the training sample;
step 5.2, adjusting the hidden layer number and the hidden layer node number of the BI color matching model just established by a trial and error method, and optimizing the hidden layer number and the hidden layer node number;
step 5.3, optimizing the selection of an initial threshold and a weight of the BI color matching model which is just built by using a particle swarm optimization algorithm;
step 5.4, replacing the activation function between the input layer and the hidden layer and the activation function between the hidden layers with a log-sigmoid function, and optimizing the activation function;
and 5.5, adding 'stop in advance' setting in the BI color matching model just established, and preventing the BP neural network and training samples from being excessively fitted and the generalization prediction capability from being deteriorated.
Secondly, the invention also provides a BI color matching system based on the BP neural network, which solves the technical problems and adopts the following technical scheme:
a BP neural network-based BI color matching system, comprising:
the collection marking module is used for collecting the existing effect graph, adding labels according to the client requirements corresponding to the effect graph, and taking the labels of the effect graph as training samples;
the collecting and extracting module is used for collecting color scheme data and extracting RGB values of main colors in the effect graph, and the RGB values of the main colors of the effect graph are used as training samples;
the construction module is used for constructing a BI color matching model based on the BP neural network;
the training module is used for taking the effect graph label of the collecting and marking module as input data of the BI color matching model, taking the RGB value of main color matching in the effect graph of the collecting and extracting module as output data of the BI color matching model, and training the BI color matching model;
the input/output module is used for inputting the effect icon into the BI color matching model and further sending the output result of the BI color matching model to the setting verification module;
the verification module is used for setting a practical threshold value of the BI color matching model, and verifying whether the output result of the input/output module meets the practical threshold value based on the RGB value of the main color matching of the acquisition and extraction module extraction effect diagram;
and the optimization module is used for optimizing the BI color matching model to obtain a final optimized BI color matching model when the verification result of the set verification module has no practicability, and the output result of the final optimized BI color matching model meets the set threshold.
Optionally, the involved collection marking module adds a label to the effect map by:
collecting the existing effect graph, and analyzing the client attribute corresponding to the effect graph;
adding labels to the effect graph according to the analysis result, wherein the labels comprise customer attribute labels, style labels and dominant hue labels, and the dominant hue labels comprise at least three colors;
and then, the acquisition and extraction module analyzes the effect graph by using a Photoshop or other drawing tools, and sequentially extracts RGB values of at least three colors with the duty ratio exceeding a set threshold value in the effect graph.
Optionally, the related construction module establishes a BI color matching model based on three layers of an input layer, an hidden layer and an output layer of the BP neural network;
the related training module trains the process of the BI color matching model comprises the following steps:
firstly, introducing a training algorithm into a BI color matching model, inputting a label of an effect diagram into the BI color matching model, and outputting RGB values of main color matching in the effect diagram by the BI color matching model;
secondly, comparing the RGB values of the main color matching in the extraction effect diagram of the acquisition and extraction module with the output result of the BI color matching model, and adjusting the BI color matching model according to the comparison result;
finally, training of the BI color matching model is completed when the comparison result meets the threshold set by the set verification module.
Optionally, when the verification result of the verification module is set to have no practicability, the related optimization module optimizes the BI color matching model through the following measures:
expanding a training sample, and optimizing the training sample;
adjusting the hidden layer number and the hidden layer node number of the BI color matching model by a trial and error method, and optimizing the hidden layer number and the hidden layer node number;
optimizing the selection of an initial threshold value and a weight of the BI color matching model by using a particle swarm optimization algorithm;
replacing an activation function between the input layer and the hidden layer and an activation function between the hidden layer with a log-sigmoid function to optimize the activation function;
and adding 'stop in advance' setting when the construction module builds the BI color matching model, so as to prevent the BP neural network and training samples from being excessively fitted and the generalization prediction capability from being deteriorated.
The BI color matching method and system based on the BP neural network have the following beneficial effects compared with the prior art:
according to the invention, the BI color matching model is obtained through the label of the existing effect graph and the RGB value training of the main color matching, and further, the effect of automatically deriving the color matching scheme according to the requirements of a customer is realized through the BI color matching model, so that a basic color matching basis is provided for a UI designer, and the efficiency of outputting the effect graph is improved.
Drawings
FIG. 1 is a flow chart of a method according to a first embodiment of the invention
Fig. 2 is a block diagram showing structural connection of a second embodiment of the present invention.
