A transparent LED display brightness regulation method and system based on environmental perception

Abstract

The application provides a kind of based on environmental perception's transparent LED display screen brightness regulation method and system, it is related to display control technical field, this method includes: obtaining historical ambient light sequence and determining future ambient light prediction result;Accordingly determine background transmission brightness value and initial foreground luminous brightness value, construct light-dark contrast relationship;Determine initial visual perception evaluation result in combination with visual adaptation parameter;When evaluation result is not up to standard and contrast relationship exceeds tolerance range, determine brightness compensation and obtain updated foreground luminous brightness value according to visual comfort threshold limit constraint adjustment;Redetermine target light-dark contrast relationship and evaluate again, when meeting conditions, output updated brightness value.The application can realize the dynamic brightness precision regulation of environmental self-adaptation, improve picture level distinctness and viewing comfort.

Description

TECHNICAL FIELD

[0001] The present application relates to the technical field of display control, in particular to a transparent LED display screen brightness regulation method and system based on environment perception. BACKGROUND

[0002] The transparent LED display screen has been widely used in commercial show windows, building curtain walls and public transportation scenes due to its characteristics of information display and space permeability. In the existing technical solutions, the brightness regulation of such display screens usually relies on the environmental light intensity data collected by photosensitive sensors in real time. The system reads the current illumination value, adjusts the driving current or duty cycle of the screen light emitting unit directly according to the preset linear correspondence or fixed threshold interval, so as to realize the synchronous change of screen brightness with the environmental light intensity. This technical route mainly focuses on the instant matching of physical light intensity value, aiming to ensure that the screen brightness maintains a certain proportional relationship with the environment on a macro level.

[0003] However, in actual application process, the existing technical solutions face the long-term challenge that the visual comfort and clarity of the display picture are difficult to balance. Since the environmental light has dynamic fluctuation, there is inherent delay between the instantaneous measurement of the sensor and the system response, and the adjustment action always lags behind the actual change of light; at the same time, the human eye has a nonlinear physiological adaptation process in the perception of light and dark contrast, which will be further amplified under the nonlinear perception mechanism, resulting in imbalance between the light and dark levels of the foreground content and the background transmission image of the screen. This imbalance manifests as obvious jumping flicker of the picture when the light changes rapidly, or the foreground content appears gray and blurred due to insufficient contrast, or even produces glare due to too high brightness under certain light conditions.

[0004] Therefore, how to realize the light and dark coordinated control between the foreground light emission and the background transmission of the transparent LED display screen under the premise of considering the dynamic change of environmental light and the visual adaptation characteristics of the human eye, so that the display picture can present clear and comfortable visual levels under different light conditions, has become a technical problem to be solved in the field. SUMMARY

[0005] In view of the above problems, the present application provides a transparent LED display screen brightness regulation method and system based on environment perception, which is used to solve the technical problems of adjustment delay caused by the dynamic change of environmental light and the lag of sensor response, and the imbalance between the foreground and background transmission of the picture caused by the lack of consideration of the visual adaptation state of the human eye and the nonlinear light and dark perception characteristics.

[0006] To achieve the above purpose, the technical scheme adopted by the present application is:

[0007] The first aspect of this application provides a method for adjusting the brightness of a transparent LED display screen based on environmental perception. The method includes: acquiring a historical ambient lighting sequence of the scene where the transparent LED display screen is located, and determining a future ambient lighting prediction result based on the historical ambient lighting sequence; determining a background transmission brightness value corresponding to the transparent LED display screen according to the future ambient lighting prediction result, and acquiring an initial foreground luminous brightness value corresponding to the transparent LED display screen; determining a contrast relationship between the foreground and background based on the background transmission brightness value and the initial foreground luminous brightness value; determining an initial visual perception evaluation result based on the contrast relationship and visual adaptation parameters characterizing the observer's visual adaptation state; and determining an initial visual perception evaluation result based on the initial visual perception evaluation result. When the visual perception evaluation result does not meet the preset visual perception evaluation conditions and the brightness-darkness contrast relationship exceeds the preset brightness-darkness transition tolerance range, a brightness compensation amount is determined based on the future ambient lighting prediction result and the brightness-darkness contrast relationship. Then, based on the visual comfort threshold determined based on the background transmittance brightness value, the foreground luminous brightness value corresponding to the brightness compensation amount is constrained and adjusted to obtain an updated foreground luminous brightness value. The target brightness-darkness contrast relationship is re-determined based on the updated foreground luminous brightness value and the background transmittance brightness value. A visual perception evaluation is then performed again based on the target brightness-darkness contrast relationship. When the re-determined visual perception evaluation result meets the preset visual perception evaluation conditions, the updated foreground luminous brightness value is output.

[0008] A second aspect of this application provides a brightness control system for a transparent LED display screen based on environmental perception. The system includes: an ambient light prediction module for acquiring historical ambient light sequences of the scene where the transparent LED display screen is located, and determining future ambient light prediction results based on the historical ambient light sequences; a contrast relationship determination module for determining the background transmission brightness value corresponding to the transparent LED display screen based on the future ambient light prediction results, acquiring the initial foreground luminous brightness value corresponding to the transparent LED display screen, and determining the contrast relationship between the foreground and background based on the background transmission brightness value and the initial foreground luminous brightness value; and an initial visual perception evaluation module for determining an initial visual perception evaluation result based on the contrast relationship and visual adaptation parameters characterizing the observer's visual adaptation state. The system includes a brightness compensation adjustment module, used to determine a brightness compensation amount based on the future ambient lighting prediction result and the brightness contrast relationship when the initial visual perception evaluation result does not meet the preset visual perception evaluation conditions and the brightness contrast relationship exceeds the preset brightness transition tolerance range. The module also constrains and adjusts the foreground luminous brightness value corresponding to the brightness compensation amount based on the visual comfort threshold determined by the background transmittance brightness value, resulting in an updated foreground luminous brightness value. A closed-loop evaluation output module is used to redetermine the target brightness contrast relationship based on the updated foreground luminous brightness value and the background transmittance brightness value, and to perform another visual perception evaluation based on the target brightness contrast relationship. When the re-determined visual perception evaluation result meets the preset visual perception evaluation conditions, the module outputs the updated foreground luminous brightness value.

[0009] Compared with the prior art, the technical effects of this application are as follows:

[0010] This application provides a method and system for adjusting the brightness of a transparent LED display screen based on environmental perception. By acquiring historical ambient light sequences and determining future ambient light prediction results based on these sequences, the brightness adjustment is transformed from the traditional "response-based lag adjustment" to "proactive prediction adjustment". This effectively overcomes the adjustment lag problem caused by the inherent delay between the instantaneous measurement of the sensor and the system response, and avoids screen jumps and flickering caused by rapid changes in ambient light.

[0011] Furthermore, this application determines the background transmittance brightness value based on the future ambient lighting prediction results and obtains the initial foreground luminance value, constructing the contrast relationship between the two. On this basis, a visual adaptation parameter characterizing the observer's visual adaptation state is introduced, mapping the objective physical brightness data to the subjective visual perception evaluation of the human eye. This makes subsequent compensation decisions no longer rely on a single physical value, but rather on the actual perception of the human eye as the basis for regulation. This solves the technical problem of the imbalance of brightness levels between the foreground and background caused by the lack of consideration for the visual adaptation state of the human eye and the nonlinear characteristics of brightness perception, and avoids the phenomenon of graying, blurring or glare in the image.

[0012] Meanwhile, this application only triggers compensation when the visual perception evaluation result does not meet the preset visual perception evaluation conditions and the contrast relationship between light and dark exceeds the preset tolerance range for light and dark transition. The joint judgment of the two conditions effectively avoids frequent erroneous adjustments caused by slight fluctuations in ambient light, and improves the stability and anti-interference ability of brightness control.

[0013] Based on this, this application determines the brightness compensation amount according to the future ambient lighting prediction results and the contrast relationship between light and dark, and combines the visual comfort threshold determined based on the background transmission brightness value to constrain and adjust the foreground luminous brightness. While compensating, it ensures that the adjusted brightness does not exceed the range of human eye comfort, thus achieving a unity of sufficient compensation and visual comfort, avoiding secondary glare caused by over-compensation or image obscuration caused by insufficient compensation.

[0014] Finally, this application establishes a "compensation-verification-confirmation" closed-loop verification mechanism by redefining the target brightness contrast relationship and conducting visual perception evaluation again. The final brightness value is only output when the re-evaluation result meets the preset visual perception evaluation conditions, ensuring that each brightness adjustment is confirmed by the human eye perception model before it takes effect. This system-level guarantee ensures the reliability of the control results and the stability of the image quality.

[0015] In summary, this application effectively solves the technical problem of the difficulty in balancing visual comfort and clarity in existing technical solutions, and significantly improves the image layering, visual stability and long-term viewing comfort of transparent LED displays under different lighting conditions. Attached Figure Description

[0016] Figure 1 A flowchart of a method for adjusting the brightness of a transparent LED display screen based on environmental perception, provided for this application;

[0017] Figure 2 The flowchart of the temporal feature extraction and long short-term memory network prediction of ambient light in the brightness control method of transparent LED display based on environmental perception provided in this application;

[0018] Figure 3 The flowchart illustrates the process of determining the background and foreground brightness in the environment-aware transparent LED display brightness control method provided in this application, which combines optical transmission characteristics and grayscale distribution matrix.

[0019] Figure 4 The flowchart illustrates the calculation of weighted brightness contrast features based on Weber fraction and retinal illuminance in the brightness control method for a transparent LED display screen based on environmental perception provided in this application.

[0020] Figure 5The flowchart illustrates the process of obtaining initial visual perception evaluation results by inputting a human eye brightness sensitivity model into the brightness control method for a transparent LED display screen based on environmental perception provided in this application.

