Projector projection brightness adaptive control method and system based on deep learning

Abstract

The invention discloses a projector projection brightness adaptive control method and system based on deep learning, and relates to the technical field of projector control, and the method comprises the steps: S1, constructing a deep learning model; s2, ambient light perception and data acquisition; s3, implementing a local brightness adjustment algorithm; s4, real-time brightness adjustment and feedback are carried out; and S5, performing user interaction and personalized setting. According to the projector projection brightness adaptive control method and system based on deep learning, a hybrid architecture combining multi-task learning and a convolutional neural network is designed, and the hybrid architecture is used for learning a complex relationship between ambient light and projection picture brightness and an optimal strategy for brightness adjustment of each region in a picture; and the adjustment strategy is applied to the projection picture, the brightness level of each area is adjusted in real time, and then the local brightness adjustment strategy is adjusted in combination with feedback information of the brightness adjustment effect, so that the optimal projection brightness adaptive control effect of the projector is achieved.

Description

Technical Field

[0001] The present invention relates to the technical field of projector control, and in particular to a method and system for adaptively controlling the brightness of a projector based on deep learning. Background Art

[0002] A projector, also known as a projector, is a device that can project images or video signals onto a screen or other flat surface. It is connected to computers, VCDs, DVDs, BDs, etc. through different interfaces to play the corresponding video or image content. It is widely used in homes, offices, schools, and entertainment venues, and is an important tool for modern multimedia presentations. The projection brightness of a projector, or "light out," is one of the main technical indicators of a projector. It describes the ability of a light source to produce a strong visual response per unit time, usually expressed in luminous flux, in lumens. Generally speaking, the higher the brightness of the projector, the brighter the image projected onto the screen, but too high a brightness may also cause eye fatigue.

[0003] Most projectors provide menu setting functions. The control method of traditional projector projection brightness is generally through menu settings. Users can adjust the screen brightness through options such as brightness, contrast, and color in the menu. This method is simple and direct and is suitable for most projectors. In addition, since the screen brightness of the projector is closely related to the projection distance and screen size, when the light source brightness is constant, appropriately reducing the projection distance or increasing the screen area can increase the screen brightness. However, this method is limited by the use environment and needs to be adjusted according to the actual situation.

[0004] With the development of technology, some high-end projectors have adopted adaptive brightness control methods to improve the viewing effect and comfort of the picture. The principle is to monitor the changes in ambient light in real time through the built-in ambient light sensor, and combine the hardware performance and software algorithm of the projector to automatically adjust the brightness, contrast, color and other parameters of the projector according to the content of the projected picture and the changes in ambient light to achieve the best viewing effect.

[0005] However, in the actual projection process, since the existing projector projection brightness adaptive control method usually adjusts and controls the overall brightness of the picture, and different areas of the projected picture may be affected by various factors such as ambient light, projection angle, screen material, etc., resulting in differences in brightness of various parts of the picture, if the projector cannot independently adjust specific areas in the picture, it is difficult to solve this problem of uneven brightness, thereby affecting the viewing comfort. Especially in some high-brightness or low-brightness areas, a single overall brightness adjustment will cause overexposure and loss of details in the high-brightness area, or too dark in the low-brightness area and difficult to identify details, which not only reduces the overall quality of the picture, but also affects the accurate communication of information. Moreover, brightness and color complement each other. If the brightness is not adjusted properly, the color may also be affected, resulting in problems such as color cast and distortion. Especially in the case of uneven brightness, the color performance of different areas may vary greatly, further reducing the viewing effect of the picture.

[0006] Therefore, there is an urgent need to improve this shortcoming. The present invention studies and improves the existing technology and its shortcomings, and provides a projector projection brightness adaptive control method and system based on deep learning. Summary of the invention

[0007] The purpose of the present invention is to provide a method and system for adaptively controlling the projection brightness of a projector based on deep learning to solve the problems raised in the above-mentioned background technology.

[0008] To achieve the above object, the present invention provides the following technical solutions:

[0009] On the one hand, a method for adaptively controlling the brightness of a projector based on deep learning is provided, comprising the following steps:

[0010] S1. Deep learning model construction:

[0011] Design a hybrid architecture that combines multi-task learning and convolutional neural networks (CNNs) to learn the complex relationship between ambient light and projected image brightness, as well as the optimal strategy for adjusting the brightness of each area within the image, and apply the designed deep learning model architecture to the projector;

[0012] S2. Ambient light perception and data collection:

[0013] The built-in ambient light sensor of the projector is used to monitor the light changes in the projection environment in real time, collect the light intensity, color temperature and other parameters in the current environment, and capture the image data of the projection screen through the image sensor of the projector. The image data includes but is not limited to the brightness distribution and color information of each area of ​​the screen;

[0014] S3, local brightness adjustment algorithm implementation:

[0015] Area division: The projection screen is divided into several areas, and the brightness of each area can be adjusted independently. At the same time, the area division can be dynamically adjusted according to factors such as screen content and brightness distribution;

[0016] Brightness assessment: Use deep learning models to analyze the brightness level, contrast, color and other information of each area to assess whether the current brightness setting is appropriate;

[0017] Strategy generation: Based on the evaluation results and the predictions of the deep learning model, a personalized brightness adjustment strategy is generated for each area, including increasing or decreasing brightness, adjusting contrast, optimizing color, etc.