The reference numerals in the drawings represent:
1. a collecting and marking module, a collecting and extracting module, a constructing module, a training module and a marking module,
5. the system comprises an input/output module, a setting verification module, a 7-optimization module, a 8-BI color matching model.
Detailed Description
In order to make the technical solution, the technical problems to be solved and the technical effects of the present invention more apparent, the technical solution of the present invention will be clearly and completely described below in conjunction with specific embodiments, and it is apparent that the described embodiments are only some embodiments of the present invention, but not all embodiments. All embodiments obtained by a person skilled in the art without making any inventive effort are within the scope of the present invention based on the embodiments of the present invention.
Embodiment one:
with reference to fig. 1, this embodiment proposes a BI color matching method based on a BP neural network, where implementation of the method includes:
step 1.1, collecting an existing effect diagram, and analyzing client attributes corresponding to the effect diagram;
step 1.2, adding labels to the effect graph according to the analysis result, wherein the labels comprise customer attribute labels, style labels and main tone labels, the main tone labels comprise four colors, and the main tone labels comprise four colors in general. Customer attribute tags may be "government", "business", "institutional", "scientific research", "public welfare organization", "personal", etc.; style labels may be "express," "science and technology," "strict," "business," etc.; the dominant hue labels may be "blue", "golden yellow", "green", etc. The labels added to the effect graph need to be as detailed as possible, in the actual process, labels with other dimensions can be added continuously, and the labels can be expressed by English.
And 2, collecting color scheme data, extracting RGB values of main colors in the effect graph, and storing the RGB values of the main colors of the effect graph as training samples. The specific implementation process of the step is as follows:
and analyzing the effect graph by using Photoshop or other drawing tools, and sequentially extracting RGB values of four colors with the duty ratio exceeding a set threshold value in the effect graph.
And 3, building a BI color matching model 8 based on the BP neural network, then using the label of the effect diagram in the step 1 as input data of the BI color matching model 8, and using the RGB value of the main color matching in the effect diagram in the step 2 as output data of the BI color matching model 8 to train the BI color matching model 8. The specific process of building and training the BI color matching model 8 includes:
step 3.1, building a BI color matching model 8 based on a three-layer structure of an input layer, an hidden layer and an output layer of the BP neural network; wherein the hidden layer of the BP neural network is a single hidden layer, and the number of hidden layer nodes is set to be 7;
step 3.2, introducing a training algorithm into the BI color matching model 8 established in the step 3.1, setting the introduced training algorithm as a Bayesian regularization algorithm, inputting the label of the effect graph in the step 1 into the BI color matching model 8, and outputting RGB values of main color matching in the effect graph by the BI color matching model 8;
step 3.3, comparing the RGB value of the main color matching in the extraction effect diagram of the step 2 with the output result of the BI color matching model 8 in the step 3.2, and adjusting the BI color matching model 8 according to the comparison result;
and 3.4, setting the comparison similarity between the extracted RGB values and the output results, and circularly executing the steps 3.2-3.3 based on different effect graphs until the comparison results exceed the set comparison similarity, so as to finish training of the BI color matching model 8.
Step 4, verifying whether the output result of the BI color matching model 8 has practicability,
a) The output result has practicability, and the BI color matching model 8 is qualified;
b) If the output result has no practicability, the BI color matching model 8 is unqualified, and the step 5 is continuously executed;
and 5, optimizing the BI color matching model 8, inputting labels required by customers into the final optimized BI color matching model 8, and outputting RGB values of main color matching of the effect map corresponding to the input labels by the final optimized BI color matching model 8. In this step, the specific content of optimizing the BI color matching model 8 includes:
step 5.1, expanding a training sample, and optimizing the training sample;
step 5.2, adjusting the hidden layer number and the hidden layer node number of the BI color matching model 8 just established by a trial and error method, and optimizing the hidden layer number and the hidden layer node number;
step 5.3, optimizing the selection of an initial threshold value and a weight of the BI color matching model 8 just established by using a particle swarm optimization algorithm;
step 5.4, replacing the activation function between the input layer and the hidden layer and the activation function between the hidden layers with a log-sigmoid function, and optimizing the activation function;
and 5.5, adding 'stop in advance' setting into the BI color matching model 8 just established, and preventing the BP neural network and training samples from being excessively fitted and the generalization prediction capability from being deteriorated.