[0021] Figure 6 The flowchart illustrates the process of determining the brightness compensation amount based on support vector machine classification and multi-class voting mechanism in the brightness control method for a transparent LED display screen based on environmental perception provided in this application.

[0022] Figure 7 The flowchart illustrates the calculation of the upper and lower values ​​of the visual comfort threshold using a visual model in the brightness control method for a transparent LED display screen based on environmental perception provided in this application.

[0023] Figure 8 The flowchart illustrates the process of adjusting the theoretically required brightness based on a visual comfort threshold in the brightness control method for a transparent LED display screen based on environmental perception provided in this application.

[0024] Figure 9 The flowchart illustrates the process of re-determining the target brightness contrast relationship and re-evaluating the output updated brightness in the environment-aware transparent LED display brightness control method provided in this application.

[0025] Figure 10 A block diagram of a transparent LED display brightness control system based on environmental perception is provided for this application. Detailed Implementation

[0026] To enable those skilled in the art to better understand the technical solution, the present application will be described in detail below with reference to the embodiments. The description in this section is only exemplary and explanatory, and should not be used to limit the scope of protection of the present application in any way.

[0027] like Figure 1 The diagram shown is a flowchart of a method for adjusting the brightness of a transparent LED display screen based on environmental perception, provided in an embodiment of this application. The method includes:

[0028] Step S110: Obtain the historical ambient lighting sequence of the scene where the transparent LED display is located, and determine the future ambient lighting prediction result based on the historical ambient lighting sequence.

[0029] The historical ambient lighting sequence reflects the dynamic trend of lighting changes over time in the scene where the transparent LED display screen is located. It is continuously acquired through one or more ambient lighting acquisition devices deployed in the scene. These devices can be photosensors or illuminance sensors, deployed in the bezel area of ​​the transparent LED display screen or near the screen itself, with the sensing direction facing forward to collect ambient lighting data from the viewing side of the screen. In one embodiment, the sampling frequency of the ambient lighting acquisition device can be flexibly configured according to the application scenario, for example, set to once per second, once every five seconds, or once per minute.

[0030] The historical ambient lighting sequence consists of illuminance values ​​continuously collected at a preset sampling frequency, where each illuminance value is associated with a corresponding collection time. The unit of illuminance can be lux. In one specific embodiment, the system continuously collects ambient lighting data from the past 72 hours at a sampling frequency of once per second, obtaining a historical ambient lighting sequence containing 259,200 illuminance values. In another embodiment, the sampling frequency is once every 5 seconds, continuously collecting data from the past 24 hours, obtaining a historical ambient lighting sequence containing 17,280 illuminance values. It should be noted that the above sampling frequency and collection duration are only illustrative examples, and those skilled in the art can adjust them according to the severity of actual lighting changes, prediction accuracy requirements, and system storage capacity. Data acquisition methods can include real-time reading from the ambient light sensor built into the display screen, or retrieval from a historical lighting data storage module connected to the display screen control system.

[0031] After acquiring the historical ambient lighting sequence, it can be preprocessed to reduce the impact of acquisition noise, outlier values, and inconsistent sampling time intervals on subsequent prediction results. Specifically, the historical ambient lighting data can be sorted according to the acquisition time and resampled according to a preset sampling period to ensure consistent time intervals between different sampled data. For missing data in the historical ambient lighting sequence, interpolation can be used to fill in the missing data based on valid lighting data before and after the missing time. For outlier values ​​whose changes exceed a preset abnormal change range compared to adjacent sampled data, they can be corrected based on adjacent valid sampled data, thus obtaining a preprocessed historical ambient lighting sequence.

[0032] Determining future ambient lighting based on historical ambient lighting sequences aims to analyze historical ambient lighting data to identify potential changes in ambient lighting over a future period, thus providing a forward-looking basis for subsequent brightness adjustments. Specifically, an ambient lighting prediction model can be established based on historical ambient lighting sequences, and this model can be used to determine predicted ambient lighting values ​​for the future prediction period based on historical ambient lighting data.

[0033] Specifically, in the initial stage of establishing the ambient lighting prediction model and preparing training data, historical ambient lighting sample data for building the model is acquired. This historical ambient lighting sample data can come from long-term ambient lighting records in the actual application scenarios of transparent LED displays, or from historical lighting records of similar scenarios with similar lighting variation patterns. To ensure that the established ambient lighting prediction model can reflect the ambient lighting variation patterns at different time scales, the historical ambient lighting sample data can cover multiple consecutive lighting variation cycles, such as multiple day-night cycles, and can include ambient lighting data under sunny, cloudy, and other different weather conditions.

[0034] Historical ambient lighting sample data is divided into model training data and model validation data according to chronological order. The model training data is used to determine the model parameters of the ambient lighting prediction model, while the model validation data is used to evaluate the prediction accuracy of the trained model. The model training and validation data can be divided according to a predetermined time ratio; for example, earlier ambient lighting data can be used as model training data, and later ambient lighting data as model validation data, to avoid using future data in model training.

[0035] Before model training, the training data can be cleaned and its continuity processed according to the actual sampling conditions. This includes deleting duplicate sampled data, correcting abnormal sampled data, and supplementing missing sampled data to ensure that the model training data has a continuous temporal order. For ambient lighting data obtained from different sampling periods, resampling can be performed according to a uniform time scale to ensure that the data input to the model has a consistent time interval.

[0036] Next, in the stage of constructing and determining the parameters of the ambient lighting prediction model, the ambient lighting prediction model is constructed based on the preprocessed model training data. The ambient lighting prediction model can be established using a time series prediction model that can determine future ambient lighting data based on historical ambient lighting data, such as an autoregressive integral moving average model.

[0037] Taking the autoregressive integral moving average model as an example, the model training data is first analyzed for stationarity to determine whether there is a significant trend in the historical ambient light data. When the model training data shows a trend and does not meet the stationarity requirement, the model training data is differentially processed. The trend component in the data is eliminated by calculating the difference in ambient light between adjacent sampling times, and the difference order is determined according to the number of differences required to reach a stationary state.

[0038] After determining the differencing order, the autoregressive order and moving average order are determined based on the differencing ambient lighting data. Specifically, the correlation between the differencing ambient lighting data at different time intervals and the degree of independent correlation between the current sampled value and historical sampled values ​​can be analyzed to obtain candidate parameters for determining the autoregressive order and moving average order, and multiple candidate ambient lighting prediction models can be established based on combinations of multiple candidate orders.

[0039] For each candidate ambient lighting prediction model, the corresponding model parameters can be estimated using the model training data. These model parameters describe the correlation between historical ambient lighting data and the random variation characteristics within the historical data. Once the model parameters are determined, a portion of the training data can be used for model fitting, and the remaining data can be used for prediction validation. Based on the prediction errors of different candidate models, the ambient lighting prediction model suitable for the current scene can be determined. For example, the error between the predicted value of each candidate model on the validation data and the actual ambient lighting value can be calculated separately, and the candidate model with the smaller prediction error can be selected as the final ambient lighting prediction model.

[0040] In another implementation, other time series prediction models can also be used to establish the ambient lighting prediction model. For different prediction models, the parameters of the corresponding model can be trained using historical ambient lighting sample data, and the prediction accuracy of the trained model can be verified using independent validation data. When the prediction error meets the preset prediction accuracy condition, the trained model is determined as the ambient lighting prediction model.

[0041] The above method allows for the creation of an ambient lighting prediction model that matches the current lighting variation patterns based on historical ambient lighting data of the scene where the transparent LED display is located, instead of directly using fixed prediction parameters to estimate future ambient lighting.

[0042] Finally, the ambient lighting prediction model is used to determine the future ambient lighting prediction results. After obtaining the trained ambient lighting prediction model, continuous historical ambient lighting data up to the current time are input into the model. The input historical ambient lighting data can be organized according to the same sampling period and data processing method as the model training phase to ensure that the model input data and training data have the same data format.

[0043] The ambient lighting prediction model performs prediction calculations based on the input historical ambient lighting data and the model parameters obtained during the training phase, outputting the predicted ambient lighting values ​​for one or more prediction times within a future prediction period. For prediction models that can continuously output prediction values ​​for multiple future times, multiple ambient lighting prediction values ​​covering a preset future prediction period can be output at once. For prediction models that predict time-by-time, the predicted ambient lighting value for the next prediction time can be determined first based on the current historical ambient lighting data, and then the already obtained prediction values ​​and historical ambient lighting data can be used together as the basis for subsequent predictions to determine the predicted ambient lighting values ​​for subsequent prediction times in sequence.

[0044] For example, if it is determined at the current moment that the change in ambient light intensity needs to be predicted within the next 10 minutes, historical ambient light intensity data within a preset time range prior to the current moment can be input into a trained ambient light intensity prediction model. The model then calculates the predicted ambient light intensity values ​​for each future prediction moment and forms a future ambient light intensity prediction sequence according to the chronological order of the prediction moments. Each predicted intensity value can be associated with corresponding prediction time information, thus obtaining a future ambient light intensity prediction result containing multiple future prediction moments and their corresponding predicted intensity values.

[0045] After obtaining the future ambient light prediction results, the trend of future ambient light changes can be determined based on the changes in predicted light values ​​between adjacent prediction times. For example, when the predicted light values ​​at multiple consecutive prediction times gradually increase, it can be determined that the future ambient light is on an upward trend; when the predicted light values ​​at multiple consecutive prediction times gradually decrease, it can be determined that the future ambient light is on a downward trend; when the changes in predicted light values ​​between adjacent prediction times are all less than a preset change threshold, it can be determined that the future ambient light is in a relatively stable state.