[0018] S4, real-time brightness adjustment and feedback:

[0019] Apply the local brightness adjustment strategy to the projected image, adjust the brightness level of each area in real time, and use the image sensor to capture the adjusted image data again to evaluate the effect of the brightness adjustment. If the effect is not good, adjust the deep learning model or the local brightness adjustment strategy according to the feedback information for iterative optimization.

[0020] S5. User interaction and personalization:

[0021] Provide a user interaction interface that allows users to adjust the sensitivity, range and other parameters of brightness adjustment according to personal preferences and needs, and provide users with personalized brightness adjustment suggestions and content recommendations based on user usage history and preference data.

[0022] Furthermore, in step S1, the deep learning model construction specifically includes the following sub-steps:

[0023] S11. Dataset construction: collect ambient light data (including brightness, color temperature, etc.) under different lighting conditions through ambient light sensors, capture image data of the projection screen for each ambient light data point (the image should contain sufficient resolution to allow area division and contain key information such as brightness, contrast, color, etc.), and generate labels for each projection screen;

[0024] S12. Deep learning model architecture design: including input layer, feature extraction layer, fusion layer, multi-task learning layer and output layer;

[0025] S13. Loss function and optimization:

[0026] Loss function: A loss function consisting of multiple parts is designed, corresponding to the prediction errors of global brightness adjustment and regional brightness adjustment, respectively. The mean square error (MSE) and structural similarity index (SSIM) are used to evaluate the difference between the predicted value and the true value.

[0027] Optimization algorithm: Select an optimization algorithm suitable for the problem, such as Adam, RMSprop, etc., to train the model. During the training process, monitor the changes in the loss function and perform model parameter adjustment and early stopping as needed.

[0028] S14. Model training and evaluation:

[0029] Use the collected data set to train the model to ensure that the model can learn the complex relationship between ambient light and projected image brightness and generate an effective brightness adjustment strategy;

[0030] During the training process, the validation set is used to evaluate the performance of the model, and after the training is completed, the test set is used to verify the generalization ability of the model. When evaluating the performance of the model, we should not only pay attention to the accuracy of brightness adjustment, but also pay attention to the picture quality and color fidelity after adjustment.

[0031] Furthermore, in step S11, labels are generated by manual labeling or automatic labeling based on an expert system, and the labels include the expected brightness value of each area, the brightness adjustment amount, and other indicators related to brightness adjustment.

[0032] Furthermore, in step S12, the design details of the input layer, feature extraction layer, fusion layer, multi-task learning layer and output layer are as follows:

[0033] 1) Input layer

[0034] Ambient light input: accepts numerical data from the ambient light sensor and performs preprocessing (such as normalization);

[0035] Projection screen input: accept the image data of the projection screen, including single-channel grayscale images and multi-channel color images, crop the image to a suitable size, and perform data enhancement (such as random cropping, rotation, etc.) to improve the generalization ability of the model;

[0036] 2) Feature extraction layer

[0037] Ambient light feature extraction: Use fully connected layers (for simple ambient light data) or simple convolutional layers (if the ambient light data is provided in the form of images) to extract the features of the ambient light;

[0038] Projection image feature extraction: Use several convolutional layers, pooling layers, and activation functions (such as ReLU) to extract the features of the projection image. The features can capture information such as brightness distribution, texture, and edges in the image.

[0039] 3) Fusion layer

[0040] The features of ambient light and the features of the projected image are effectively fused by concatenation (joining the feature vectors (or feature maps, if they are obtained through some form of convolution) of ambient light with the feature maps of the projected image along a certain dimension (usually the channel dimension)), addition (adding the feature maps of ambient light and the feature maps of the projected image on corresponding elements. This method requires that the sizes of the two feature maps must be exactly the same. Addition can be regarded as a linear combination of the feature maps in spatial positions, which helps the model learn the direct impact of ambient light on the brightness of the projected image), or attention mechanism (allowing the model to dynamically adjust the importance of features from different sources when fusing features. In the brightness adjustment scenario, the attention mechanism can help the model focus on the image areas that are most sensitive to changes in ambient light and generate more accurate brightness adjustment strategies accordingly).