Embodiment two:
referring to fig. 2, this embodiment proposes a BI color matching system based on a BP neural network, which includes:
the collection marking module 1 is used for collecting the existing effect graph, adding labels according to the client requirements corresponding to the effect graph, and taking the labels of the effect graph as training samples;
the collecting and extracting module 2 is used for collecting color scheme data and extracting RGB values of main color in the effect graph, wherein the RGB values of the main color of the effect graph are used as training samples;
the construction module 3 is used for constructing a BI color matching model 8 based on the BP neural network;
the training module 4 is configured to use the effect graph label of the collection and extraction module 1 as input data of the BI color matching model 8, and use RGB values of main color matching in the effect graph of the collection and extraction module 2 as output data of the BI color matching model 8 to train the BI color matching model 8;
the input/output module 5 is used for inputting the effect icon into the BI color matching model 8 and further sending the output result of the BI color matching model 8 to the setting verification module 6;
the setting verification module 6 is used for setting a practical threshold value of the BI color matching model 8 and verifying whether the output result of the input/output module 5 meets the practical threshold value based on the RGB value of the main color matching of the effect diagram extracted by the acquisition and extraction module 2;
the optimizing module 7 is used for optimizing the BI color matching model 8 to obtain a final optimized BI color matching model 8 when the verification result of the setting verification module 6 has no practicability, and the output result of the final optimized BI color matching model 8 meets the set threshold.
In the present embodiment, the collection marking module 1 concerned adds a label to the effect map by:
collecting the existing effect graph, and analyzing the client attribute corresponding to the effect graph;
adding labels to the effect graph according to the analysis result, wherein the labels comprise customer attribute labels, style labels and dominant hue labels, and the dominant hue labels comprise at least three colors; specifically, the customer attribute tags may be "government", "enterprise", "public institution", "scientific research institution", "public welfare organization", "individual", etc.; style labels may be "express," "science and technology," "strict," "business," etc.; the dominant hue labels may be "blue", "golden yellow", "green", etc. The labels added for the effect graph need to be as detailed as possible, in the actual process, the labels with other dimensions can be continuously added, and the labels can be expressed by English;
subsequently, the collection and extraction module 2 uses Photoshop or other drawing tools to analyze the effect graph, and sequentially extracts RGB values of at least three colors in the effect graph, wherein the RGB values are more than a set threshold.
In the embodiment, the related construction module 3 establishes a BI color matching model 8 based on three layers of an input layer, an hidden layer and an output layer of the BP neural network;
the process of training the BI color matching model 8 by the training module 4 involved includes:
firstly, introducing a training algorithm into a BI color matching model 8, inputting a label of an effect graph into the BI color matching model 8, and outputting RGB values of main color matching in the effect graph by the BI color matching model 8;
secondly, comparing the RGB value of the main color matching in the extraction effect diagram of the acquisition and extraction module 2 with the output result of the BI color matching model 8, and adjusting the BI color matching model 8 according to the comparison result;
finally, training of the BI color matching model 8 is completed when the comparison result meets the threshold set by the set verification module 6.
In the present embodiment, when the verification result of the verification module 6 is set to have no practicality, the optimization module 7 optimizes the BI color matching model 8 by:
expanding a training sample, and optimizing the training sample;
adjusting the hidden layer number and the hidden layer node number of the BI color matching model 8 by a trial and error method, and optimizing the hidden layer number and the hidden layer node number;
optimizing the selection of an initial threshold and weight of the BI color matching model 8 by using a particle swarm optimization algorithm;
replacing an activation function between the input layer and the hidden layer and an activation function between the hidden layer with a log-sigmoid function to optimize the activation function;
and adding a setting of 'stop in advance' when the construction module 3 constructs the BI color matching model 8, so as to prevent the BP neural network and the training sample from being excessively fitted and the generalization prediction capability from being deteriorated.
In summary, by adopting the BI color matching method and system based on the BP neural network, the BI color matching model 8 is obtained through the label of the existing effect graph and the RGB value training of main color matching, and further the effect of automatically deriving a color matching scheme according to the requirements of a customer is realized through the BI color matching model 8, so that a basic color matching basis is provided for a UI designer, and the efficiency of outputting an effect graph is improved.