[0046] Finally, the predicted ambient light values ​​and corresponding prediction time information for each predicted time in the future are used to construct the future ambient light prediction results. These results are then provided to the subsequent background transmission brightness determination and foreground luminous brightness adjustment processes, enabling the transparent LED display to adjust its brightness in advance based on future ambient light change trends, thus reducing the response lag caused by adjusting solely based on the current instantaneous ambient light.

[0047] like Figure 2The diagram shows a flowchart of temporal feature extraction and long short-term memory network prediction of ambient light in a brightness control method for a transparent LED display screen based on environmental perception, provided in an embodiment of this application. This embodiment provides a specific implementation method based on temporal feature extraction and a long short-term memory network for the process of determining the predicted ambient light value for a future prediction period using an ambient light prediction model based on historical ambient light data, as described above. The method includes the following steps:

[0048] Step S210: Extract temporal features from the historical ambient light sequence to obtain ambient light change features.

[0049] Historical ambient light sequences are continuously collected according to a preset sampling period from ambient light acquisition devices deployed around the transparent LED display screen. The historical ambient light sequences can be the pre-processed historical ambient light sequences from step S110, thereby ensuring that the input data has a continuous temporal order and consistent sampling intervals.

[0050] To characterize the variation patterns of ambient light at different time scales, temporal features were extracted from historical ambient light sequences to obtain ambient light variation characteristics. These characteristics are used to describe information such as instantaneous changes, overall trends, and periodic variations in historical ambient light sequences.

[0051] Specifically, the historical ambient light sequence can first be divided into segments using a sliding time window. Within each time window, the difference in illumination or the rate of change in illumination between adjacent sampling moments can be calculated to obtain the characteristics reflecting the degree of instantaneous change in ambient illumination. When the difference in illumination between adjacent sampling moments is large, it indicates that the ambient illumination has changed significantly within the current time period; when the difference in illumination between multiple consecutive sampling moments is small, it indicates that the ambient illumination is relatively stable within the current time period.

[0052] Furthermore, statistical characteristics, including average, maximum, minimum, standard deviation, and variation range, can be calculated for historical ambient light data within a time window to characterize the overall level and fluctuation of ambient light within the corresponding time window. For example, the variation range of light can be determined based on the difference between the maximum and minimum ambient light values ​​within a preset time window, and the light fluctuation characteristics can be determined based on the dispersion between multiple light sampling values ​​within the same time window.

[0053] To further obtain periodic information in changes in ambient light, frequency domain analysis can be performed on historical ambient light sequences. For example, a Fast Fourier Transform (FFT) can be used to transform the ambient light sequence within a time window. Based on the transformed frequency domain data, frequency components with higher energy can be identified, and the frequency values ​​and energy levels corresponding to these components can be used as periodic variation characteristics. Through this processing, periodic variations in historical ambient light data can be identified, such as periodic light changes caused by the alternation of day and night.

[0054] In another implementation, autocorrelation features can be determined based on the correlation between data from different time intervals in a historical ambient lighting sequence to characterize the temporal dependence between current and historical ambient lighting. When ambient lighting data separated by specific time intervals have a high correlation, the corresponding time intervals and correlation levels can be included as part of the ambient lighting change characteristics.

[0055] In another implementation, wavelet transform can be used to perform multi-scale analysis on the historical ambient lighting sequence, decomposing the changes in ambient lighting into components of change at different time scales, and determining the characteristics of changes in ambient lighting at different time scales based on each component of change, thereby simultaneously characterizing both short-term abrupt changes and slow changes in ambient lighting over longer time scales.

[0056] After the above processing, the obtained instantaneous change features, statistical features, periodic change features, and time-related features are combined according to their corresponding time order to form ambient light change features. For each time sampling point, the ambient light change features may include the light value, light change rate, statistical features of the time window, and corresponding periodic change features, etc., corresponding to that time sampling point; the ambient light change features corresponding to multiple consecutive time sampling points are arranged in chronological order to form a feature sequence for processing by the Long Short-Term Memory Network.

[0057] In one specific implementation, ambient light data from the past hour is used as a time window. The average light value, standard deviation, maximum value, minimum value, variation range, and rate of change between adjacent sampling points are calculated based on the light data within the time window. At the same time, the main periodic components are determined through frequency domain analysis. The above features are combined in chronological order to obtain the corresponding ambient light change feature sequence.

[0058] Step S230: Input the ambient light change characteristics into the long short-term memory network for prediction to obtain an ambient light prediction sequence covering the future prediction time period, which is used as the future ambient light prediction result.

[0059] Long Short-Term Memory (LSTM) networks are recurrent neural networks used to process time-series data. They select information input at different time steps through input gates, forget gates, and output gates, and store information that needs to be used continuously over a long period of time through cell states. Therefore, LTM networks can be used to learn the temporal dependencies in the characteristics of ambient light variation to predict future ambient light based on historical ambient light variation features.

[0060] Before inputting the ambient light change features into the Long Short-Term Memory (LSTM) network for prediction, a trained LTM network model is required. Specifically, multiple historical ambient light sample sequences can be obtained, and the historical ambient light sample sequences can be processed using the same feature extraction method as in step S210 to obtain the corresponding ambient light change feature samples.

[0061] The model uses a continuous range of ambient light variation features as input and actual ambient light data collected after that input time period as predicted labels, thus constructing a correspondence between multiple input samples and predicted labels. For example, continuous ambient light variation features over a past period can be used as input, and actual ambient light values ​​collected at multiple subsequent time points can be used as output labels. Multiple training samples are formed by shifting the input time window along the time direction.

[0062] After obtaining the training samples, a Long Short-Term Memory (LSTM) network model is constructed. The LSM network model includes an input layer, one or more LSM network hidden layers, and an output layer. The input layer is used to receive ambient light change features arranged in chronological order, the hidden layers are used to encode the feature relationships between different time steps, and the output layer is used to output the predicted ambient light values ​​for the future prediction time period.

[0063] In one implementation, the number of nodes in the hidden layer of the Long Short-Term Memory (LSTM) network can be set according to the number of training samples and the complexity of ambient lighting changes, for example, to 128 nodes; when it is necessary to improve the model's ability to fit complex ambient lighting changes, multiple hidden layers can also be used. The above-mentioned number of network layers and nodes are merely exemplary settings and do not constitute a limitation of this application.

[0064] The training samples are input into the Long Short-Term Memory (LSTM) network model for training. During training, the LTM network processes the ambient light change features corresponding to each time step in the time sequence of the training samples. It uses the input gate to determine the information that needs to be written into the cell state in the current time step, the forget gate to determine the information that needs to be retained in the historical cell state, and the output gate to determine the information that needs to be output in the current time step, thus forming the hidden state corresponding to the current time step.

[0065] The predicted ambient light value for the future prediction period is determined based on the hidden state formed at each time step. This predicted ambient light value is then compared with the actual ambient light value at the corresponding time point to determine the prediction error. Based on this error, the weight and bias parameters in the Long Short-Term Memory network are updated using a backpropagation algorithm, gradually bringing the prediction result closer to the actual ambient light value.

[0066] In one implementation, mean squared error can be used as the loss function during network training. The loss value is determined based on the difference between the predicted ambient light value and the actual ambient light value, and the network parameters are iteratively updated using an adaptive moment estimation optimization method. When the decrease in the loss value over several consecutive training cycles is less than a preset threshold, or when a preset number of training cycles is reached, network training can be stopped, and the current network parameters can be defined as the completed Long Short-Term Memory (LSTM) network model.

[0067] To verify the predictive ability of the trained Long Short-Term Memory (LSTM) network model, ambient light variation feature samples not used in model training can be used for prediction verification. The predicted ambient light values ​​output by the model are compared with the actual ambient light values ​​corresponding to the verification samples, and the prediction accuracy of the model is evaluated based on the prediction error. When the prediction error meets the preset accuracy condition, the trained LSM network model is used for actual ambient light prediction; when the prediction error does not meet the preset accuracy condition, the network structure or training parameters can be further adjusted and the model can be retrained.

[0068] After obtaining the trained Long Short-Term Memory (LSTM) network model, the ambient light change features obtained in step S210 are input into the LTM network model in chronological order. Based on the input ambient light change feature sequence, and combined with the weight and bias parameters obtained during network training, the LTM network model calculates the temporal dependencies in historical ambient light changes and outputs predicted ambient light values ​​for one or more prediction times within the future prediction period.

[0069] For example, when it is necessary to predict changes in ambient light within the next 10 minutes, the characteristic sequence of ambient light changes formed within a preset time range before the current moment can be input into a trained Long Short-Term Memory (LSTM) network model. The model then outputs the predicted ambient light values ​​for each prediction moment within the next 10 minutes. The predicted values ​​are arranged in chronological order of prediction moments and associated with the corresponding prediction time information, thus forming an ambient light prediction sequence covering the future prediction period.

[0070] In another implementation, a multi-step prediction method can be used, allowing the Long Short-Term Memory network to output ambient light prediction values ​​for multiple future time points at once; alternatively, a recursive prediction method can be used, first predicting the ambient light value for the next time point, and then combining the already obtained prediction results to determine the ambient light prediction values ​​for subsequent time points. The specific prediction method can be set according to the prediction time range and the system's computing resources.

[0071] Finally, the predicted ambient light values ​​for each future prediction time point output by the Long Short-Term Memory Network, along with the corresponding prediction time information, constitute the future ambient light prediction results. These results are then provided to the subsequent processes of determining the background transmittance brightness and adjusting the foreground luminance.

[0072] In this embodiment, instead of directly using historical ambient light values ​​to predict future lighting, the method described above first extracts ambient light change features reflecting instantaneous changes, statistical fluctuations, periodic changes, and time correlations from historical ambient light sequences. Then, it uses a long short-term memory network to learn the correlation between ambient light change features at different time scales, thereby obtaining future ambient light prediction results and providing a forward-looking basis for ambient light change for subsequent brightness control of transparent LED displays.