[0041] 4) Multi-task learning layer

[0042] Global brightness adjustment prediction: Use a branch network to predict the global brightness value that should be adjusted for the entire projected image. The specific operation is to use a fully connected layer to output a scalar value.

[0043] Regional brightness adjustment prediction: Another branch network is used to predict the brightness adjustment amount of each area within the picture. The specific operation is to upsample the feature map or use transposed convolution to achieve this, and finally output a brightness adjustment map of the same size as the input picture, where each pixel value represents the brightness adjustment amount of the corresponding area;

[0044] 5) Output layer

[0045] The output layer generates the final brightness adjustment strategy based on the results of the multi-task learning layer.

[0046] Furthermore, the attention mechanism specifically includes:

[0047] Calculate attention weights: Calculate an attention weight map based on the characteristics of the ambient light and the projection image. The weight map is a two-dimensional matrix of the same size as the projection image feature map, in which each element represents the importance that should be given to the corresponding image area in the fusion process;

[0048] Apply attention weights: perform element-wise multiplication (or weighted sum) of the attention weight map and the feature map of the projected image to generate a weighted feature map that contains both the original information of the projected image and selectively emphasizes or suppresses it based on changes in ambient light.

[0049] Fusion features: The weighted projection image feature map (or the feature map after further processing) is concatenated or added with the features of the ambient light to generate the final fusion features.

[0050] Furthermore, in step S4, the implementation of real-time brightness adjustment relies on the hardware support and algorithm processing capability of the projector, and specifically adjusts the local brightness by adjusting the hardware parameters of the projector (such as light source brightness, grating control, etc.).

[0051] Furthermore, the hardware support specifically includes:

[0052] Adjustable light source: The projector must be equipped with a light source with adjustable brightness, such as LED or laser light source, and the light source must support fine-grained brightness control so that the brightness of different areas can be adjusted according to the local brightness adjustment strategy;

[0053] Dynamic grating technology: The projector uses dynamic grating technology, which allows the projector to project light of different brightness in different areas. By controlling the switch and transmittance of the grating, more precise brightness control can be achieved;

[0054] High-resolution display chip: Using high-resolution display chips such as DLP or LCD, carefully controlling each pixel of the projected image, thus achieving smoother brightness transitions in different areas;

[0055] High-speed image processing unit: The projector should be equipped with a powerful image processing unit (GPU or similar hardware) to quickly process the brightness adjustment map generated by the local brightness adjustment algorithm and apply it to the projected image.

[0056] Furthermore, the algorithm processing capability specifically includes:

[0057] Efficient algorithm implementation: The local brightness adjustment algorithm needs to be implemented efficiently to minimize computational latency, which includes using optimized data structures, reducing unnecessary computations, and leveraging parallel computing and hardware acceleration techniques (such as GPU acceleration).

[0058] Real-time optimization: The algorithm needs to be designed to respond to changes in ambient light in real time, which means that the algorithm needs to complete the calculation and application of brightness adjustment in a very short time to avoid the audience to perceive obvious delays or lags;

[0059] Feedback mechanism: Implement a feedback mechanism to capture the image of the projected screen after brightness adjustment and compare it with the expected brightness effect. If there is a difference, the algorithm will automatically adjust its parameters or strategies to gradually approach the optimal brightness adjustment effect;

[0060] Multi-core processing or distributed computing: For complex local brightness adjustment algorithms, multi-core processors or distributed computing frameworks are used to speed up the calculation process. By distributing the calculation tasks to multiple processing units, the calculation time can be significantly shortened.

[0061] On the other hand, a projector projection brightness adaptive control system based on deep learning is provided, which is applied to the above-mentioned projector projection brightness adaptive control method based on deep learning, including:

[0062] Model building module: responsible for building a deep learning model architecture to learn the complex relationship between ambient light and projection screen brightness, as well as the optimal strategy for adjusting the brightness of each area within the screen;

[0063] Real-time monitoring module: responsible for using the built-in ambient light sensor of the projector to monitor the light changes in the projection environment in real time, collect the parameters in the current environment, and capture the image data of the projection screen through the image sensor of the projector;

[0064] Area division module: responsible for dividing the projection screen into several areas that can be independently adjusted for brightness;

[0065] Brightness assessment module: responsible for using deep learning models to analyze the brightness level, contrast, color and other information of each area to evaluate whether the current brightness setting is appropriate;

[0066] Strategy generation module: responsible for generating personalized brightness adjustment strategies for each area based on the evaluation results and predictions of the deep learning model;

[0067] Brightness adjustment module: responsible for applying the local brightness adjustment strategy to the projection screen, adjusting the brightness level of each area in real time, and the brightness adjustment result will be fed back to the real-time monitoring module;

[0068] User interaction interface: responsible for providing an intuitive and easy-to-use interface, allowing users to view the brightness distribution of the current projection screen, adjust the sensitivity, priority and other parameters of the brightness adjustment, and provide personalized brightness adjustment suggestions and content recommendations based on the user's usage history and preference data. The user interaction interface uses a graphical interface to display a real-time preview of the projection screen and a visual representation of the brightness adjustment area, and provides touch or button operation methods to facilitate users to make personalized settings.