Based on the above-mentioned embodiments of the present invention, any improvements and modifications made by those skilled in the art without departing from the principles of the present invention should fall within the scope of the present invention.
Claims (10)
1. The BI color matching method based on the BP neural network is characterized by comprising the following steps of:
step 1, collecting an existing effect diagram, and adding a label according to the client requirement corresponding to the effect diagram;
step 2, collecting color scheme data, extracting RGB values of main colors in an effect graph, and taking the RGB values of the main colors of the effect graph as training samples;
step 3, building a BI color matching model based on a BP neural network, then using the label of the effect diagram in the step 1 as input data of the BI color matching model, and using the RGB value of main color matching in the effect diagram in the step 2 as output data of the BI color matching model to train the BI color matching model;
step 4, verifying whether the output result of the BI color matching model has practicability,
a) The output result has practicability, and the BI color matching model is qualified;
b) If the output result has no practicability, the BI color matching model is unqualified, and the step 5 is continuously executed;
and 5, optimizing the BI color matching model, inputting labels required by customers into the final optimized BI color matching model, and outputting RGB values of main color matching of the effect graph corresponding to the input labels by the final optimized BI color matching model.
2. The BI color matching method based on the BP neural network according to claim 1, wherein in step 1, an existing effect graph is collected, and a label is added according to a customer requirement corresponding to the effect graph, and the specific operation comprises:
step 1.1, collecting an existing effect diagram, and analyzing client attributes corresponding to the effect diagram;
and 1.2, adding labels to the effect graph according to the analysis result, wherein the labels comprise customer attribute labels, style labels and main tone labels, and the main tone labels comprise at least three colors.
3. The BI color matching method based on the BP neural network according to claim 2, wherein in step 2, color matching scheme data is collected, and RGB values of main color matching in the effect map are extracted, which specifically comprises:
and analyzing the effect graph by using Photoshop or other drawing tools, and sequentially extracting RGB values of at least three colors with the duty ratio exceeding a set threshold value in the effect graph.
4. The method for BI color matching based on BP neural network according to claim 3, wherein in step 3, the specific process of building and training the BI color matching model comprises:
step 3.1, building a BI color matching model based on a three-layer structure of an input layer, an hidden layer and an output layer of the BP neural network;
step 3.2, introducing a training algorithm into the BI color matching model established in the step 3.1, inputting the label of the effect diagram in the step 1 into the BI color matching model, and outputting RGB values of main color matching in the effect diagram by the BI color matching model;
step 3.3, comparing the RGB value of the main color matching in the extraction effect diagram of the step 2 with the output result of the BI color matching model of the step 3.2, and adjusting the BI color matching model according to the comparison result;
and 3.4, setting the comparison similarity between the extracted RGB values and the output results, and circularly executing the steps 3.2-3.3 based on different effect graphs until the comparison results exceed the set comparison similarity, so as to complete training of the BI color matching model.
5. The BI color matching method based on the BP neural network according to claim 4, wherein in the process of building and training the BI color matching model, an hidden layer of the BP neural network is a single hidden layer, the number of hidden layer nodes is set to be 7, and an introduced training algorithm is set to be a Bayesian regularization algorithm.
6. The method for BI color matching based on BP neural network according to claim 4, wherein in step 5, optimizing the BI color matching model specifically comprises:
step 5.1, expanding a training sample, and optimizing the training sample;
step 5.2, adjusting the hidden layer number and the hidden layer node number of the BI color matching model just established by a trial and error method, and optimizing the hidden layer number and the hidden layer node number;
step 5.3, optimizing the selection of an initial threshold and a weight of the BI color matching model which is just built by using a particle swarm optimization algorithm;
step 5.4, replacing the activation function between the input layer and the hidden layer and the activation function between the hidden layers with a log-sigmoid function, and optimizing the activation function;
and 5.5, adding 'stop in advance' setting in the BI color matching model just established, and preventing the BP neural network and training samples from being excessively fitted and the generalization prediction capability from being deteriorated.