[0073] Step S130: Determine the background transmission brightness value corresponding to the transparent LED display screen based on the future ambient light prediction results, and obtain the initial foreground luminous brightness value corresponding to the transparent LED display screen. Determine the contrast relationship between the foreground and the background based on the background transmission brightness value and the initial foreground luminous brightness value.

[0074] Background transmission brightness is used to characterize the brightness level of ambient light entering the observer's visual range after passing through the transparent LED display screen under future ambient lighting conditions. Because the transparent LED display screen has certain optical transmission characteristics, ambient light will still enter the observer's field of vision after passing through the display screen. Therefore, the background transmission brightness value is related to the future ambient lighting prediction result and the optical characteristics of the transparent LED display screen itself. After obtaining the future ambient lighting prediction result, the corresponding background transmission brightness value can be determined for each prediction moment within the future prediction period, thereby characterizing the background brightness level under future ambient lighting changes. The specific method for determining the background transmission brightness value will be further explained in subsequent embodiments.

[0075] The initial foreground luminance value characterizes the luminance level of the foreground display content of the transparent LED display screen before the brightness adjustment of this application is performed. The foreground display content may include text, graphics, images, and other display information. The initial foreground luminance value can be obtained based on the current display state of the transparent LED display screen, for example, based on the brightness parameters or luminance driving state determined during the current display control process. For transparent LED displays containing multiple display areas, the initial foreground luminance value can also be determined separately for each display area to reflect the foreground luminance state of different areas.

[0076] After obtaining the background transmitted brightness value and the initial foreground luminance value, the contrast between the foreground and background is determined based on the brightness relationship between the two. The contrast is used to characterize the relative brightness between the displayed foreground content and the transmitted background. A significant difference in brightness between the initial foreground luminance value and the background transmitted brightness value indicates a strong distinction between the foreground and background; conversely, a closer similarity in brightness levels indicates a lower distinction between the foreground and background.

[0077] The contrast between light and dark can be represented using numerical contrast features, and the corresponding contrast can be determined separately for different display areas. The specific methods for calculating contrast and the processing methods related to the observer's visual perception will be further explained in subsequent implementation methods.

[0078] Therefore, the obtained background transmittance brightness value, initial foreground luminance value, and contrast relationship can serve as the basis for subsequent visual perception evaluation and brightness compensation, enabling subsequent brightness control to simultaneously consider the predicted ambient light and the contrast relationship between the foreground and background of the transparent LED display.

[0079] In one alternative implementation, such as Figure 3 The diagram shown is a flowchart illustrating a method for determining background and foreground brightness by combining optical transmission characteristics and a grayscale distribution matrix, according to an embodiment of this application. The method includes:

[0080] Step S310: Obtain the optical transmission characteristic parameters of the transparent LED display screen, and determine the background transmission brightness value corresponding to the transparent LED display screen based on the optical transmission characteristic parameters and the future ambient light prediction results.

[0081] The optical transmission characteristics of a transparent LED display screen are used to characterize the transmission properties of ambient light after passing through the screen. These parameters can include the overall screen transmittance, transmittance at different wavelengths, and transmittance at different incident directions. These optical transmission characteristics can be obtained during the factory calibration or system initialization process of the transparent LED display screen.

[0082] In one implementation, during the screen calibration stage, the manufacturing parameters and factory calibration data of the transparent LED display device are combined. A light intensity measurement device is used to measure the light intensity under different illumination conditions on both the incident and transmission sides of the screen. The optical transmission characteristic parameters of the screen are determined based on the correspondence between the measurement results on the incident and transmission sides. For cases requiring differentiation of different wavelengths or incident directions, the wavelength or incident direction of the incident light can be changed for measurement to obtain the corresponding transmission characteristic parameters. The determined optical transmission characteristic parameters are then correlated with the device identifier of the transparent LED display and stored in the storage module of the display control system.

[0083] When adjusting brightness, the ambient light level at the corresponding prediction time is determined based on the future ambient light prediction results. Then, the optical transmission characteristic parameters corresponding to the current transparent LED display are called in, and the corresponding background transmission brightness value is determined based on the transmission change of ambient light after passing through the transparent LED display. When the future ambient light prediction results include multiple prediction times, the corresponding background transmission brightness value is determined separately for each prediction time, thus forming a background transmission brightness sequence corresponding to the time of the future ambient light prediction results.

[0084] Step S330: Determine the total equivalent brightness value based on the future ambient light prediction results. Subtract the background transmitted brightness value from the total equivalent brightness value and then separate the initial foreground luminous brightness value by combining it with the grayscale distribution matrix of the current image to be displayed.

[0085] First, the predicted ambient light value from the future ambient light prediction results is converted into a total equivalent luminance value in luminance units by combining it with the optical transmission characteristics of the transparent LED display. Specifically, the predicted ambient light value (unit: lx) is multiplied by the effective transmission area ratio and optical transmittance of the display to obtain the total equivalent luminance value in the corresponding luminance unit (cd / m²).

[0086] Then, the brightness information attributable to the active display portion of the transparent LED display is separated from the total equivalent luminance value. Specifically, the total equivalent luminance value is divided into two parts: one part is the background transmitted luminance value through the transparent LED display (i.e., the background brightness reaching the observer's eye after ambient light passes through the screen, as determined in step S310), and the other part is the foreground brightness contributed by the transparent LED display's own emission. Since the background transmitted luminance value and the total equivalent luminance value are in the same luminance dimension, subtracting the background transmitted luminance value from the total equivalent luminance value yields the total luminance value that can be allocated to the foreground display portion under the current predicted ambient light.

[0087] After obtaining the total brightness value that can be allocated to the foreground display, it is combined with the grayscale distribution matrix of the current image to be displayed to spatially allocate the foreground luminance. The grayscale distribution matrix is ​​composed of the grayscale values ​​corresponding to each pixel position in the current image to be displayed, and is used to characterize the brightness distribution of the foreground display content at different spatial positions. Based on the grayscale ratio of each pixel position in the grayscale distribution matrix, the total brightness value that can be allocated to the foreground display is spatially allocated, so that areas with higher grayscale receive a larger proportion of foreground luminance, and areas with lower grayscale receive a smaller proportion of foreground luminance, thereby obtaining the initial foreground luminance value corresponding to the spatial structure of the current image to be displayed.

[0088] In one possible implementation, such as Figure 4 The diagram shown is a flowchart illustrating a method for calculating weighted contrast features based on Weber fraction and retinal illuminance, according to an embodiment of this application. The method includes:

[0089] Step S410: Using the relative difference calculation method defined by the Weber fraction threshold, calculate the initial image brightness contrast between the background transmittance brightness value and the initial foreground luminance brightness value.

[0090] According to Weber's Law, the degree to which the human eye perceives a difference in brightness is related to the level of background brightness. Therefore, in this embodiment, the relative difference between the foreground brightness and the background brightness is used to characterize the degree of contrast between the foreground and the background.

[0091] Specifically, the background transmitted brightness value and the initial foreground luminance value determined in the above steps are obtained. First, the brightness difference between the initial foreground luminance value and the background transmitted brightness value is calculated. Then, using the background transmitted brightness value as a normalization benchmark, the brightness difference is divided by the background transmitted brightness value to obtain the relative brightness difference between the foreground and the background. This relative brightness difference is then determined as the initial image contrast ratio. Therefore, the initial image contrast ratio not only reflects the absolute brightness difference between the foreground and the background but also reflects the magnitude of this brightness difference relative to the current background brightness level.

[0092] For example, when the background transmitted brightness is 100 candela per square meter and the initial foreground luminance is 120 candela per square meter, the brightness difference between the two is 20 candela per square meter. Normalizing this brightness difference with the background transmitted brightness of 100 candela per square meter yields a relative brightness difference of 0.20. Therefore, 0.20 is determined as the initial image contrast ratio. This visually reflects the relative prominence of the foreground brightness compared to the background brightness.

[0093] After obtaining the initial contrast ratio of the image, a pre-determined Weber fraction threshold is used to determine its discernibility. The Weber fraction threshold is used to characterize the minimum relative brightness difference required for an observer to stably distinguish the difference in brightness between the foreground and the background.

[0094] The Weber fraction threshold can be calibrated through visual recognition experiments. Specifically, different relative brightness differences between the foreground and background are set at multiple different background brightness levels, and observers are organized to evaluate the discernibility of the foreground content. The foreground recognition rate under each relative brightness difference condition is statistically analyzed, and the critical relative brightness difference that reaches the preset recognition rate requirement is determined as the Weber fraction threshold. The preset recognition rate can be set according to actual display requirements, such as 80% or 90%.

[0095] In one specific implementation, the critical relative brightness difference obtained through the above visual recognition experiment is approximately 0.15, therefore 0.15 is used as the Weber fraction threshold. Based on the aforementioned calculations, when the background transmittance is 100 candela per square meter and the initial foreground luminance is 120 candela per square meter, the initial image contrast ratio is 0.20. Further comparing the initial image contrast ratio of 0.20 with the Weber fraction threshold of 0.15, since 0.20 is greater than 0.15, it is determined that the current brightness difference between the foreground and background reaches the preset visual recognition level, and the initial image contrast ratio is used as an effective contrast feature.

[0096] If the initial image contrast ratio is lower than the Weber score threshold, it is determined that the difference in brightness between the foreground and background does not meet the preset visual recognition level. For example, if the initial image contrast ratio is 0.10, since 0.10 is less than 0.15, it is determined that the difference in brightness between the foreground and background does not meet the preset visual recognition level.

[0097] Step S430: Obtain the retinal illuminance value, calculate the corresponding visual weight value based on the retinal illuminance value, and perform multiplicative weighted processing on the initial image brightness contrast using the visual weight value to form a weighted brightness contrast feature, which serves as the brightness contrast relationship.