[0069] Furthermore, the system also includes:

[0070] A processor, configured to execute computer instructions so that the system executes: a model building module, a real-time monitoring module, a region division module, a brightness evaluation module, a strategy generation module, a brightness adjustment module, and a user interaction interface;

[0071] The memory is used to store computer instructions. A computer program is stored in the memory. When the computer program is executed by the processor, the method described above is implemented.

[0072] The present invention provides a method and system for adaptively controlling the brightness of a projector based on deep learning, which has the following beneficial effects:

[0073] The present invention designs a hybrid architecture that combines multi-task learning and convolutional neural network (CNN) to learn the complex relationship between ambient light and projection screen brightness, as well as the optimal strategy for brightness adjustment of each area within the screen, and applies the adjustment strategy to the projection screen, adjusts the brightness level of each area in real time, and then adjusts the local brightness adjustment strategy in combination with the feedback information of the brightness adjustment effect to achieve the best projector projection brightness adaptive control effect. The present invention can realize independent brightness adjustment of different areas of the screen, improve the uniformity of the overall brightness of the screen and viewing comfort, while avoiding overexposure of highlight areas, loss of details, and over-darkness of low-brightness areas, making details difficult to recognize, which helps to improve the overall quality of the screen and is conducive to the accurate transmission of information. BRIEF DESCRIPTION OF THE DRAWINGS

[0074] Figure 1 A schematic diagram of the steps of a method for adaptively controlling brightness of a projector based on deep learning according to the present invention;

[0075] Figure 2 This is a logic block diagram of a projector brightness adaptive control system based on deep learning in the present invention. DETAILED DESCRIPTION

[0076] The following embodiments of the present invention are described in further detail in conjunction with the accompanying drawings and examples. The following examples are used to illustrate the present invention, but are not intended to limit the scope of the present invention.

[0077] Embodiment 1

[0078] like Figure 1 As shown, a method for adaptively controlling the brightness of a projector based on deep learning comprises the following steps:

[0079] S1. Deep learning model construction:

[0080] Design a hybrid architecture that combines multi-task learning and convolutional neural network (CNN) to learn the complex relationship between ambient light and projected image brightness, as well as the optimal strategy for adjusting the brightness of each area within the image, and apply the designed deep learning model architecture to the projector; specifically, the following sub-steps are included:

[0081] S11. Dataset construction: Ambient light data (including brightness, color temperature, etc.) under different lighting conditions are collected through ambient light sensors. For each ambient light data point, image data of the projection screen is captured (the image should contain sufficient resolution to allow regional division and contain key information such as brightness, contrast, and color), and labels are generated for each projection screen. The labels are generated by manual labeling or automatic labeling based on an expert system, and the labels include the expected brightness value of each area, the brightness adjustment amount, and other indicators related to brightness adjustment.

[0082] S12. Deep learning model architecture design: including input layer, feature extraction layer, fusion layer, multi-task learning layer and output layer. The design details of each layer are as follows:

[0083] 1) Input layer

[0084] Ambient light input: accepts numerical data from the ambient light sensor and performs preprocessing (such as normalization);

[0085] Projection screen input: accept the image data of the projection screen, including single-channel grayscale images and multi-channel color images, crop the image to a suitable size, and perform data enhancement (such as random cropping, rotation, etc.) to improve the generalization ability of the model;

[0086] 2) Feature extraction layer

[0087] Ambient light feature extraction: Use fully connected layers (for simple ambient light data) or simple convolutional layers (if the ambient light data is provided in the form of images) to extract the features of the ambient light;

[0088] Projection image feature extraction: Use several convolutional layers, pooling layers, and activation functions (such as ReLU) to extract the features of the projection image. The features can capture information such as brightness distribution, texture, and edges in the image.

