Brightness prediction method and device, electronic equipment and storage medium

Abstract

The invention relates to a brightness prediction method and device, electronic equipment and a storage medium, and the brightness prediction method comprises the steps: obtaining backlight module configuration data corresponding to a to-be-predicted backlight module, extracting to-be-predicted factor information from the backlight module configuration data, and predicting the to-be-predicted factor information through a brightness prediction model. According to the method and the device, the brightness prediction information corresponding to the to-be-predicted backlight module is obtained, so that the brightness prediction model can be used for performing brightness prediction on the backlight module, the problems of long period and low efficiency in the design process of the backlight module in the prior art are solved, the design period of the backlight module can be effectively shortened, and the efficiency can be improved.

Description

Technical Field

[0001] The present invention relates to the field of computer technology, and in particular to a brightness prediction method, device, electronic device and storage medium. Background Art

[0002] With the development of society and the continuous improvement of people's living standards, display devices have gradually penetrated into all aspects of life. Among the various parameters of display devices, the brightness parameter is a particularly important indicator. In the design of the backlight module, engineers need to evaluate and design a preliminary optical solution based on the brightness indicator. In the absence of similar reference solutions, it is difficult to accurately calculate the actual brightness of the optical solution and give an accurate cost estimate in the early stage of product design, which requires repeated design and production in the future, resulting in long backlight module design cycle and low efficiency. Summary of the invention

[0003] The present application provides a brightness prediction method, device, electronic device and storage medium to solve the problems of long cycle and low efficiency in the backlight module design process in the prior art.

[0004] In a first aspect, the present application provides a brightness prediction method, comprising:

[0005] Obtaining backlight module configuration data corresponding to the backlight module to be predicted;

[0006] Extracting information of factors to be predicted from the backlight module configuration data;

[0007] The to-be-predicted factor information is predicted by a brightness prediction model to obtain brightness prediction information corresponding to the to-be-predicted backlight module.

[0008] Optionally, predicting the to-be-predicted factor information by using a brightness prediction model includes:

[0009] Get the preset backlight brightness data;

[0010] Extracting brightness factor information from the backlight brightness data;

[0011] Using the brightness factor information to train a preset model to obtain model error information;

[0012] adjusting the brightness factor information according to the model error information to obtain target brightness factor information;

[0013] The preset model is trained using the target brightness factor information to obtain the brightness prediction model;

[0014] The to-be-predicted factor information is predicted using the brightness prediction model.

[0015] Optionally, the using the first sample training set and the first sample test set to train a preset model to obtain model error information includes:

[0016] Using the first sample training set to train a preset model to obtain a first training error value;

[0017] When the first training error value falls within a preset training error value range, determining the prediction model as the first model;

[0018] Using the first sample test set to perform an error test on the first model to obtain a first test error value;

[0019] The model error information is generated according to the first test error value.

[0020] Optionally, the adjusting the brightness factor information according to the model error information to obtain target brightness factor information includes:

[0021] extracting a first test error value from the model error information;

[0022] Determining whether the first test error value falls within a preset test error value range;

[0023] When the first test error value does not fall within the preset test error value range, the brightness factor information is subjected to dimensionality reduction adjustment to obtain target brightness factor information.

[0024] Optionally, performing dimensionality reduction adjustment on the brightness factor information to obtain target brightness factor information includes:

[0025] Determine the number of primary factors and the number of secondary factors according to the brightness factor information, wherein the number of primary factors is the number of primary factor information included in the brightness factor information, and the number of secondary factors is the number of secondary factor information included in the brightness factor information;

[0026] When the number of the secondary factors is not zero, at least one secondary factor information in the brightness factor information is removed to obtain the target brightness factor information;

[0027] When the number of the secondary factors is zero, at least one primary factor information in the brightness factor information is removed to obtain the target brightness factor information.

[0028] Optionally, the adopting the target brightness factor information to train the preset model to obtain a brightness prediction model includes:

[0029] Generate a second sample training set and a second sample test set according to the target brightness factor information;

[0030] Using the second sample training set to train the preset model to obtain a second training error value;

[0031] When the second training error value falls within a preset training error value range, determining the prediction model as a second model;

[0032] Using the second sample test set to perform an error test on the second model to obtain a second test error value;

[0033] When the second test error value is within the preset test error value range, the second model is determined as the brightness prediction model.

[0034] Optionally, the adopting the target brightness factor information to train the preset model to obtain a brightness prediction model further includes:

[0035] When the second test error value does not fall within the preset test error value range, updating the model error information according to the second test error value to obtain updated model error information;

[0036] Based on the updated model error information, return to the step of adjusting the brightness factor information according to the model error information.

[0037] Optionally, also include:

[0038] Obtaining backlight module configuration data corresponding to the backlight module to be predicted;

[0039] Extracting information of factors to be predicted from the backlight module configuration data;

[0040] The brightness prediction model is used to predict the factor information to be predicted, so as to obtain brightness prediction information corresponding to the backlight module to be predicted.

[0041] In a second aspect, the present application provides a brightness prediction device, comprising:

[0042] An acquisition module, used to acquire preset backlight brightness data;

[0043] An extraction module, used for extracting brightness factor information from the backlight brightness data;

[0044] A first training module is used to train a preset model using the brightness factor information to obtain model error information;

[0045] An adjustment module, used for adjusting the brightness factor information according to the model error information to obtain target brightness factor information;

[0046] The second training module is used to train the preset model using the target brightness factor information to obtain a brightness prediction model.

[0047] In a third aspect, an electronic device is provided, comprising a processor, a communication interface, a memory and a communication bus, wherein the processor, the communication interface and the memory communicate with each other via the communication bus;

[0048] Memory, used to store computer programs;

[0049] The processor is used to implement the brightness prediction method described in any one of the first aspects when executing the program stored in the memory.

[0050] In a fourth aspect, a computer-readable storage medium is provided, on which a computer program is stored. When the computer program is executed by a processor, the brightness prediction method as described in any one of the first aspects is implemented.