[0098] The retinal illuminance value is used to characterize the level of light entering the observer's eye and acting on the retina. It can be determined based on the background transmittance on the observer's side, the pupil state of the observer's eye, and preset eye optical parameters. Specifically, the pupil state corresponding to the current lighting conditions can be determined based on the background transmittance, and the illuminance level reaching the retina can be estimated by combining the pupil state and preset eye optical parameters, thereby obtaining the corresponding retinal illuminance value.

[0099] In one implementation, visual recognition experimental data under different retinal illuminance conditions are collected in advance to establish the correspondence between retinal illuminance and visual weight. Specifically, different foreground and background brightness differences are set at multiple retinal illuminance levels, and observers are organized to evaluate the recognizability of the foreground displayed content. The corresponding visual sensitivity is determined based on the visual recognition results obtained at each retinal illuminance level, and the visual sensitivity is used as the calibration benchmark for visual weight.

[0100] Based on this, the model parameters in the visual weighting model are calibrated. Specifically, multiple candidate attenuation parameters are used to calculate visual weights based on their corresponding retinal illuminance. The calculated visual weights are then compared with the visual sensitivity obtained from the aforementioned visual recognition experiments to determine the fitting error for each candidate attenuation parameter. Candidate attenuation parameters whose fitting errors meet preset accuracy requirements are selected as the attenuation parameters for the visual weighting model. When multiple candidate attenuation parameters all meet the preset accuracy requirements, the candidate attenuation parameter with the smallest fitting error is selected. The preset accuracy requirements can be set according to the visual recognition requirements of the actual display scenario.

[0101] In one implementation, the visual weighting model uses an exponential function to establish the correspondence between retinal illuminance and visual weights, and calculates the visual weights using the attenuation parameters obtained from the calibration process described above. The currently obtained retinal illuminance value is input into the visual weighting model, and the visual weight value under the current retinal illuminance condition is calculated based on the correspondence between the retinal illuminance value and the attenuation parameter. The visual weight value is used to characterize the degree to which the current retinal illuminance conditions modulate the perception of contrast.

[0102] After obtaining the visual weight value, the visual weight value is multiplied and weighted with the initial image brightness contrast obtained in step S410 to obtain the weighted brightness contrast feature. The weighted brightness contrast feature is used as the brightness contrast relationship for subsequent visual perception evaluation.

[0103] For example, in one implementation, after visual recognition experiments and parameter calibration, a visual weight model and its attenuation parameters corresponding to the current display scene are obtained. The current retinal illuminance value is input into the visual weight model to obtain a visual weight value of 0.8; if the initial image brightness contrast ratio obtained in step S410 is 0.20, then 0.20 is multiplied by 0.8 to obtain a weighted brightness contrast feature of 0.16, and 0.16 is used as the brightness contrast relationship between the current foreground and background.

[0104] Step S150: Determine the initial visual perception evaluation result based on the contrast relationship between light and dark and the visual adaptation parameters that characterize the observer's visual adaptation state.

[0105] Visual adaptation parameters characterize the visual adaptation state of an observer's eyes under current lighting conditions. These parameters can include pupil diameter, retinal illuminance level, and visual adaptation time, reflecting the degree of visual adaptation. Visual adaptation parameters can be obtained based on current ambient lighting conditions and the observer's physiological response characteristics. For example, a correspondence between ambient lighting levels and visual adaptation parameters can be pre-established. When adjusting brightness, this correspondence can be queried based on the current ambient lighting conditions to obtain visual adaptation parameters that match the current lighting environment.

[0106] When determining the initial visual perception evaluation results, the contrast relationship and visual adaptation parameters obtained in step S130 are used as inputs to the visual perception evaluation model. The visual perception evaluation model is used to establish the correspondence between the contrast relationship, the visual adaptation state, and the observer's visual perception.

[0107] In one implementation, display images under different ambient lighting conditions are pre-collected, and different foreground luminance and background transmittance are set to correspond to different brightness and darkness contrast relationships under different display conditions. Simultaneously, visual adaptation parameters corresponding to each display condition are acquired, and observers are organized to evaluate the clarity and comfort of the displayed images. Based on the observers' evaluation results, visual perception evaluation labels or scores corresponding to each group of brightness and darkness contrast relationships and visual adaptation parameters are generated, forming a visual perception evaluation sample.

[0108] The visual perception evaluation model is trained based on visual perception evaluation samples, enabling it to learn the correspondence between light and dark contrast relationships, visual adaptation parameters, and visual perception evaluation results. During training, the light and dark contrast relationships and visual adaptation parameters are used as model inputs, while the evaluation scores or labels obtained from observers are used as model outputs. By adjusting the model parameters, the error between the model output and the observer evaluation results is made to meet preset requirements, thus obtaining the trained visual perception evaluation model.

[0109] When adjusting brightness, the current contrast ratio and visual adaptation parameters are input into the trained visual perception evaluation model. The model outputs the visual perception evaluation result corresponding to the current display state based on the correspondence between the two. The visual perception evaluation result can be represented by an evaluation score, for example, set to a scoring range of 0 to 100, where a higher score indicates a higher level of visual clarity and visual comfort for the observer in the current display state; alternatively, it can be represented by evaluation levels such as clear, basically clear, blurry, or glaring.

[0110] For example, in one implementation, the current contrast ratio and visual adaptation parameters are input into a trained visual perception evaluation model, and the model outputs a visual perception evaluation result of 82 points. The pre-set visual perception evaluation condition is that the evaluation score is not lower than 80 points. Since the current evaluation result of 82 points meets the preset visual perception evaluation condition, it is determined that the current foreground luminance meets the visual perception requirements. If the evaluation result output by the model is lower than 80 points, it is determined that the current brightness setting does not meet the preset visual perception requirements, and further brightness compensation adjustments are made in subsequent steps.

[0111] In one possible implementation, such as Figure 5 The diagram shown is a flowchart illustrating how to obtain an initial visual perception evaluation result by inputting a human eye brightness sensitivity model according to an embodiment of this application. The method includes:

[0112] Step S510: Input the light-dark contrast relationship, along with the temporal adaptation state characterized by the pupil adaptation curve and the dark adaptation time constant, into the pre-established human eye light-dark sensitivity model to obtain model response data.

[0113] The brightness contrast relationship is used to characterize the relative brightness difference between the foreground content and the background environment of a transparent LED display screen. The pupil adaptation curve is used to characterize the adjustment process of the observer's pupil state with changes in ambient light. The dark adaptation time constant is used to characterize how quickly the observer's visual sensitivity recovers over time when changing from a high-brightness environment to a low-brightness environment. By inputting the above parameters into the human eye brightness sensitivity model, the model can simultaneously consider the influence of the current brightness and darkness stimulus intensity as well as the observer's visual adaptation state on the visual response.

[0114] The human eye's light-dark sensitivity model is used to establish the correspondence between light-dark contrast, pupil adaptation state, dark adaptation time characteristics, and human visual response. In one implementation, the human eye's light-dark sensitivity model employs a nonlinear response model based on the Weber-Fechner law. It uses the light-dark contrast as the input characterizing the intensity of external visual stimuli, and the pupil adaptation curve and dark adaptation time constant as the input characterizing the visual adaptation state. By performing a nonlinear transformation on the visual stimulus intensity under different input conditions, the corresponding model response data is obtained.

[0115] Before establishing a model of human eye brightness sensitivity, visual recognition experimental data were collected under different lighting conditions and different foreground and background brightness relationships. Specifically, multiple different background brightness levels were set, and multiple different foreground luminance levels were set at each background brightness level, so that the observer was in different brightness adaptation states. Under each test condition, the corresponding foreground and background brightness, the observer's pupil state, and the duration after entering the current test lighting state from the current lighting state were recorded. The observer was then organized to identify and evaluate the brightness differences of the displayed content to obtain the corresponding visual response data.

[0116] Training samples were established based on visual recognition experimental data. Each training sample included the corresponding light-dark contrast relationship, pupil adaptation curve, dark adaptation time constant, and the observer's visual response result. The observer's visual response result could be represented by foreground recognition rate, light-dark difference recognition level, or visual sensitivity score. Using the visual response results obtained from the visual recognition experiment as the model calibration benchmark, the model parameters in the human eye light-dark sensitivity model were fitted. This ensured that the difference between the model response obtained based on the light-dark contrast relationship and visual adaptation state and the experimental visual response met the preset accuracy requirements, thus obtaining the pre-established human eye light-dark sensitivity model.

[0117] The model parameters can be calibrated iteratively using multiple sets of visual recognition experimental data. For each set of experimental data, the corresponding brightness contrast relationship, pupil adaptation curve, and dark adaptation time constant are input into the model to be calibrated to obtain the corresponding model response. The model response is then compared with the visual response results in that set of experimental data, and the response error between the two is calculated. The model parameters are adjusted based on the response errors corresponding to multiple sets of experimental data, and the above process is repeated until the overall response error meets the preset accuracy requirements. In this way, the human eye brightness sensitivity model can reflect the changes in human eye visual response under different brightness contrasts and different visual adaptation states.

[0118] When evaluating visual perception, the obtained light-dark contrast relationship, pupil adaptation curve, and dark adaptation time constant are input into the calibrated human eye light-dark sensitivity model. The model performs nonlinear response calculations based on the current light-dark stimuli and visual adaptation state, and outputs the corresponding model response data. The model response data is used to characterize the degree of response of the human eye's visual system to light-dark stimuli under the current display conditions, and serves as the basis for subsequent visual perception evaluation.

[0119] Step S530: Extract response data within the cone cell response range from the model response data, and map the response data to the foveal sensitive area using a spatial weighting algorithm to obtain the initial visual perception evaluation results.