[0089] 3) Fusion layer

[0090] By concatenating (joining the feature vectors (or feature maps, if they are obtained through some form of convolution) of the ambient light with the feature maps of the projected image along a certain dimension (usually the channel dimension); assuming there are two feature maps and Among them, H and W are the height and width of the feature map, C1 and C2 are the number of channels, and the concatenated feature map is ), addition (addition of the feature map of ambient light and the feature map of the projection image on the corresponding elements. This method requires that the sizes of the two feature maps must be exactly the same. Addition can be regarded as a linear combination of the feature maps in space, which helps the model learn the direct impact of ambient light on the brightness of the projection image. Assume that the dimensions of F1 and F2 are the same, that is, F1, Then the added feature map It can be calculated by the following formula: sum (i,j,k)=F1(i,j,k)+F2(i,j,k), where i,j,k represent the row, column and channel index of the feature map respectively) or attention mechanism (which allows the model to dynamically adjust the importance of features from different sources when fusing features. In the brightness adjustment scenario, the attention mechanism can help the model focus on those areas of the screen that are most sensitive to changes in ambient light and generate a more accurate brightness adjustment strategy accordingly) to effectively fuse the features of the ambient light and the features of the projected image. The attention mechanism specifically includes:

[0091] Calculate attention weights: Calculate an attention weight map based on the characteristics of the ambient light and the projection image. The weight map is a two-dimensional matrix of the same size as the projection image feature map, in which each element represents the importance that should be given to the corresponding image area in the fusion process.

[0092] Apply attention weights: perform element-wise multiplication (or weighted sum) of the attention weight map and the feature map of the projected image to generate a weighted feature map that contains both the original information of the projected image and selectively emphasizes or suppresses it based on changes in ambient light.

[0093] Fusion features: The weighted projection image feature map (or the feature map after further processing) is concatenated or added with the features of the ambient light to generate the final fusion features;

[0094] A specific example of the attention mechanism is as follows:

[0095] A. Calculate the similarity matrix S between F1 and F2 using dot product, cosine similarity or other similarity metrics;

[0096] B. Use the softmax function to normalize the similarity matrix to obtain the attention weight A;

[0097] C. Use attention weight A to weightedly fuse F1 and F2;

[0098] Assume that the normalized attention weight A∈R H×W×C (Note that some form of flattening or rearrangement is required to adapt to the specific implementation), then the weighted fused feature map F attn It is calculated as follows (shown here in simplified form):

[0099]

[0100] Or more commonly, attention weights might be applied to only one feature map, and then supplemented with another:

[0101]

[0102] in, Represents element-wise multiplication;

[0103] 4) Multi-task learning layer

[0104] Global brightness adjustment prediction: Use a branch network to predict the global brightness value that should be adjusted for the entire projected image. The specific operation is to use a fully connected layer to output a scalar value.

[0105] Regional brightness adjustment prediction: Another branch network is used to predict the brightness adjustment amount of each area within the picture. The specific operation is to upsample the feature map or use transposed convolution to achieve this, and finally output a brightness adjustment map of the same size as the input picture, where each pixel value represents the brightness adjustment amount of the corresponding area;

[0106] 5) Output layer

[0107] The output layer generates the final brightness adjustment strategy based on the results of the multi-task learning layer;

[0108] S13. Loss function and optimization:

[0109] Loss function: A loss function consisting of multiple parts is designed, corresponding to the prediction errors of global brightness adjustment and regional brightness adjustment, respectively. The mean square error (MSE) and structural similarity index (SSIM) are used to evaluate the difference between the predicted value and the true value.

[0110] Optimization algorithm: Select an optimization algorithm suitable for the problem, such as Adam, RMSprop, etc., to train the model. During the training process, monitor the changes in the loss function and perform model parameter adjustment and early stopping as needed.

[0111] S14. Model training and evaluation:

[0112] Use the collected data set to train the model to ensure that the model can learn the complex relationship between ambient light and projected image brightness and generate an effective brightness adjustment strategy;

[0113] During the training process, the validation set is used to evaluate the performance of the model, and after the training is completed, the test set is used to verify the generalization ability of the model. When evaluating the performance of the model, it is necessary to pay attention not only to the accuracy of brightness adjustment, but also to the quality of the adjusted picture, color fidelity, etc.

[0114] S2. Ambient light perception and data collection:

[0115] The built-in ambient light sensor of the projector is used to monitor the light changes in the projection environment in real time, collect the light intensity, color temperature and other parameters in the current environment, and capture the image data of the projection screen through the image sensor of the projector. The image data includes but is not limited to the brightness distribution and color information of each area of ​​the screen;

[0116] S3, local brightness adjustment algorithm implementation:

[0117] Area division: The projection screen is divided into several areas, and the brightness of each area can be adjusted independently. At the same time, the area division can be dynamically adjusted according to factors such as screen content and brightness distribution;

[0118] Brightness assessment: Use deep learning models to analyze the brightness level, contrast, color and other information of each area to assess whether the current brightness setting is appropriate;

[0119] Strategy generation: Based on the evaluation results and the predictions of the deep learning model, a personalized brightness adjustment strategy is generated for each area, including increasing or decreasing brightness, adjusting contrast, optimizing color, etc.