[0051] The embodiment of the present application obtains the backlight module configuration data corresponding to the backlight module to be predicted, extracts the to-be-predicted factor information from the backlight module configuration data, and then predicts the to-be-predicted factor information through a brightness prediction model to obtain the brightness prediction information corresponding to the backlight module to be predicted. The brightness prediction model can then be used to predict the brightness of the backlight module, thereby solving the problems of long cycle and low efficiency in the backlight module design process in the existing related technologies, and can effectively shorten the design cycle of the backlight module and improve efficiency. BRIEF DESCRIPTION OF THE DRAWINGS

[0052] Figure 1 A schematic diagram of a flow chart of a brightness prediction method provided in an embodiment of the present application;

[0053] Figure 2 Another schematic diagram of a brightness prediction method provided in an embodiment of the present application;

[0054] Figure 3 A schematic diagram of an application scenario of a brightness prediction method provided in an embodiment of the present application;

[0055] Figure 4 A schematic diagram of a training scene for a brightness prediction method provided in an embodiment of the present application;

[0056] Figure 5 A schematic diagram of another application scenario of a brightness prediction method provided in an embodiment of the present application;

[0057] Figure 6A schematic diagram of the structure of a brightness prediction device provided in an embodiment of the present application;

[0058] Figure 7 A schematic diagram of the structure of an electronic device provided in an embodiment of the present application. DETAILED DESCRIPTION

[0059] The following will describe the embodiments of the present invention with reference to the accompanying drawings and preferred embodiments. Those skilled in the art can easily understand other advantages and effects of the present invention from the contents disclosed in this specification. The present invention can also be implemented or applied through other different specific embodiments, and the details in this specification can also be modified or changed in various ways based on different viewpoints and applications without departing from the spirit of the present invention. It should be understood that the preferred embodiments are only for illustrating the present invention, not for limiting the scope of protection of the present invention.

[0060] In the process of backlight module design in the existing related technology, in order to predict the brightness, two methods are usually adopted. One is to actually assemble the backlight module according to the plan and test it. Although this method can obtain accurate brightness data, it takes a long time and a large cost to produce and assemble the backlight module, that is, there is a problem of lengthening the product design cycle and reducing efficiency; the other method is to use optical simulation software to model and simulate the backlight module. Although this method can avoid the problem of a long time period for actually assembling the backlight module, a large amount of accurate data is required for the specific modeling, and the material-related characteristic parameters need to be simulated. It is difficult to implement and its simulation accuracy will fluctuate according to different software, that is, there is still a problem of low efficiency.

[0061] In order to solve the problems of long cycle and low efficiency in the backlight module design process of the existing related technologies, the present application provides a brightness prediction method, device, electronic device and storage medium, by obtaining the backlight module configuration data corresponding to the backlight module to be predicted, and extracting the factor information to be predicted from the backlight module configuration data, and then predicting the factor information to be predicted by a brightness prediction model to obtain the brightness prediction information corresponding to the backlight module to be predicted, so that the brightness prediction model can be used to predict the brightness of the backlight module, thereby solving the problems of long cycle and low efficiency in the backlight module design process of the existing related technologies, and can effectively shorten the design cycle of the backlight module and improve efficiency.

[0062] Figure 1A flow chart of a brightness prediction method provided for an embodiment of the present application. The method can be applied to one or more electronic devices such as a computer, a client, and a server. In addition, the execution subject of the method can be hardware or software. When the above-mentioned execution subject is hardware, the execution subject can be one or more of the above-mentioned electronic devices. For example, a single electronic device can execute the method, or a plurality of electronic devices can cooperate with each other to execute the method. When the above-mentioned execution subject is software, the method can be implemented as a plurality of software or software modules, or as a single software or software module. No specific limitation is made herein.

[0063] like Figure 1 As shown, a brightness prediction method provided in an embodiment of the present application may specifically include the following steps:

[0064] Step S110: acquiring backlight module configuration data corresponding to the backlight module to be predicted;

[0065] Step S120: extracting information of factors to be predicted from the backlight module configuration data;

[0066] Step S130: predicting the to-be-predicted factor information using a brightness prediction model to obtain brightness prediction information corresponding to the to-be-predicted backlight module.

[0067] This embodiment can obtain the backlight module configuration data corresponding to the backlight module to be predicted according to the specific scenario and prediction requirements, wherein the backlight module to be predicted can represent the backlight module that needs to be predicted for brightness prediction, and the backlight module can be a module in the design stage or an actual module. This embodiment does not make specific limitations on this. The backlight module configuration data can represent the specific configuration data of the backlight module to be predicted, such as the size, hardware configuration, device type, etc. of the backlight module; then the information of the factors to be predicted can be extracted from the backlight module configuration data. The information of the factors to be predicted can be the information of the factors that affect the brightness of the backlight module to be predicted. The elements that constitute the information of the factors to be predicted, for example, can include but are not limited to at least one of the following: LED chip size, LED quantity, LED horizontal spacing, LED vertical spacing, LED total power, lens, OD value, diaphragm type, O / C transmittance, blue light wavelength, spectrum type, LED driving current, reflector, diffuser plate transmittance, diffuser plate thickness, etc.; thereby, the brightness prediction model can be used to predict the information of the factors to be predicted, and the brightness prediction information corresponding to the backlight module to be predicted can be obtained. The specific prediction method can be to input the information of the factors to be predicted into the brightness prediction model, and obtain the output information of the brightness prediction model for the information of the factors to be predicted, and determine the brightness prediction information corresponding to the backlight module to be predicted based on the output information. The brightness prediction information can represent the module backlight brightness corresponding to the backlight module to be predicted, so that the brightness prediction model can be used to predict the brightness of the backlight module, thereby solving the problems of long cycle and low efficiency in the design process of the backlight module in the existing related technology, and can effectively shorten the design cycle of the backlight module and improve the efficiency.

[0068] like Figure 2 As shown, in an optional embodiment of the present application, step S130: predicting the to-be-predicted factor information by using a brightness prediction model may specifically include the following sub-steps:

[0069] Step S210: Obtaining preset backlight brightness data.

[0070] Among them, the preset backlight brightness data can be represented as the relevant brightness data contained in the backlight module that has been tested or produced in advance. Specifically, the way to obtain the backlight brightness data can be to obtain it from a preset database, or to obtain it based on the user's input operation. Of course, other methods can also be used to obtain it, and this is not specifically limited in this embodiment.