[0120] For the model response data, the corresponding response data is first extracted based on the pre-determined cone cell response range to obtain target response data related to the human eye's visual response under photopic vision conditions. The cone cell response range can be pre-determined based on the response characteristics of cone cells to light stimuli of different wavelengths and stored in the visual perception evaluation module. During evaluation, response data within the response range is extracted based on the spectral information corresponding to the model response data.

[0121] Subsequently, the extracted response data is spatially weighted to characterize the contribution of different retinal regions to visual perception when an observer gazes at a transparent LED display screen. In one embodiment, the corresponding spatial location of the retina is determined based on the observer's gaze area, and a spatial coordinate region corresponding to the foveal sensitive area is established. Weights are assigned to different spatial locations according to their distance from the center of the fovea, giving higher weights to the central foveal region and lower weights to regions farther from the foveal center, thus obtaining spatially weighted response data.

[0122] The spatially weighted response data is aggregated and processed to obtain an evaluation value characterizing the visual response level under the current display condition, and this evaluation value is determined as the initial visual perception evaluation result. The evaluation value can be normalized according to a preset evaluation range, for example, converted into an evaluation score of 0 to 100, where a higher score indicates a better visual perception under the current display conditions.

[0123] For example, when the model response data is extracted from the cone cell response range and weighted by the fovea space, and a normalized visual response evaluation value of 82 is obtained, and the preset visual perception evaluation condition is that the evaluation value is not lower than 80, then it is determined that the initial visual perception evaluation result corresponding to the current display state meets the preset visual perception evaluation condition; when the evaluation value is lower than 80, it is determined that the current display state does not meet the preset visual perception requirements, providing a basis for subsequent brightness compensation adjustment.

[0124] Step S170: When the initial visual perception evaluation result does not meet the preset visual perception evaluation conditions and the contrast relationship between light and dark exceeds the preset tolerance range for light and dark transition, determine the brightness compensation amount based on the future ambient lighting prediction result and the contrast relationship between light and dark, and constrain and adjust the foreground luminous brightness value corresponding to the brightness compensation amount based on the visual comfort threshold determined based on the background transmission brightness value, so as to obtain the updated foreground luminous brightness value.

[0125] Preset visual perception evaluation conditions are used to determine whether the current display state meets preset visual viewing requirements. These conditions can be determined in advance through visual evaluation experiments before system deployment. Specifically, multiple different ambient lighting conditions and different combinations of foreground and background brightness are set to create multiple different display states on the transparent LED display screen, and corresponding visual perception evaluation results are obtained for each state. Simultaneously, multiple observers are organized to subjectively evaluate the image clarity and visual comfort under each display state. Based on the observer evaluation results, the range of visual perception evaluation results corresponding to the display state that meets normal viewing requirements is determined, and this range is defined as the preset visual perception evaluation conditions.

[0126] For example, when using evaluation scores as the visual perception evaluation result, the acceptance of the display state corresponding to different evaluation scores by multiple observers can be statistically analyzed, and the lowest evaluation score that reaches the preset acceptance ratio can be determined as the visual perception evaluation threshold. In one specific implementation, when the visual perception evaluation result reaches 80 points, at least a preset proportion of observers believe that the current display state meets normal viewing needs, so 80 points is determined as the visual perception evaluation threshold; when the initial visual perception evaluation result is lower than 80 points, it is determined that the initial visual perception evaluation result does not meet the preset visual perception evaluation conditions. 80 points is merely an illustrative example; in different display scenarios, the visual perception evaluation threshold can be re-determined based on the corresponding observer evaluation data.

[0127] The preset tolerance range for light and dark transitions is used to limit the range of light and dark variations between the foreground and background that an observer can accept in the current display scene. The preset tolerance range for light and dark transitions can also be determined through pre-conducted visual adaptation experiments.

[0128] Specifically, multiple combinations of foreground luminance and background transmittance were set under different ambient lighting conditions to create varying contrasts. Under each test condition, the foreground luminance or background luminance was varied, allowing the observer to experience different degrees of brightness variation. The observer's evaluation of image sharpness, perceived brightness abruptness, and visual comfort was then obtained.

[0129] Based on observer evaluations obtained under multiple test conditions, a range of light and dark contrast relationships that simultaneously meet preset sharpness requirements and do not cause significant visual discomfort is determined. The lower and upper limits of this range are then defined as the boundaries of the preset tolerable range for light and dark transitions. Specifically, the evaluation results from multiple observers can be statistically analyzed. If, within a certain range of light and dark contrast relationships, a preset proportion of observers consider the foreground content stably recognizable without noticeable brightness jumps or visual discomfort, then this range is defined as the preset tolerable range for light and dark transitions.

[0130] For example, in a specific display scenario, if the acceptable range of brightness contrast is determined to be 5:1 to 20:1 through the aforementioned visual adaptation experiment, then 5:1 is taken as the lower limit of the preset brightness transition tolerance range, and 20:1 is taken as the upper limit. When the current brightness contrast is lower than 5:1, it indicates that the difference in brightness between the foreground and background is insufficient, which may affect the recognition of foreground content; when the current brightness contrast is higher than 20:1, it indicates that the difference in brightness between the foreground and background is too large, which may cause the observer to experience obvious brightness jumps or visual discomfort. It should be noted that the above 5:1 to 20:1 is only an illustrative example. In different application scenarios, different preset brightness transition tolerance ranges can be determined based on the corresponding visual adaptation experiment results.

[0131] After obtaining the initial visual perception evaluation result and the current brightness-dark contrast relationship, they are compared with the preset visual perception evaluation conditions and the preset brightness-dark transition tolerance range, respectively. Only when the initial visual perception evaluation result does not meet the preset visual perception evaluation conditions, and the current brightness-dark contrast relationship is lower than the lower limit of the preset brightness-dark transition tolerance range or higher than the upper limit of the preset brightness-dark transition tolerance range, is it determined that the current display state requires brightness compensation.

[0132] After the above triggering conditions are met, the brightness compensation amount is determined based on the future ambient lighting prediction results and the current brightness-dark contrast relationship. The future ambient lighting prediction results are used to characterize the changing trend of ambient lighting within the predicted time period, and the brightness-dark contrast relationship is used to characterize the degree of brightness imbalance between the foreground and background. The brightness compensation amount determined based on these two factors is used to adjust the foreground luminescence brightness so that the adjusted foreground luminescence brightness can adapt to the predicted changes in ambient lighting and improve the current brightness-dark contrast state.

[0133] After determining the brightness compensation amount, the corresponding visual comfort threshold is determined based on the background transmittance brightness value. The visual comfort threshold is used to limit the allowable range of foreground luminance brightness after adjustment, so as to avoid excessive increase or decrease in foreground luminance brightness during the compensation process.

[0134] When adjusting the brightness, the current foreground luminance is adjusted according to the determined brightness compensation amount, and the adjusted foreground luminance is constrained by the visual comfort threshold so that the adjusted foreground luminance is within a visual comfort range that is compatible with the current background transmittance, thereby obtaining the updated foreground luminance value.

[0135] By simultaneously setting preset visual perception evaluation conditions and preset tolerance ranges for brightness transitions, frequent brightness adjustments can be avoided due to an abnormality in a single evaluation indicator. Brightness compensation is only performed when the current display state fails to meet the preset visual perception requirements, and the brightness relationship between the foreground and background exceeds the range of changes acceptable to the observer, thereby improving the targeting and stability of the brightness control process.

[0136] Specifically, such as Figure 6 The diagram shown is a flowchart of a method for determining a brightness compensation amount according to an embodiment of this application. The method includes:

[0137] Step S610: Construct the future ambient lighting prediction results and the brightness contrast relationship as the input feature vector of the support vector machine.

[0138] Support Vector Machines (SVMs) are used to determine the brightness compensation requirement category corresponding to the current display state based on the current ambient lighting conditions and the contrast between the foreground and background of a transparent LED display. To enable SVMs to classify using both future ambient lighting predictions and contrast relationships, these two elements need to be converted into a unified numerical feature form, constructing an input feature vector that the SVM can directly process.

[0139] Specifically, feature information characterizing future ambient lighting conditions is extracted from the future ambient lighting prediction results, such as the average light intensity, maximum light intensity, and light variation range corresponding to the future ambient lighting prediction results; simultaneously, weighted brightness-darkness contrast feature values ​​are obtained from the brightness-darkness contrast relationship. The above features are then subjected to numerical normalization to reduce the impact of differences in the dimensions and numerical ranges between different features on subsequent classification results.

[0140] Subsequently, following a preset feature arrangement order, the normalized ambient lighting features and weighted brightness-to-darkness contrast features are concatenated to form an input feature vector containing multiple numerical features. For example, the average light intensity, maximum light intensity, light change amplitude, and current weighted brightness-to-darkness contrast feature value corresponding to the future ambient lighting prediction result can be sequentially arranged to form the input feature vector. In this way, the future ambient lighting state and the brightness-to-darkness contrast state of the current display screen are transformed into an input sample in the support vector machine's feature space. This allows the support vector machine to classify the current display state based on the correspondence between this input sample and different pre-learned compensation requirement categories.

[0141] The specific selection method for each ambient lighting feature and the weighted brightness contrast feature corresponding to the future ambient lighting prediction results can be determined according to the actual brightness adjustment requirements of the transparent LED display screen, and should be consistent with the feature dimensions and arrangement order used in the support vector machine training stage.

[0142] Step S630: Use the hyperplane segmentation boundary obtained by training with radial basis kernel function and soft margin relaxation factor to classify the input feature vector and obtain the initial classification result.

[0143] Support Vector Machines (SVMs) are used to classify and determine the brightness compensation requirements corresponding to the current display state based on the correspondence between changes in ambient lighting and the current foreground and background brightness contrast. During the model training phase, SVMs use ambient lighting data and brightness contrast obtained from historical brightness adjustment processes as input samples, and the compensation category to which the corresponding actual brightness adjustment result belongs as the sample label. It learns the correspondence between different input features and brightness compensation requirement categories from historical adjustment data.