[0120] S4, real-time brightness adjustment and feedback:

[0121] Apply the local brightness adjustment strategy to the projected image, adjust the brightness level of each area in real time, and use the image sensor to capture the adjusted image data again to evaluate the effect of the brightness adjustment. If the effect is not good, adjust the deep learning model or the local brightness adjustment strategy according to the feedback information for iterative optimization. The implementation of real-time brightness adjustment relies on the hardware support and algorithm processing capabilities of the projector. Specifically, the local brightness is adjusted by adjusting the hardware parameters of the projector (such as light source brightness, grating control, etc.).

[0122] Hardware support specifically includes:

[0123] Adjustable light source: The projector must be equipped with a light source with adjustable brightness, such as LED or laser light source, and the light source must support fine-grained brightness control so that the brightness of different areas can be adjusted according to the local brightness adjustment strategy;

[0124] Dynamic grating technology: The projector uses dynamic grating technology, which allows the projector to project light of different brightness in different areas. By controlling the switch and transmittance of the grating, more precise brightness control can be achieved;

[0125] High-resolution display chip: Using high-resolution display chips such as DLP or LCD, carefully controlling each pixel of the projected image, thus achieving smoother brightness transitions in different areas;

[0126] High-speed image processing unit: The projector should be equipped with a powerful image processing unit (GPU or similar hardware) to quickly process the brightness adjustment map generated by the local brightness adjustment algorithm and apply it to the projected image;

[0127] Among them, the algorithm processing capabilities specifically include:

[0128] Efficient algorithm implementation: The local brightness adjustment algorithm needs to be implemented efficiently to minimize computational latency, which includes using optimized data structures, reducing unnecessary computations, and leveraging parallel computing and hardware acceleration techniques (such as GPU acceleration).

[0129] Real-time optimization: The algorithm needs to be designed to respond to changes in ambient light in real time, which means that the algorithm needs to complete the calculation and application of brightness adjustment in a very short time to avoid the audience to perceive obvious delays or lags;

[0130] Feedback mechanism: Implement a feedback mechanism to capture the image of the projected screen after brightness adjustment and compare it with the expected brightness effect. If there is a difference, the algorithm will automatically adjust its parameters or strategies to gradually approach the optimal brightness adjustment effect;

[0131] Multi-core processing or distributed computing: For complex local brightness adjustment algorithms, multi-core processors or distributed computing frameworks are used to accelerate the calculation process. By distributing the calculation tasks to multiple processing units, the calculation time can be significantly shortened.

[0132] S5. User interaction and personalization:

[0133] Provide a user interaction interface that allows users to adjust the sensitivity, range and other parameters of brightness adjustment according to personal preferences and needs, and provide users with personalized brightness adjustment suggestions and content recommendations based on user usage history and preference data.

[0134] Embodiment 2

[0135] like Figure 2 As shown, a projector projection brightness adaptive control system based on deep learning is applied to the above-mentioned projector projection brightness adaptive control method based on deep learning, including:

[0136] Model building module: responsible for building a deep learning model architecture to learn the complex relationship between ambient light and projection screen brightness, as well as the optimal strategy for adjusting the brightness of each area within the screen;

[0137] Real-time monitoring module: responsible for using the built-in ambient light sensor of the projector to monitor the light changes in the projection environment in real time, collect the parameters in the current environment, and capture the image data of the projection screen through the image sensor of the projector;

[0138] Area division module: responsible for dividing the projection screen into several areas that can be independently adjusted for brightness;

[0139] Brightness assessment module: responsible for using deep learning models to analyze the brightness level, contrast, color and other information of each area to evaluate whether the current brightness setting is appropriate;

[0140] Strategy generation module: responsible for generating personalized brightness adjustment strategies for each area based on the evaluation results and predictions of the deep learning model;

[0141] Brightness adjustment module: responsible for applying the local brightness adjustment strategy to the projection screen, adjusting the brightness level of each area in real time, and the brightness adjustment result will be fed back to the real-time monitoring module;

[0142] User interaction interface: responsible for providing an intuitive and easy-to-use interface, allowing users to view the brightness distribution of the current projection screen, adjust the sensitivity, priority and other parameters of brightness adjustment, and provide personalized brightness adjustment suggestions and content recommendations based on the user's usage history and preference data. The user interaction interface adopts a graphical interface to display the real-time preview of the projection screen and the visual representation of the brightness adjustment area, and provides touch or key operation methods to facilitate users to make personalized settings;

[0143] A processor, configured to execute computer instructions so that the system executes: a model building module, a real-time monitoring module, a region division module, a brightness evaluation module, a strategy generation module, a brightness adjustment module, and a user interaction interface;

[0144] A memory is used to store computer instructions. A computer program is stored in the memory. When the computer program is executed by the processor, the projector projection brightness adaptive control method based on deep learning as described above is implemented.