[0071] Step S220: extracting brightness factor information from the backlight brightness data.

[0072] Specifically, after obtaining the backlight brightness data, the brightness factor information can be extracted from the backlight brightness data. The brightness factor information can be used to represent the factors that affect the brightness of the backlight module, including but not limited to LED chip size, blue light wavelength, spectrum type, LED quantity, LED horizontal spacing, LED vertical spacing, LED driving current, lens, OD value, reflector, diffuser plate transmittance, diffuser plate thickness, diaphragm type, O / C transmittance, O / C size and other information. For details, please refer to Figure 3 As shown, this embodiment does not make any specific limitation to this.

[0073] It should be noted that the brightness factor information in this embodiment may include continuous features and discrete features. Specifically, in the process of extracting brightness factor information from backlight brightness data, the following sub-steps may also be included: extracting brightness features from backlight brightness data; determining the feature type of the brightness features; when the feature type is a continuous type, obtaining a first feature parameter corresponding to the brightness feature; when the feature type is a discrete type, encoding the brightness features to obtain a second feature parameter corresponding to the brightness feature; and generating brightness factor information based on the first feature parameter and the second feature parameter.

[0074] In the present embodiment, in the process of extracting brightness factor information from backlight brightness data, first, brightness features can be extracted from the backlight brightness data. The brightness features can represent features that affect the brightness of the backlight module, so that the feature type of the brightness features can be determined. The feature type represents the type to which the brightness feature specifically corresponds, such as a continuous type corresponding to a continuous feature and a discrete type corresponding to a discrete feature. In the case where the feature type is a continuous type, the first feature parameter corresponding to the brightness feature can be directly obtained. In the case where the feature type is a discrete type, the brightness feature can be encoded to obtain a second feature parameter corresponding to the brightness feature. The specific encoding method can be to convert the brightness feature into its corresponding encoding parameter according to a preset rule, and then generate the brightness factor information based on the first feature parameter and the second feature parameter. Specifically, the first feature parameter and the second feature parameter can be integrated to generate the brightness factor information.

[0075] In one example, for example, the brightness feature D is extracted from the backlight brightness data, D={(LED chip size 1, blue light wavelength 1, spectrum type 1, LED quantity 1, LED lateral spacing 1, LED longitudinal spacing 1, LED driving current 1, lens 1, OD value 1, reflector 1, diffuser plate transmittance 1, diffuser plate thickness 1, diaphragm type 1, O / C transmittance 1, O / C size 1), (LED chip size 2, blue light wavelength 2, spectrum type 2, LED quantity 2, LED lateral spacing 2, LED longitudinal spacing 2, LED driving current 2, lens 2, OD value 2, reflector 2, diffuser plate transmittance 2, diffuser plate thickness 2, diaphragm type 2, O / C transmittance 2, O / C size 2)...(LED chip size m, blue light wavelength m, spectrum type m, LED quantity m, LED lateral spacing m, LED longitudinal spacing m, LED driving current m, lens m, OD value m, reflector m, diffuser plate transmittance m, diffuser plate thickness m , diaphragm type m, O / C transmittance m, O / C size m)}, where 1-m is the brightness factor information extracted from each backlight brightness data, that is, the D set contains m brightness factor information; specifically, the feature type of "LED chip size, LED quantity, LED horizontal spacing, LED vertical spacing, LED total power, lens, OD value, diaphragm type, O / C transmittance" in each brightness factor information is a continuous type. When extracting this part of the features, the first feature parameter corresponding to the brightness feature of this part can be directly obtained. The first feature parameter is the specific parameter corresponding to the brightness feature, such as a specific size value, quantity value, spacing value, etc., and the feature type of "blue light wavelength, spectrum type, LED driving current, reflector, diffuser plate transmittance, diffuser plate thickness, O / C size" is a discrete type. At this time, the brightness feature of this part can be encoded to obtain the second feature parameter corresponding to the brightness feature. The specific encoding method can refer to the following example:

[0076] The wavelengths of blue light are divided into (short wave, long wave), and the corresponding codes are (1, 0), (0, 1)

[0077] The spectrum types are divided into (YAG spectrum, KSF spectrum, QD spectrum), and the corresponding codes are (1, 0, 0).

[0078] (0, 1, 0), (0, 0, 1)

[0079] Lenses are divided into (no lens, refractive lens, reflective lens), and the corresponding codes are (1, 0, 0), (0, 1, 0), (0, 0, 1)

[0080] Reflectors are divided into (glossy reflectors, coated reflectors), and the corresponding codes are (1, 0), (0, 1)

[0081] The diaphragm types are (no diaphragm, single expansion, double expansion, DOP, MOP, P+DOP, DPP, CPP), and the corresponding ordinal codes are (1.0, 1.27, 1.37, 1.63, 1.79, 2.1, 2.2, 3.3).

[0082] The second characteristic parameter is a parameter obtained by encoding the brightness feature. Then, the brightness factor information can be generated based on the first characteristic parameter and the second characteristic parameter. At this time, the brightness factor information x=(x1, x2,……, x i ), where each x is a first characteristic parameter or a second characteristic parameter, and there are i characteristic parameters in total. Of course, the above is only an example for illustration, and in specific implementation, adaptive adjustments can be made according to scenarios and requirements, and this embodiment does not make specific limitations on this.

[0083] In addition, before extracting the brightness factor information from the backlight brightness data, the brightness value can also be extracted from the backlight brightness data, and it is determined whether the brightness value in the backlight brightness data belongs to a preset brightness value range. When the brightness value does not belong to the preset brightness value range, the backlight brightness data is discarded. When the brightness value belongs to the preset brightness value range, the step of extracting the brightness factor information from the backlight brightness data can be performed; wherein the brightness value represents a specific brightness value corresponding to the backlight brightness data, such as a color coordinate; the preset brightness value range can represent a brightness value range configured in advance according to requirements, such as a color coordinate range within (0.28±0.15, 0.29±0.15). Specifically, when the brightness value in the backlight brightness data does not belong to the range within (0.28±0.15, 0.29±0.15), the backlight brightness data can be discarded, and when the brightness value in the backlight brightness data belongs to the range within (0.28±0.15, 0.29±0.15), the step of extracting the brightness factor information from the backlight brightness data can be performed. Of course, the preset brightness value range can be adaptively adjusted according to the scene and needs, and this embodiment does not make any specific limitation to this.