[0144] Specifically, firstly, multiple sets of historical training samples are acquired. Each set includes the ambient lighting characteristics determined under historical lighting conditions, the corresponding brightness contrast relationship, and the actual brightness compensation results used. Based on the brightness compensation results, multiple compensation categories are pre-defined, such as no compensation, minor compensation, medium compensation, and major compensation. Each historical training sample is then labeled according to its compensation category. This process forms a training sample set with category labels.

[0145] During model training, each training sample is input into a support vector machine. Since the brightness compensation requirements under different ambient lighting conditions and contrast states are usually not a simple linear relationship, a radial basis function (RBF) kernel is used to map the nonlinear relationship between input features to a high-dimensional feature space. This allows training samples of different compensation categories to form a more easily distinguishable category distribution within this feature space. Simultaneously, a soft-margin relaxation factor is set to allow a small number of training samples near the category boundary to have classification errors, thus avoiding excessive shift in the classification boundary due to abnormal lighting data or measurement noise.

[0146] Based on this, the Support Vector Machine (SVM) determines the maximum margin between categories according to the distribution of training samples of different compensation categories in the high-dimensional feature space, and determines the segmentation position between each category based on the support vectors near the category boundary, thus forming a hyperplane segmentation boundary used to distinguish different brightness compensation requirement categories. The hyperplane segmentation boundary is used to characterize the boundary position of different brightness compensation categories in the feature space, and it is determined by the trained model parameters and stored in the SVM model. When a new input feature is located in the segmentation region corresponding to a certain category, its brightness compensation category can be determined according to the positional relationship of the input feature relative to the hyperplane segmentation boundary.

[0147] In the actual control process, the input feature vector obtained in step S610 is input into the trained support vector machine. The support vector machine calculates the region to which the input feature vector belongs in the feature space based on the pre-trained hyperplane segmentation boundary and outputs the corresponding compensation category as the initial classification result. For example, when the input feature vector is located in the segmentation region corresponding to the no-compensation category, the initial classification result is determined as no compensation is required; when the input feature vector is located in the segmentation region corresponding to the small-compensation, medium-compensation, or large-compensation categories, it is determined as the corresponding compensation category. Thus, the ambient light and brightness contrast state is converted into the classification result required for subsequent brightness compensation amount determination.

[0148] Step S650: Based on the initial classification results, a multi-classification voting mechanism is used to conduct voting statistics in the preset candidate compensation interval, and the compensation value corresponding to the candidate compensation interval with the highest number of votes is determined as the brightness compensation amount.

[0149] Before establishing the support vector machine model, candidate compensation intervals are determined based on the brightness adjustment resolution and actual brightness adjustment range of the transparent LED display screen. Specifically, actual brightness adjustment data under different ambient lighting conditions can be collected, and the brightness adjustment amplitude corresponding to the preset visual perception evaluation conditions can be statistically analyzed. The statistically obtained brightness adjustment amplitude is then divided into multiple continuous compensation intervals. For example, based on historical adjustment data, the brightness adjustment amplitude can be divided into multiple candidate compensation intervals corresponding to no compensation, small compensation, medium compensation, and large compensation. The boundaries of each candidate compensation interval can be determined based on the differences in visual perception evaluation results corresponding to adjacent compensation amplitudes, so that adjacent intervals can correspond to different levels of brightness compensation requirements.

[0150] For cases requiring multiple candidate compensation categories, a one-to-one approach is used to establish multiple support vector machine classifiers, with each classifier classifying two compensation categories. During operation, the input feature vector obtained in step S610 is input into each classifier, and voting statistics are performed based on the classification results of each classifier. The compensation category with the most votes is determined as the initial compensation category. Subsequently, the candidate compensation interval corresponding to this compensation category is determined, and the preset representative value of this candidate compensation interval is used as the brightness compensation amount. The representative value can be set to the median of the corresponding interval based on the actual brightness adjustment resolution of the display screen, so that the output brightness compensation amount can be directly converted into the brightness adjustment command of the display screen.

[0151] By using the above method, a correspondence between ambient light, contrast and brightness compensation requirements is established using historical display adjustment data. Support vector machines are then used to classify different compensation requirements, thereby determining the corresponding brightness compensation amount based on the current predicted ambient light and display status.

[0152] Furthermore, such asFigure 7 As shown, this application also provides a process for determining the visual comfort threshold based on background transmittance brightness values, which specifically includes:

[0153] Step S710: Input the background transmission brightness value into the preset visual model, and calculate the range of human eye visual comfort brightness corresponding to the background transmission brightness value through the visual model.

[0154] A pre-defined visual model is used to establish the correspondence between background transmittance brightness values ​​and the range of brightness comfortable for human vision. This visual model is pre-established using visual experimental data during the system initialization phase.

[0155] Specifically, during the model building process, different background transmission brightness conditions were selected, and multiple different foreground luminous brightness levels were set for each background transmission brightness condition. Observers viewed the test images displayed on the transparent LED display screen and evaluated the comfort level of the current display effect. The test images included at least text, graphics, or images with different foreground and background brightness differences, so that the evaluation results could reflect the viewing experience in actual transparent display scenarios.

[0156] For each combination of background transmittance and foreground emission brightness, the observer's comfort evaluation results are obtained, and the evaluation results are categorized into comfortable, dim, bright, or glare states according to preset evaluation rules. For example, when multiple observers agree that the current brightness is sufficient for continuous viewing and there is no obvious lack of brightness or glare, the brightness is marked as comfortable; when observers evaluate it as significantly dim, it is marked as below the comfortable range; when observers evaluate it as significantly bright or glare-inducing, it is marked as above the comfortable range.

[0157] For each background transmittance brightness condition, the corresponding lower and upper limits of comfortable brightness are determined based on the evaluation results of multiple observers. Specifically, the lower limit of comfortable brightness is defined as the position in the foreground luminance that transitions from "dim" to "comfortable" below the lower limit, and the upper limit of comfortable brightness is defined as the position in the foreground luminance that transitions from "comfortable" to "bright" or "glare" above the upper limit. By statistically analyzing the evaluation results of multiple observers, the brightness range in which a preset majority of observers rate it as comfortable is defined as the visually comfortable brightness range under the corresponding background transmittance brightness.

[0158] After obtaining the visual comfort brightness range under multiple background transmittance conditions, a training sample set for the visual model is constructed using background transmittance brightness as the model input and the corresponding lower and upper comfort brightness limits as the model output labels. A regression model is used to fit the training sample set, enabling the model to learn the nonlinear correspondence between changes in background transmittance brightness and the upper and lower limits of visual comfort brightness. During model training, a portion of the collected samples are used as training samples, and the remaining samples are used as validation samples. The model parameters are iteratively adjusted by comparing the error between the model's predicted upper and lower comfort brightness limits and those determined by visual experiments. When the prediction error on the validation samples reaches a preset requirement, the trained visual model is obtained.

[0159] During the operation phase, the background transmittance brightness value determined in step S310 is input into the trained visual model to obtain the lower limit and upper limit of comfortable brightness under the current background transmittance brightness conditions, and the lower limit and upper limit of comfortable brightness constitute the visual comfortable brightness range.

[0160] For example, in visual experiments, upper and lower limits of comfortable brightness are obtained under multiple different background transmittance brightness conditions, and these experimental data are used to train a visual model. When the current background transmittance brightness value is input, the visual model outputs a lower limit and an upper limit of comfortable brightness that match the current background brightness, based on the learned correspondence between background brightness and the comfortable brightness range.

[0161] Step S730: Use the upper and lower limits of the visual comfort brightness range as the visual comfort threshold.

[0162] The lower limit of comfortable brightness output by the visual model is determined as the lower limit of the visual comfort threshold, and the upper limit of comfortable brightness is determined as the upper limit of the visual comfort threshold, thus forming the visual comfort threshold under the current background transmittance brightness condition.

[0163] The visual comfort threshold is not a fixed brightness value, but rather changes dynamically with variations in background transmitted brightness. When the background transmitted brightness changes, the changed background transmitted brightness is re-input into the visual model to obtain the corresponding upper and lower limits of visual comfort brightness. Based on this, the visual comfort threshold is updated, enabling subsequent brightness adjustments to adapt to changes in the observer's visual adaptation state under different ambient lighting conditions.

[0164] The above method transforms the subjective comfort evaluation of the human eye obtained from visual experiments into calculable upper and lower limits of visual comfort brightness. The trained visual model is then used to establish a mapping relationship between background transmission brightness and visual comfort threshold, thereby providing dynamic upper and lower boundaries for subsequent constraint adjustment of foreground luminous brightness.

[0165] Furthermore, such as Figure 8The diagram shown is a flowchart of a method for constraining and adjusting theoretically required brightness based on a visual comfort threshold, according to an embodiment of this application. The method specifically includes:

[0166] Step S810: Obtain the initial foreground luminance value and luminance compensation amount, and sum the initial foreground luminance value and luminance compensation amount to obtain the theoretical required luminance value.

[0167] In this embodiment, the initial foreground luminance value represents the foreground luminance of the transparent LED display screen before this luminance compensation is performed, and the luminance compensation amount represents the adjustment range of the foreground luminance determined based on the current ambient light prediction result and the contrast relationship between light and dark. To enable direct superposition of the two, the luminance compensation amount and the initial foreground luminance value use the same luminance unit, and the luminance direction corresponding to the initial foreground luminance value is used as the reference. A positive luminance compensation amount represents an increase in foreground luminance, and a negative luminance compensation amount represents a decrease in foreground luminance.