[0145] The embodiments of the present invention are given for the purpose of illustration and description, and are not intended to be exhaustive or to limit the invention to the disclosed forms. Many modifications and variations will be apparent to those of ordinary skill in the art. The embodiments are selected and described in order to better illustrate the principles and practical applications of the present invention and to enable those of ordinary skill in the art to understand the present invention and thereby design various embodiments with various modifications suitable for specific uses.

Claims

1. A projector projection brightness adaptive control method based on deep learning, characterized in that: The following steps are involved: S1. Deep learning model construction: Design a hybrid architecture that combines multi-task learning and convolutional neural networks to learn the complex relationship between ambient light and projected image brightness, as well as the optimal strategy for adjusting the brightness of each area within the image, and apply the designed deep learning model architecture to the projector; S2. Ambient light perception and data collection: The built-in ambient light sensor of the projector is used to monitor the light changes in the projection environment in real time, collect the parameters of the current environment, and capture the image data of the projection screen through the image sensor of the projector; S3, local brightness adjustment algorithm implementation: Area division: the projection screen is divided into several areas, and the brightness of each area can be adjusted independently; Brightness assessment: Use deep learning models to analyze information such as brightness level, contrast, and color in each area to assess whether the current brightness setting is appropriate; Strategy generation: Generates personalized brightness adjustment strategies for each area based on the evaluation results and predictions of the deep learning model, including increasing or decreasing brightness, adjusting contrast, and optimizing color; S4, real-time brightness adjustment and feedback: Apply the local brightness adjustment strategy to the projected image, adjust the brightness level of each area in real time, and use the image sensor to capture the adjusted image data again to evaluate the effect of the brightness adjustment. If the effect is not good, adjust the deep learning model or the local brightness adjustment strategy according to the feedback information for iterative optimization. S5. User interaction and personalization: Provides a user interaction interface that allows users to adjust the various parameters of brightness adjustment according to personal preferences and needs, and provides users with personalized brightness adjustment suggestions and content recommendations based on user usage history and preference data.

2. According to the method of adaptively controlling the brightness of a projector based on deep learning in claim 1, it is characterized in that: The step S1 specifically includes the following sub-steps: S11, Dataset construction: Collect ambient light data under different lighting conditions through ambient light sensors, capture image data of the projection screen corresponding to each ambient light data point, and generate labels for each projection screen; S12. Deep learning model architecture design: including input layer, feature extraction layer, fusion layer, multi-task learning layer and output layer; S13. Loss function and optimization: Loss function: A loss function consisting of multiple parts is designed, corresponding to the prediction errors of global brightness adjustment and regional brightness adjustment, respectively. The mean square error and structural similarity index are used to evaluate the difference between the predicted value and the true value. Optimization algorithm: Select an optimization algorithm suitable for the problem to train the model. During the training process, monitor the changes in the loss function and adjust the model parameters and perform early stopping operations as needed. S14. Model training and evaluation: Use the collected data set to train the model to ensure that the model can learn the complex relationship between ambient light and projection screen brightness and generate an effective brightness adjustment strategy; During the training process, the validation set is used to evaluate the performance of the model, and after the training is completed, the test set is used to verify the generalization ability of the model.

3. According to the method of adaptively controlling the brightness of a projector based on deep learning in claim 2, it is characterized in that: In the step S11, the label is generated by manual labeling or automatic labeling based on an expert system, and the label includes the expected brightness value of each area, the brightness adjustment amount, and other indicators related to the brightness adjustment.

4. The method for adaptively controlling the brightness of a projector based on deep learning according to claim 2, characterized in that: In step S12, the design details of the input layer, feature extraction layer, fusion layer, multi-task learning layer and output layer are as follows: 1) Input layer Ambient light input: accepts numerical data from the ambient light sensor and performs preprocessing; Projection screen input: accepts image data of the projection screen, including single-channel grayscale images and multi-channel color images, crops the image to a suitable size, and performs data enhancement; 2) Feature extraction layer Ambient light feature extraction: Use a fully connected layer or a simple convolutional layer to extract the features of ambient light; Projection image feature extraction: Use several convolutional layers, pooling layers, and activation functions to extract the features of the projection image, which can capture various information in the image; 3) Fusion layer Through concatenation, addition or attention mechanism, the characteristics of ambient light and projection image are effectively integrated; 4) Multi-task learning layer Global brightness adjustment prediction: Use a branch network to predict the global brightness value that should be adjusted for the entire projected image. The specific operation is to use a fully connected layer to output a scalar value. Regional brightness adjustment prediction: Another branch network is used to predict the brightness adjustment amount of each area within the picture. The specific operation is to upsample the feature map or use transposed convolution to achieve this, and finally output a brightness adjustment map of the same size as the input picture, where each pixel value represents the brightness adjustment amount of the corresponding area; 5) Output layer The output layer generates the final brightness adjustment strategy based on the results of the multi-task learning layer.