[0084] Step S230: using the brightness factor information to train the preset model to obtain model error information.

[0085] Specifically, after obtaining the brightness factor information, the brightness factor information can be used to train the preset model to obtain model error information, where the preset model can be a regression model, a neural network model, etc., and the model error information can represent the error information after the preset model is trained using the brightness factor information.

[0086] Specifically, in this embodiment, the preset model can be one or more models, and the training of each preset model and the execution of subsequent steps can be independent. For example, the preset model is a decision tree regression model, a support vector regression model, etc.

[0087] Step S240: adjusting the brightness factor information according to the model error information to obtain target brightness factor information.

[0088] Specifically, after obtaining the model error information, the brightness factor information can be adjusted according to the model error information to obtain the target brightness factor information, wherein the specific adjustment method can be to reduce the dimension of the model error information, that is, to remove some factor information in the model error information, so that the adjusted brightness factor information can be determined as the target brightness factor information.

[0089] It should be noted that after obtaining the model error information, it can be determined whether the error value corresponding to the model error information belongs to the preset error value range. If the error value belongs to the preset error value range, it means that the preset model training has met expectations. At this time, the preset model can be directly determined as the brightness prediction model. If the error value does not belong to the preset error value range, it means that the preset model training has not met expectations. At this time, the brightness factor information can be adjusted according to the model error information to obtain the target brightness factor information, and the target brightness factor information can be used to train the preset model again to obtain the brightness prediction model, that is, execute step S250.

[0090] Step S250: using the target brightness factor information to train the preset model to obtain a brightness prediction model.

[0091] Specifically, after obtaining the target brightness factor information, the preset model may be trained again using the target brightness factor information to obtain a brightness prediction model, which may represent a model for predicting the brightness of the backlight module.

[0092] It should be noted that in the process of using the target brightness factor information to train the preset model, model error information can still be collected, and it is determined whether the error value corresponding to the model error information belongs to the preset error value range. If the error value belongs to the preset error value range, it means that the preset model training has met expectations. At this time, the preset model can be directly determined as a brightness prediction model. If the error value does not belong to the preset error value range, it means that the preset model training has not met expectations. At this time, the target brightness factor information can be updated and adjusted again based on the model error information to obtain updated target brightness factor information, and the preset model can be trained again using the updated target brightness factor information to obtain a brightness prediction model, that is, return to execute step S250, and repeat this cycle until the error value corresponding to the model error information belongs to the preset error value range.

[0093] It can be seen that this embodiment obtains the backlight module configuration data corresponding to the backlight module to be predicted, extracts the to-be-predicted factor information from the backlight module configuration data, and then predicts the to-be-predicted factor information through a brightness prediction model to obtain the brightness prediction information corresponding to the backlight module to be predicted. The brightness prediction model can then be used to predict the brightness of the backlight module, thereby solving the problems of long cycle and low efficiency in the backlight module design process of the existing related technologies, and can effectively shorten the design cycle of the backlight module and improve efficiency.

[0094] In an optional embodiment of the present application, the preset model is trained using brightness factor information to obtain model error information, which may specifically include the following sub-steps: based on the brightness factor information, generating a sample data set corresponding to the backlight brightness data; performing sample division on the sample data set to obtain a first sample training set and a first sample test set; using the first sample training set and the first sample test set to train the preset model to obtain model error information.

[0095] In this embodiment, in the process of training the preset model using brightness factor information, a sample data set corresponding to the backlight brightness data can be generated based on the brightness factor information. The sample data set represents a data set used to train the preset model, wherein the sample data set may include at least two sets of brightness factor information, so that the sample data set can be divided into samples to obtain a first sample training set and a first sample test set. The first sample training set represents a training set used in the training stage of the preset model, and the first sample test set can represent a test set used for testing the preset model after the training stage is completed. The first sample training set and the first sample test set can then be used to train the preset model to obtain model error information.

[0096] It should be noted that, in the present embodiment, when dividing the sample data set to obtain the first sample training set and the first sample test set, since the backlight brightness data contained in the sample data set is the relevant brightness data contained in the backlight module that has been tested or produced, the amount of backlight brightness data in the actual scene may not be large enough. In this case, the "self-service method" can be used for sample division. For example, for the brightness factor information corresponding to m backlight brightness data, a sample data set D corresponding to the backlight brightness data is generated. At this time, a sample data is randomly selected from the D set and put into the D' set each time, and then the sample data is returned to the D set so that it may still be sampled in the next sampling. After repeating this process m times, a data set D' containing m samples is obtained. In this way, the probability that a certain sample is never sampled in m samplings is Taking the limit of it gives:

[0097]

[0098] That is, about 36.8% of the samples in the initial sample D do not appear in the sampled data set D', so D' can be used as the first sample training set and D\D' as the first sample test set, where D\D' represents the sample data included in the D set but not in the D' set. At this time, for the scenario where the backlight brightness data of the backlight module is limited, the sample characteristics of the difficulty in effectively dividing the training set and the test set can be overcome. The self-service sampling method can be used to efficiently determine the experimental data set and improve the efficiency of model training.

[0099] In an optional embodiment of the present application, a first sample training set and a first sample test set are used to train a preset model to obtain model error information, which may specifically include the following sub-steps: training the preset model with the first sample training set to obtain a first training error value; when the first training error value falls within a preset training error value range, determining the prediction model as the first model; performing an error test on the first model with the first sample test set to obtain a first test error value; and generating model error information based on the first test error value.