7. A BI color matching system based on a BP neural network, comprising:
the collection marking module is used for collecting the existing effect graph, adding labels according to the client requirements corresponding to the effect graph, and taking the labels of the effect graph as training samples;
the collecting and extracting module is used for collecting color scheme data and extracting RGB values of main colors in the effect graph, and the RGB values of the main colors of the effect graph are used as training samples;
the construction module is used for constructing a BI color matching model based on the BP neural network;
the training module is used for taking the effect graph label of the collecting and marking module as input data of the BI color matching model, taking the RGB value of main color matching in the effect graph of the collecting and extracting module as output data of the BI color matching model, and training the BI color matching model;
the input/output module is used for inputting the effect icon into the BI color matching model and further sending the output result of the BI color matching model to the setting verification module;
the verification module is used for setting a practical threshold value of the BI color matching model, and verifying whether the output result of the input/output module meets the practical threshold value based on the RGB value of the main color matching of the acquisition and extraction module extraction effect diagram;
and the optimization module is used for optimizing the BI color matching model to obtain a final optimized BI color matching model when the verification result of the set verification module has no practicability, and the output result of the final optimized BI color matching model meets the set threshold.
8. The BP neural network-based BI color matching system of claim 7, wherein the collection labeling module adds labels to the effect map by:
collecting the existing effect graph, and analyzing the client attribute corresponding to the effect graph;
adding labels to the effect graph according to the analysis result, wherein the labels comprise customer attribute labels, style labels and dominant hue labels, and the dominant hue labels comprise at least three colors;
and then, the acquisition and extraction module analyzes the effect graph by using a Photoshop or other drawing tools, and sequentially extracts RGB values of at least three colors with the duty ratio exceeding a set threshold value in the effect graph.
9. The BI color matching system based on the BP neural network according to claim 8, wherein the construction module establishes a BI color matching model based on a three-layer structure of an input layer, an hidden layer and an output layer of the BP neural network;
the process of training the BI color matching model by the training module comprises the following steps:
firstly, introducing a training algorithm into a BI color matching model, inputting a label of an effect diagram into the BI color matching model, and outputting RGB values of main color matching in the effect diagram by the BI color matching model;
secondly, comparing the RGB values of the main color matching in the extraction effect diagram of the acquisition and extraction module with the output result of the BI color matching model, and adjusting the BI color matching model according to the comparison result;
finally, training of the BI color matching model is completed when the comparison result meets the threshold set by the set verification module.
10. The BI color matching system based on the BP neural network of claim 7, wherein the optimizing module optimizes the BI color matching model by:
expanding a training sample, and optimizing the training sample;
adjusting the hidden layer number and the hidden layer node number of the BI color matching model by a trial and error method, and optimizing the hidden layer number and the hidden layer node number;
optimizing the selection of an initial threshold value and a weight of the BI color matching model by using a particle swarm optimization algorithm;
replacing an activation function between the input layer and the hidden layer and an activation function between the hidden layer with a log-sigmoid function to optimize the activation function;
and adding 'stop in advance' setting when the construction module builds the BI color matching model, so as to prevent the BP neural network and training samples from being excessively fitted and the generalization prediction capability from being deteriorated.
Priority Applications (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN201911365727.9A CN111080739B (en) | 2019-12-26 | 2019-12-26 | BI color matching method and system based on BP neural network |
Applications Claiming Priority (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN201911365727.9A CN111080739B (en) | 2019-12-26 | 2019-12-26 | BI color matching method and system based on BP neural network |
Publications (2)
| Publication Number | Publication Date |
|---|---|
| CN111080739A CN111080739A (en) | 2020-04-28 |
| CN111080739B true CN111080739B (en) | 2023-05-09 |
Family
ID=70318038
Family Applications (1)
| Application Number | Title | Priority Date | Filing Date |
|---|---|---|---|