5. The method for adaptively controlling the brightness of a projector based on deep learning according to claim 4, characterized in that: The attention mechanism specifically includes: Calculate attention weights: Calculate an attention weight map based on the characteristics of the ambient light and the projection image. The weight map is a two-dimensional matrix of the same size as the projection image feature map, in which each element represents the importance that should be given to the corresponding image area in the fusion process; Apply attention weights: perform element-wise multiplication of the attention weight map with the feature map of the projected image to generate a weighted feature map. Fusion features: The weighted projection image feature map is concatenated or added with the features of the ambient light to generate the final fusion features.

6. The method for adaptively controlling the brightness of a projector based on deep learning according to claim 1, characterized in that: In step S4, the implementation of real-time brightness adjustment relies on the hardware support and algorithm processing capability of the projector, and specifically, the local brightness is adjusted by adjusting the hardware parameters of the projector.

7. The method for adaptively controlling the brightness of a projector based on deep learning according to claim 6, characterized in that: The hardware support specifically includes: Adjustable light source: The projector must be equipped with a light source with adjustable brightness, and the light source supports fine-grained brightness control; Dynamic grating technology: The projector uses dynamic grating technology, which allows the projector to project light of different brightness in different areas. By controlling the switch and transmittance of the grating, more precise brightness control can be achieved; High-resolution display chip: Using DLP or LCD display chip, each pixel of the projected image is carefully controlled, so as to achieve smoother brightness transition in different areas; High-speed image processing unit: The projector should be equipped with a powerful image processing unit to quickly process the brightness adjustment map generated by the local brightness adjustment algorithm and apply it to the projected picture.

8. The method for adaptively controlling the brightness of a projector based on deep learning according to claim 6, characterized in that: The algorithm processing capabilities specifically include: Efficient algorithm implementation: The local brightness adjustment algorithm needs to be implemented efficiently to minimize computational latency, which includes using optimized data structures, reducing unnecessary computations, and leveraging parallel computing and hardware acceleration techniques. Real-time optimization: The algorithm needs to be designed to respond to changes in ambient light in real time, which means that the algorithm needs to complete the calculation and application of brightness adjustment in a very short time to avoid the audience to perceive obvious delays or lags; Feedback mechanism: Implement a feedback mechanism to capture the image of the projected screen after brightness adjustment and compare it with the expected brightness effect. If there is a difference, the algorithm will automatically adjust its parameters or strategies to gradually approach the optimal brightness adjustment effect; Multi-core processing or distributed computing: For complex local brightness adjustment algorithms, multi-core processors or distributed computing frameworks are used to speed up the calculation process. By distributing the calculation tasks to multiple processing units, the calculation time can be significantly shortened.

9. A projector projection brightness adaptive control system based on deep learning, applied to the projector projection brightness adaptive control method based on deep learning as claimed in any one of claims 1 to 8, characterized in that: include: Model building module: responsible for building a deep learning model architecture to learn the complex relationship between ambient light and projection screen brightness, as well as the optimal strategy for adjusting the brightness of each area within the screen; Real-time monitoring module: responsible for using the built-in ambient light sensor of the projector to monitor the light changes in the projection environment in real time, collect the parameters in the current environment, and capture the image data of the projection screen through the image sensor of the projector; Area division module: responsible for dividing the projection screen into several areas that can be independently adjusted for brightness; Brightness evaluation module: responsible for analyzing the information of each area using a deep learning model to evaluate whether the current brightness setting is appropriate; Strategy generation module: responsible for generating personalized brightness adjustment strategies for each area based on the evaluation results and predictions of the deep learning model; Brightness adjustment module: responsible for applying the local brightness adjustment strategy to the projection screen, adjusting the brightness level of each area in real time, and the brightness adjustment result will be fed back to the real-time monitoring module; User interaction interface: Responsible for providing an intuitive and easy-to-use interface that allows users to view the brightness distribution of the current projection image, adjust brightness adjustment parameters, and provide personalized brightness adjustment suggestions and content recommendations based on user usage history and preference data.

10. The projector projection brightness adaptive control system based on deep learning according to claim 9, characterized in that: include: A processor, configured to execute computer instructions so that the system executes: a model building module, a real-time monitoring module, a region division module, a brightness evaluation module, a strategy generation module, a brightness adjustment module, and a user interaction interface; A memory for storing computer instructions, wherein a computer program is stored in the memory, and when the computer program is executed by a processor, the method according to any one of claims 1 to 8 is implemented.

Images