[0100] In this embodiment, in the process of training the preset model using the first sample training set and the first sample test set, the first sample training set can be used to perform model training on the preset model to obtain a first training error value, which indicates the error value generated by the preset model during the training stage. Then, it can be determined whether the first training error value falls within the preset training error value range. If it does not fall within the preset training error value range, it means that the error of the preset model training does not meet expectations. At this time, the model parameters can be adjusted and the step of training the preset model using the first sample training set to obtain the first training error value can be performed again. When the first training error value falls within the preset training error value range, it means that the error of the preset model training meets expectations. At this time, the prediction model is determined to be the first model. Therefore, the first sample test set can be used to perform an error test on the first model to obtain a first test error value. The first test error value can indicate the error value generated by the first model during the error test process. Then, model error information can be generated based on the first test error value.

[0101] Specifically, in the process of generating model error information based on the first test error value, it can be determined whether the first test error value is within the preset test error value range. If it is not within the preset test error value range, it means that the error of the first model during the test process does not meet expectations. At this time, the model error information can be generated based on the first test error value and the step of adjusting the brightness factor information based on the model error information to obtain the target brightness factor information can be continued until the first test error value is within the preset test error value range. This means that the error of the first model during the test process meets expectations. At this time, there is no need to execute subsequent steps and the first model can be directly determined as a brightness prediction model.

[0102] In an example, there are differences in the training process of different models. The following example uses a decision tree regression model and a support vector regression model as examples. Figure 4 Said:

[0103] For the decision tree regression model, some of its structures are shown in the attached figure. Figure 5 As shown in the figure, the branching principle of the decision tree is the key to the decision tree model. For the regression tree predicting brightness, the output prediction value is a continuous value, and its heterogeneity measurement is calculated by variance. The variance of the regression tree node t is defined as

[0104]

[0105] Among them, y i (t) is the output variable value of the sample observation Xi in node t, is the average value of the output variable in node t. If t is the parent node, the branch is based on the weighted mean of the variance of t's left child node tleft and the variance of the right child node tright.

[0106]

[0107] In the process of using the first sample training set to train the preset model and obtain the first training error value, the variance weighted mean of the child nodes under each brightness influencing factor is recursively calculated, the minimum value is recorded, and the corresponding brightness influencing factor is used as the condition for this branch, and the cost complexity is used as the first training error value to decide whether to continue branching. in represents the number of leaf nodes of tree T, α is a complexity parameter, and its value is [0,∞], and R(T) is the mean square error MSE based on the test set; when the first training error value is within the preset training error value range, that is, the cost complexity is minimized, the prediction model is determined as the first model, that is, the result of the decision tree regression model is determined. At this time, the first sample test set is used to perform an error test on the first model to obtain a predicted value, and a first test error value, that is, MSE and variance, is calculated based on the predicted value, and it is determined whether the MSE and variance are within the preset test error value range. If they are not within the preset test error value range, it means that the error of the first model during the test process does not meet expectations. At this time, model error information can be generated based on the first test error value, and the brightness factor information can be adjusted based on the model error information to obtain the target brightness factor information. The step is performed until the first test error value is within the preset test error value range, indicating that the error of the first model test during the test process meets expectations. At this time, there is no need to perform subsequent steps, and the first model can be directly determined as the brightness prediction model.

[0108] For the support vector regression model, it analyzes the quantitative relationship between the N-dimensional influencing factors of the brightness sample data and the brightness output, and finds the maximum boundary regression plane to achieve robust prediction of new data. The training process aims to find a regression plane that can reasonably split the brightness sample data. If the brightness sample data is linearly separable in the existing dimensions, the basic form of the regression plane is as follows:

[0109]

[0110] The upper boundary is expressed as

[0111]

[0112] The lower boundary is expressed as

[0113]

[0114] where ω i is the normal vector, which determines the direction of the hyperplane, b is the displacement term, and ε is an adjustable parameter, which is called ε-band in support vector regression. The corresponding loss function is the ε-insensitive loss function, that is, if the residual e of the brightness data predicted by the model is different from the true value i =y i -f(x) is not greater than the pre-given adjustable parameter ε>0, it is not included in the loss function, so there is

[0115]

[0116] In order to reduce the impact of data noise, it is necessary to observe each brightness sample point X iSet a non-negative slack variable

[0117]

[0118] The slack variable of the brightness data point above the regression hyperplane is recorded as The slack variable at the lower point is denoted by ξ i , so that the sample observation point ξ outside the regression hyperplane i >0, point ξ within the ε-band i =0.

[0119] Furthermore, the regression hyperplane is obtained by solving the objective function under the constraints. According to the above description, for each sample X i , the following constraints need to be met:

[0120]

[0121] The objective function of support vector regression is set as

[0122]

[0123] The second term represents the loss of sample observations. Support vector regression hopes to find the regression hyperplane where the sum of these two terms is minimized. C is the loss penalty parameter, which can be adjusted according to the tolerance for prediction errors. The larger C is, the more complex the model will be. Like ε, cross-validation is required to determine the appropriate parameters.

[0124] According to the constraints and objective function, the solution of regression hyperplane coefficients is given by Lagrange multiplier method:

[0125]

[0126] And the regression prediction value of the sample observation point X0 is:

[0127]

[0128] Among them, there are L support vectors

[0129] If the brightness sample data is linearly inseparable in N-dimensional space, it can be made linearly separable in high-dimensional space through nonlinear transformation. That is, whether it is linearly separable is used as a criterion for judging whether the first training error value is within the preset training error value range. If it is linearly inseparable, it is determined that the first training error value does not fall within the preset training error value range, and a Gaussian kernel function is used for nonlinear transformation. If it is linearly separable, it is determined that the first training error value falls within the preset training error value range. The support vector regression model is determined as the first model, and the first test sample set can be input.

[0130] Here we use the Gaussian kernel function

[0131]

[0132] At this time, the regression prediction value of the sample observation point X0 is

[0133]

[0134] That is, the first model is tested using the first test sample set to obtain the test error and variance and determine whether the test error and variance are within the preset test error value range. If they are not within the preset test error value range, it means that the error of the first model during the test process does not meet expectations. At this time, based on the first test error value, model error information can be generated and the brightness factor information can be adjusted based on the model error information to obtain the target brightness factor information. Until the first test error value is within the preset test error value range, it means that the error of the first model during the test process meets expectations. At this time, there is no need to execute subsequent steps and the first model can be directly determined as a brightness prediction model.