| CN201911365727.9A Active CN111080739B (en) | 2019-12-26 | 2019-12-26 | BI color matching method and system based on BP neural network |
Country Status (1)
| Country | Link |
|---|---|
| CN (1) | CN111080739B (en) |
Families Citing this family (1)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN113345053B (en) * | 2021-06-30 | 2023-12-26 | 北京华录新媒信息技术有限公司 | Intelligent color matching method and system |
Citations (3)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| US5590251A (en) * | 1993-08-06 | 1996-12-31 | Toyota Jidosha Kabushiki Kaisha | Color reproducing device for reproducing matched colors and an outputting device for outputting information for reproducing a color of a coated surface |
| CN104504716A (en) * | 2014-12-31 | 2015-04-08 | 广州厚邦木业制造有限公司 | Solid wood floor paint automatic color blending method based on BP neural network |
| CN104700153A (en) * | 2014-12-05 | 2015-06-10 | 江南大学 | PH (potential of hydrogen) value predicting method of BP (back propagation) neutral network based on simulated annealing optimization |
-
2019
- 2019-12-26 CN CN201911365727.9A patent/CN111080739B/en active Active
Patent Citations (3)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| US5590251A (en) * | 1993-08-06 | 1996-12-31 | Toyota Jidosha Kabushiki Kaisha | Color reproducing device for reproducing matched colors and an outputting device for outputting information for reproducing a color of a coated surface |
| CN104700153A (en) * | 2014-12-05 | 2015-06-10 | 江南大学 | PH (potential of hydrogen) value predicting method of BP (back propagation) neutral network based on simulated annealing optimization |
| CN104504716A (en) * | 2014-12-31 | 2015-04-08 | 广州厚邦木业制造有限公司 | Solid wood floor paint automatic color blending method based on BP neural network |
Non-Patent Citations (1)
| Title |
|---|
| 王莹 ; 张秉森 ; .基于OWO-HWO算法的MLP网络在织物染色配色中的应用.系统仿真技术.2007,(02),全文. * |
Also Published As
| Publication number | Publication date |
|---|---|
| CN111080739A (en) | 2020-04-28 |
Similar Documents
| Publication | Publication Date | Title |
|---|---|---|
| CN109767255B (en) | Method for realizing intelligent operation and accurate marketing through big data modeling | |
| CN107506822B (en) | Deep neural network method based on space fusion pooling | |
| CN110458077B (en) | Vehicle color identification method and system | |
| CN109697233A (en) | A kind of knowledge mapping system building method | |
| CN107368700A (en) | Based on the microbial diversity interaction analysis system and method for calculating cloud platform | |
| CN106294738B (en) | A kind of Intelligent household scene configuration method | |
| CN106709017A (en) | Big data-based aid decision making method | |
| CN110489749B (en) | Business process optimization method of intelligent office automation system | |
| CN104679827A (en) | Big data-based public information association method and mining engine | |
| CN107392982A (en) | Photographing On-line method, apparatus and system | |
| CN109492750A (en) | A kind of zero sample image classification method and system based on convolutional neural networks and factor Spaces | |
| CN109460354A (en) | A method of test case reduction is carried out based on RDF reasoning | |
| CN113282612A (en) | Author conference recommendation method based on scientific cooperation heterogeneous network analysis | |
| CN111651566B (en) | Multi-task small sample learning-based referee document dispute focus extraction method | |
| CN111967930A (en) | Clothing style recognition recommendation method based on multi-network fusion | |
| CN110245984A (en) | A kind of shopping at network behavior analysis method and system based on causal inference | |
| CN111080739B (en) | BI color matching method and system based on BP neural network | |
| WO2023155508A1 (en) | Graph convolutional neural network and knowledge base-based paper correlation analysis method | |
| CN112365139A (en) | Crowd danger degree analysis method under graph convolution neural network | |
| CN103929499B (en) | A kind of Internet of Things isomery index identification method and system | |
| CN102799667B (en) | Hierarchical clustering method based on asymmetric distance | |
| CN110245310A (en) | A kind of behavior analysis method of object, device and storage medium | |
| CN117237559A (en) | Digital twin city-oriented three-dimensional model data intelligent analysis method and system | |
| CN112818229A (en) | Ornament recommendation method, system, device and medium based on home space | |
| CN112765313A (en) | False information detection method based on original text and comment information analysis algorithm |
Legal Events
| Date | Code | Title | Description |
|---|---|---|---|
| PB01 | Publication | ||
| PB01 | Publication | ||
| SE01 | Entry into force of request for substantive examination | ||
| SE01 | Entry into force of request for substantive examination | ||
| TA01 | Transfer of patent application right | ||
| TA01 | Transfer of patent application right |
Effective date of registration: 20230413 Address after: No. 1036, Shandong high tech Zone wave road, Ji'nan, Shandong Applicant after: Inspur Genersoft Co.,Ltd. Address before: 250100 No. 2877 Kehang Road, Sun Village Town, Jinan High-tech District, Shandong Province Applicant before: SHANDONG INSPUR GENESOFT INFORMATION TECHNOLOGY Co.,Ltd. |
|
| GR01 | Patent grant | ||
| GR01 | Patent grant |