[0135] In an optional embodiment of the present application, the brightness factor information is adjusted according to the model error information to obtain the target brightness factor information, which may specifically include the following sub-steps: extracting a first test error value from the model error information; determining whether the first test error value falls within a preset test error value range; and when the first test error value does not fall within the preset test error value range, performing dimensionality reduction adjustment on the brightness factor information to obtain the target brightness factor information.

[0136] In this embodiment, in the process of adjusting the brightness factor information according to the model error information to obtain the target brightness factor information, a first test error value can be extracted from the model error information. The first test error value can represent the error value generated by the preset model during the error test, and determine whether the first test error value falls within the preset test error value range. If it does not fall within the preset test error value range, it means that the error of the preset model during the test process does not meet expectations. At this time, the model error information can be generated based on the first test error value and the step of adjusting the brightness factor information according to the model error information to obtain the target brightness factor information is continued until the first test error value falls within the preset test error value range, indicating that the error of the first model test during the test process meets expectations. At this time, there is no need to execute subsequent steps, and the first model can be directly determined as the brightness prediction model. It should be noted that the preset model in this implementation can be the first model in the aforementioned embodiment.

[0137] In an optional embodiment of the present application, the brightness factor information is subjected to dimensionality reduction adjustment to obtain target brightness factor information, which may specifically include the following sub-steps: determining the number of main factors and the number of secondary factors based on the brightness factor information; when the number of secondary factors is not zero, removing at least one secondary factor information in the brightness factor information to obtain target brightness factor information; when the number of secondary factors is zero, removing at least one main factor information in the brightness factor information to obtain target brightness factor information.

[0138] In this embodiment, in the process of performing dimensionality reduction adjustment on the brightness factor information to obtain the target brightness factor information, the number of main factors and the number of secondary factors can be determined based on the brightness factor information, wherein the number of main factors is the number of main factor information included in the brightness factor information, and the number of secondary factors is the number of secondary factor information included in the brightness factor information, and the main factor information can be a pre-configured factor that mainly affects the brightness of the backlight module, and the secondary factor information can be a pre-configured factor that secondarily affects the brightness of the backlight module; then it can be determined whether the number of secondary factors is zero. When the number of secondary factors is not zero, at least one secondary factor information in the brightness factor information can be removed to obtain the target brightness factor information; when the number of secondary factors is zero, it means that all the secondary factor information has been removed through the previous steps, or there is no secondary factor information. At this time, at least one main factor information in the brightness factor information can be removed to obtain the target brightness factor information.

[0139] It can be seen that in the present embodiment, in the process of performing dimensionality reduction adjustment on the brightness factor information to obtain the target brightness factor information, the target brightness factor information is obtained by removing the secondary factor information or the main factor information in the brightness factor information, and thereby using the target brightness factor information after dimensionality reduction to again execute the step of using the target brightness factor information to train the preset model, which can reduce the complexity of the model when the error is large, thereby improving the model training effect and efficiency.

[0140] Specifically, when the brightness factor information includes LED chip size, LED quantity, LED lateral spacing, LED longitudinal spacing, LED total power, lens, OD value, diaphragm type, O / C transmittance, blue light wavelength, spectrum type, LED driving current, reflector, diffuser plate transmittance, and diffuser plate thickness, the main factor information can be LED chip size, LED quantity, LED lateral spacing, LED longitudinal spacing, LED total power, lens, OD value, diaphragm type, and O / C transmittance; the secondary factor information can be blue light wavelength, spectrum type, LED driving current, reflector, diffuser plate transmittance, and diffuser plate thickness; at this time, the number of main factors is 9 and the number of secondary factors is 6, that is, the number of secondary factors is not zero, and at this time it can be At least one secondary factor information in the brightness factor information is removed to obtain the target brightness factor information. The specific removal method may be to remove one of the six secondary factor information in turn to obtain seven target brightness factor information, that is, the six target brightness factor information is to remove one different secondary factor information from the brightness factor information to obtain the target brightness factor information. It may also be possible to remove two of the six secondary factor information in turn to obtain three target brightness factor information, that is, the three target brightness factor information is to remove two different secondary factor information from the brightness factor information to obtain the target brightness factor information. Of course, one or more removal methods may be used in specific implementation, and this embodiment does not specifically limit this.

[0141] In an optional embodiment of the present application, the preset model is trained using the target brightness factor information to obtain a brightness prediction model, which may specifically include the following sub-steps:

[0142] Based on the target brightness factor information, a second sample training set and a second sample test set are generated; the preset model is trained using the second sample training set to obtain a second training error value; when the second training error value falls within the preset training error value range, the prediction model is determined as the second model; the second model is error tested using the second sample test set to obtain a second test error value; when the second test error value falls within the preset test error value range, the second model is determined as the brightness prediction model.

[0143] In this embodiment, in the process of using the target brightness factor information to train the preset model to obtain the brightness prediction model, first, a second sample training set and a second sample test set can be generated based on the target brightness factor information, the first sample training set represents a training set used for re-training the preset model, and the first sample test set can represent a test set used for testing the preset model after completing the training stage again; then the second sample training set can be used to train the preset model to obtain a second training error value, and it is determined whether the second training error value falls within the preset training error value range. If the second training error value does not fall within the preset training error value range, it means that the error of the preset model training does not meet expectations, and the model can continue to be trained. After the parameters are adjusted, the step of using the second sample training set to train the preset model to obtain the second training error value is performed again, until the second training error value falls within the preset training error value range, indicating that the error of the preset model training meets expectations, and the prediction model is determined as the second model at this time; thereby, the second sample test set can be used to perform an error test on the second model to obtain a second test error value, the second test error value can represent the error value generated by the second model during the error testing process, and determine whether the second test error value falls within the preset test error value range. When the first test error value falls within the preset test error value range, it means that the error of the first model test during the test process meets expectations, and the second model can be directly determined as the brightness prediction model.

[0144] In an optional embodiment of the present application, the preset model is trained using the target brightness factor information to obtain a brightness prediction model, which may further include the following sub-steps: when the second test error value does not fall within the preset test error value range, the model error information is updated according to the second test error value to obtain updated model error information; based on the updated model error information, return to execute the step of adjusting the brightness factor information according to the model error information.

[0145] In this embodiment, when it is determined that the second test error value does not fall within the preset test error value range, it means that the error of the preset model during the test process does not meet expectations. At this time, the model error information can be updated according to the second test error value to obtain updated model error information, and based on the updated model error information, return to execute the step of adjusting the brightness factor information according to the model error information, until the second test error value falls within the preset test error value range, the second model is determined as the brightness prediction model.

[0146] like Figure 6 As shown, the present application also discloses an embodiment, providing a brightness prediction device, which may specifically include:

[0147] An acquisition module 610 is used to acquire backlight module configuration data corresponding to the backlight module to be predicted;

[0148] An extraction module 620, configured to extract information of factors to be predicted from the backlight module configuration data;

[0149] The prediction module 630 is used to predict the to-be-predicted factor information by using a brightness prediction model to obtain brightness prediction information corresponding to the to-be-predicted backlight module.

[0150] In an optional embodiment of the present application, the prediction module 630 may include:

[0151] An acquisition unit, used for acquiring preset backlight brightness data;

[0152] An extraction unit, used for extracting brightness factor information from the backlight brightness data;

[0153] A first training unit, used to train a preset model using the brightness factor information to obtain model error information;

[0154] an adjusting unit, configured to adjust the brightness factor information according to the model error information to obtain target brightness factor information;

[0155] The second training unit is used to train the preset model using the target brightness factor information to obtain a brightness prediction model.

[0156] The prediction unit is used to predict the information of the factor to be predicted by using a brightness prediction model.

[0157] In an optional embodiment of the present application, the first training unit may specifically include:

[0158] A first generating subunit, configured to generate a sample data set corresponding to the backlight brightness data based on the brightness factor information;

[0159] A sample division subunit, used to divide the sample data set into samples to obtain a first sample training set and a first sample test set;

[0160] The first training subunit is used to train a preset model using the first sample training set and the first sample test set to obtain model error information.

[0161] In an optional embodiment of the present application, the first training subunit may specifically include:

[0162] A second training subunit is used to train a preset model using the first sample training set to obtain a first training error value;

[0163] A first determining subunit, configured to determine the prediction model as a first model when the first training error value falls within a preset training error value range;

[0164] A third training subunit is used to perform an error test on the first model using the first sample test set to obtain a first test error value;

[0165] The second generating subunit is used to generate the model error information according to the first test error value.

[0166] In an optional embodiment of the present application, the adjustment unit 540 may specifically include:

[0167] an extraction subunit, configured to extract a first test error value from the model error information;

[0168] A second determining subunit, configured to determine whether the first test error value falls within a preset test error value range;

[0169] The dimension reduction adjustment subunit is used to perform dimension reduction adjustment on the brightness factor information to obtain target brightness factor information when the first test error value does not fall within the preset test error value range.

[0170] In an optional embodiment of the present application, the dimension reduction adjustment subunit may specifically include:

[0171] A third determining subunit is used to determine the number of primary factors and the number of secondary factors according to the brightness factor information, wherein the number of primary factors is the number of primary factor information included in the brightness factor information, and the number of secondary factors is the number of secondary factor information included in the brightness factor information;

[0172] A first removing subunit is used for removing at least one secondary factor information in the brightness factor information to obtain the target brightness factor information when the number of the secondary factors is not zero;

[0173] The second removing subunit is used to remove at least one main factor information in the brightness factor information when the number of the secondary factors is zero, so as to obtain the target brightness factor information.

[0174] In an optional embodiment of the present application, the second training unit 550 may specifically include:

[0175] A third generating subunit, used for generating a second sample training set and a second sample test set according to the target brightness factor information;

[0176] a fourth training subunit, configured to train the preset model using the second sample training set to obtain a second training error value;

[0177] A fourth determining subunit, configured to determine the prediction model as a second model when the second training error value is within a preset training error value range;

[0178] a fifth training subunit, configured to perform an error test on the second model using the second sample test set to obtain a second test error value;

[0179] The fifth determining subunit is configured to determine the second model as the brightness prediction model when the second test error value is within the preset test error value range.

[0180] In an optional embodiment of the present application, the second training unit may further include:

[0181] an updating subunit, configured to update the model error information according to the second test error value to obtain updated model error information when the second test error value does not fall within the preset test error value range;

[0182] The return execution subunit is used to return to the step of adjusting the brightness factor information according to the model error information based on the updated model error information.

[0183] The implementation process of the functions and effects of each module in the above-mentioned device is specifically described in the implementation process of the corresponding steps in the above-mentioned method, which will not be repeated here.

[0184] like Figure 7 As shown, an embodiment of the present application provides an electronic device, including a processor 710, a communication interface 720, a memory 730 and a communication bus 740, wherein the processor 710, the communication interface 720, and the memory 730 communicate with each other through the communication bus 740.

[0185] Memory 730, for storing computer programs;

[0186] In one embodiment of the present application, the processor 710, when used to execute the program stored in the memory 730, implements the brightness prediction method provided by any of the aforementioned method embodiments, by obtaining the backlight module configuration data corresponding to the backlight module to be predicted, and extracting the factor information to be predicted from the backlight module configuration data, and then predicting the factor information to be predicted through a brightness prediction model to obtain the brightness prediction information corresponding to the backlight module to be predicted, so that the brightness prediction model can be used to predict the brightness of the backlight module, thereby solving the problems of long cycle and low efficiency in the backlight module design process in the existing related technologies, and can effectively shorten the design cycle of the backlight module and improve efficiency.

[0187] The embodiment of the present application also provides a computer-readable storage medium having a computer program stored thereon. When the computer program is executed by a processor, the steps of the brightness prediction method provided in any of the aforementioned method embodiments are implemented, by obtaining the backlight module configuration data corresponding to the backlight module to be predicted, and extracting the factor information to be predicted from the backlight module configuration data, and then predicting the factor information to be predicted by a brightness prediction model to obtain the brightness prediction information corresponding to the backlight module to be predicted, so that the brightness prediction model can be used to predict the brightness of the backlight module, thereby solving the problems of long cycle and low efficiency in the backlight module design process in the existing related technologies, and can effectively shorten the design cycle of the backlight module and improve efficiency.

[0188] The device embodiments described above are merely illustrative, wherein the units described as separate components may or may not be physically separated, and the components shown as units may or may not be physical units, that is, they may be located in one place or distributed on multiple network units. Some or all of the modules may be selected according to actual needs to achieve the purpose of the solution of this embodiment.

[0189] Through the description of the above implementation methods, those skilled in the art can clearly understand that each implementation method can be implemented by means of software plus a general hardware platform, and of course, by hardware. Based on this understanding, the above technical solution is essentially or the part that contributes to the relevant technology can be embodied in the form of a software product, and the computer software product can be stored in a computer-readable storage medium, such as ROM / RAM, a disk, an optical disk, etc., including a number of instructions for a computer device (which can be a personal computer, a server, or a network device, etc.) to execute the methods described in each embodiment or some parts of the embodiments.

[0190] It should be understood that the terms used in the text are only for the purpose of describing specific example embodiments, and are not intended to be limiting. Unless the context clearly indicates otherwise, the singular forms "one", "an" and "said" as used in the text may also be meant to include plural forms. The terms "include", "comprise", "contain", and "have" are inclusive, and therefore specify the existence of stated features, steps, operations, elements and / or parts, but do not exclude the existence or addition of one or more other features, steps, operations, elements, parts, and / or combinations thereof. The method steps, processes, and operations described herein are not interpreted as necessarily requiring them to be performed in the specific order described or illustrated, unless the execution order is clearly indicated. It should also be understood that additional or alternative steps may be used.

[0191] The foregoing is merely a specific embodiment of the present invention, which enables those skilled in the art to understand or implement the present invention. Various modifications to these embodiments will be apparent to those skilled in the art, and the general principles defined herein may be implemented in other embodiments without departing from the spirit or scope of the present invention. Therefore, the present invention will not be limited to the embodiments shown herein, but rather to the widest scope consistent with the principles and novel features claimed herein.

Claims

1. A brightness prediction method, characterized in that: include: Obtaining backlight module configuration data corresponding to the backlight module to be predicted; Extracting information of factors to be predicted from the backlight module configuration data; The to-be-predicted factor information is predicted by a brightness prediction model to obtain brightness prediction information corresponding to the to-be-predicted backlight module.

2. The method according to claim 1, characterized in that The predicting the to-be-predicted factor information by using a brightness prediction model includes: Get the preset backlight brightness data; Extracting brightness factor information from the backlight brightness data; Using the brightness factor information to train a preset model to obtain model error information; adjusting the brightness factor information according to the model error information to obtain target brightness factor information; The preset model is trained using the target brightness factor information to obtain the brightness prediction model; The to-be-predicted factor information is predicted using the brightness prediction model.

3. The method according to claim 2, characterized in that The step of using the brightness factor information to train a preset model to obtain model error information includes: Based on the brightness factor information, generating a sample data set corresponding to the backlight brightness data; Performing sample division on the sample data set to obtain a first sample training set and a first sample test set; The preset model is trained using the first sample training set and the first sample test set to obtain model error information.

4. The method according to claim 3, characterized in that The step of training a preset model using the first sample training set and the first sample test set to obtain model error information includes: Using the first sample training set to train a preset model to obtain a first training error value; When the first training error value falls within a preset training error value range, determining the prediction model as the first model; Using the first sample test set to perform an error test on the first model to obtain a first test error value; The model error information is generated according to the first test error value.

5. The method according to claim 2, characterized in that: The step of adjusting the brightness factor information according to the model error information to obtain target brightness factor information includes: extracting a first test error value from the model error information; Determining whether the first test error value is within a preset test error value range; When the first test error value does not fall within the preset test error value range, the brightness factor information is subjected to dimensionality reduction adjustment to obtain target brightness factor information.

6. The method according to claim 5, characterized in that The step of performing dimension reduction adjustment on the brightness factor information to obtain target brightness factor information includes: Determine the number of primary factors and the number of secondary factors according to the brightness factor information, wherein the number of primary factors is the number of primary factor information included in the brightness factor information, and the number of secondary factors is the number of secondary factor information included in the brightness factor information; When the number of the secondary factors is not zero, at least one secondary factor information in the brightness factor information is removed to obtain the target brightness factor information; When the number of the secondary factors is zero, at least one primary factor information in the brightness factor information is removed to obtain the target brightness factor information.

7. The method according to any one of claims 1 to 6, characterized in that: The step of training the preset model using the target brightness factor information to obtain a brightness prediction model includes: Generate a second sample training set and a second sample test set according to the target brightness factor information; Using the second sample training set to train the preset model to obtain a second training error value; When the second training error value falls within a preset training error value range, determining the prediction model as a second model; Using the second sample test set to perform an error test on the second model to obtain a second test error value; When the second test error value is within the preset test error value range, the second model is determined as the brightness prediction model.

8. The method according to claim 7, characterized in that The adopting the target brightness factor information to train the preset model to obtain a brightness prediction model also includes: When the second test error value does not fall within the preset test error value range, updating the model error information according to the second test error value to obtain updated model error information; Based on the updated model error information, return to the step of adjusting the brightness factor information according to the model error information.

9. A brightness prediction device, characterized in that: include: An acquisition module, used for acquiring backlight module configuration data corresponding to the backlight module to be predicted; An extraction module, used for extracting information of factors to be predicted from the backlight module configuration data; The prediction module is used to predict the to-be-predicted factor information through a brightness prediction model to obtain brightness prediction information corresponding to the to-be-predicted backlight module.

10. An electronic device, characterized in that: It includes a processor, a communication interface, a memory and a communication bus, wherein the processor, the communication interface and the memory communicate with each other through the communication bus; Memory, used to store computer programs; A processor, configured to implement the brightness prediction method according to any one of claims 1 to 8 when executing a program stored in a memory.

11. A computer-readable storage medium having a computer program stored thereon, characterized in that: When the computer program is executed by a processor, the brightness prediction method according to any one of claims 1 to 8 is implemented.

Images