[0168] Specifically, the initial foreground luminance value in the current display state is obtained, along with the luminance compensation amount determined in the preceding steps. The luminance compensation amount is then added as an increment relative to the initial foreground luminance value, causing a corresponding change in the initial foreground luminance value along the luminance compensation direction. For example, when the initial foreground luminance value is 200 candela per square meter and the luminance compensation amount is 50 candela per square meter, it indicates that the current luminance needs to be increased by 50 candela per square meter. In this case, the two are summed to obtain a theoretical required luminance value of 250 candela per square meter. When the initial foreground luminance value is 200 candela per square meter and the luminance compensation amount is -30 candela per square meter, it indicates that the current luminance needs to be decreased by 30 candela per square meter. The summation yields a theoretical required luminance value of 170 candela per square meter.

[0169] Therefore, the theoretical required brightness value represents the target brightness obtained after adjusting the initial foreground luminance based on the current brightness compensation requirement, without considering the visual comfort threshold constraint. This theoretical required brightness value only reflects the adjustment result required by the brightness compensation mechanism and has not yet been determined whether it is within the visual comfort range corresponding to the current background transmittance brightness. Therefore, it is used as the input for subsequent comparison and constraint adjustment with the visual comfort threshold.

[0170] Step S830: Obtain the upper and lower limits of the visual comfort threshold, and compare the theoretical required brightness value with the upper and lower limits.

[0171] In this embodiment, the visual comfort threshold includes an upper limit and a lower limit of the visual comfort brightness range corresponding to the current background transmittance brightness value. The upper limit limits the foreground luminance brightness to a level acceptable to the human eye under the current background light environment, while the lower limit limits the foreground luminance brightness to a level that maintains the visibility of foreground content under the current background light environment. The upper and lower limits are determined based on the current background transmittance brightness value by the aforementioned visual comfort threshold determination process and serve as constraint boundaries corresponding to the current display state during this brightness adjustment.

[0172] Specifically, after obtaining the theoretical required brightness value, the visual comfort threshold corresponding to the current background transmittance brightness value is read, and the upper and lower limits are obtained respectively. Since the theoretical required brightness value, upper limit, and lower limit are all used to characterize the foreground luminous brightness of the transparent LED display, the same brightness unit and value benchmark are used for comparison. Subsequently, the range of the theoretical required brightness value is determined by comparing the theoretical required brightness value with the upper limit and comparing the theoretical required brightness value with the lower limit.

[0173] Specifically, when the theoretically required brightness value is higher than the upper limit, it is determined to exceed the current visual comfort range; when the theoretically required brightness value is lower than the lower limit, it is determined to be lower than the current visual comfort range; when the theoretically required brightness value is neither lower than the lower limit nor higher than the upper limit, it is determined to be within the current visual comfort range. The above comparison results are used to determine whether boundary constraints need to be applied to the theoretically required brightness value, and if constraints are needed, which boundary value from the visual comfort thresholds should be used for adjustment.

[0174] Step S850: If the theoretical required brightness value is greater than the upper limit value, then adjust the theoretical required brightness value to the upper limit value; if the theoretical required brightness value is less than the lower limit value, then adjust the theoretical required brightness value to the lower limit value; otherwise, use the theoretical required brightness value as the updated foreground luminous brightness value.

[0175] Based on the comparison results obtained in step S830, boundary constraints are applied to the theoretical required brightness value. When the theoretical required brightness value is greater than the upper limit of the visual comfort threshold, it indicates that the theoretical required brightness obtained according to the brightness compensation amount has exceeded the visual comfort range corresponding to the current background light environment. At this time, the theoretical required brightness value is adjusted to the upper limit value to limit the foreground luminous brightness from not exceeding the maximum allowable value of the current visual comfort range.

[0176] When the theoretical required brightness value is less than the lower limit of the visual comfort threshold, it indicates that the theoretical required brightness obtained according to the brightness compensation amount is lower than the visual comfort range corresponding to the current background light environment. At this time, the theoretical required brightness value is adjusted to the lower limit to avoid insufficient brightness of the foreground light source, which would result in insufficient difference in brightness between the foreground content and the background.

[0177] When the theoretical required brightness value is not lower than the lower limit and not higher than the upper limit, it indicates that the theoretical required brightness value is already within the brightness range limited by the visual comfort threshold. At this time, no further adjustment is needed, and the theoretical required brightness value is directly determined as the updated foreground luminous brightness value.

[0178] Through the above boundary constraint processing, the theoretical required brightness obtained by the brightness compensation mechanism is limited to the range defined by the current visual comfort threshold. This allows the brightness compensation to respond to changes in ambient light and changes in the contrast between the foreground and background, while also preventing the compensated foreground luminance from exceeding the acceptable brightness range for the human eye. This results in an updated foreground luminance value for subsequent display control.

[0179] Step S190: Based on the updated foreground luminance value and background transmission luminance value, redetermine the target brightness contrast relationship, and perform visual perception evaluation again based on the target brightness contrast relationship. When the visual perception evaluation result is redetermined and meets the preset visual perception evaluation conditions, output the updated foreground luminance value.

[0180] After obtaining the updated foreground luminance value, it is used together with the background transmitted luminance value as a parameter to characterize the display state after brightness adjustment, and the target brightness contrast relationship between the foreground and background is redefined. The target brightness contrast relationship is used to reflect the actual brightness relationship formed between the foreground luminance and the background transmitted luminance of the transparent LED display after brightness compensation. Its determination process corresponds to the aforementioned determination process of the brightness contrast relationship, so that the brightness contrast relationship before and after adjustment can be compared using a consistent evaluation scale.

[0181] After determining the target's brightness contrast, a visual perception evaluation is performed again based on the adjusted display state. Specifically, the target's brightness contrast and the visual adaptation parameters corresponding to the current observation state are used as inputs for the re-evaluation. The aforementioned visual perception evaluation process is then used to evaluate the display state after brightness adjustment, resulting in a re-determined visual perception evaluation result. This re-determined visual perception evaluation result characterizes whether the observer's visual perception of the current foreground and background brightness relationship reaches a preset acceptable level after adopting the updated foreground luminance value.

[0182] Furthermore, the re-determined visual perception evaluation result is compared with the preset visual perception evaluation conditions. When the re-determined visual perception evaluation result meets the preset visual perception evaluation conditions, it indicates that the display state after brightness compensation has reached the expected visual perception requirements. At this time, the updated foreground luminance value is determined as the final brightness control result and output to the display driver module of the transparent LED display screen to drive the display screen to display according to the updated foreground luminance value.

[0183] By redefining the contrast between light and dark after brightness compensation and evaluating visual perception, the actual visual improvement effect of brightness adjustment can be verified in a closed loop. This avoids directly outputting the brightness adjustment result based solely on a predetermined brightness compensation amount, thereby improving the consistency between the brightness control result and the target visual perception.

[0184] In one alternative embodiment, such as Figure 9 The diagram shown is a flowchart illustrating a method for redetermining the target brightness contrast and re-evaluating the output updated brightness, as provided in an embodiment of this application. The method includes:

[0185] Step S910: Using the relative difference calculation method defined by the Weber fraction threshold, calculate the target image brightness contrast between the updated foreground luminance value and the background transmittance value, as the target brightness contrast relationship.

[0186] After adjusting the foreground luminance, in order to verify whether the adjusted display effect meets expectations, it is necessary to redetermine the contrast relationship between the foreground and the background based on the updated foreground luminance value.

[0187] Specifically, the updated foreground luminance and background transmittance values ​​obtained in the preceding steps are acquired. Using the same relative difference calculation method as the initial image contrast ratio, the relative luminance difference between the updated foreground luminance and background transmittance values ​​is determined. Then, a Weber fraction threshold is used to determine whether this relative luminance difference reaches a level of visual recognition. When the relative luminance difference reaches or exceeds the Weber fraction threshold, the corresponding relative luminance difference is determined as the target image contrast ratio, and this target image contrast ratio is used as the target contrast ratio.

[0188] For example, when the updated foreground luminance is 360.6 candela per square meter and the background transmittance is 120.2 candela per square meter, the luminance difference is 240.4 candela per square meter. This luminance difference is normalized relative to the background transmittance, resulting in a relative luminance difference of 2.0. If this relative luminance difference reaches a preset Weber fraction threshold, then 2.0 is determined as the target image's contrast ratio, thus obtaining the adjusted target contrast relationship.

[0189] By using the same calculation method as the initial image's brightness and contrast, the brightness and contrast relationship before and after the compensation adjustment can have a consistent evaluation scale, which facilitates the subsequent visual evaluation of the brightness compensation effect.

[0190] It should be noted that both this step and step S410 use the relative difference calculation method defined by the Weber fraction threshold. However, the calculation in step S410 is used to obtain the image brightness contrast in the initial state to determine whether the current image needs brightness compensation. The calculation in step S910 uses the same Weber fraction relative difference calculation method, but the input is the foreground luminance value after compensation and update. Its purpose is to verify, using the same metric, whether the image brightness contrast after compensation has met the preset visual evaluation conditions. The use of the same contrast quantification method in both steps ensures the consistency and comparability of the image brightness contrast evaluation results before and after compensation, thus providing a reliable basis for the closed-loop verification mechanism.

[0191] Step S930: Input the target brightness contrast relationship along with the current pupil adaptation curve state back into the human eye brightness sensitivity model. After processing by the human eye brightness sensitivity model, obtain the target visual perception evaluation result in the foveal sensitive area.

[0192] After determining the target's brightness contrast, it is input together with the current pupil adaptation curve state into the human eye brightness sensitivity model established in the previous steps to re-evaluate the visual perception of the display effect after brightness compensation.

[0193] The current pupil adaptation curve state is used to characterize the observer's dynamic pupil adaptation state under the current ambient lighting conditions. It can be determined based on the current ambient lighting change process and the corresponding pupil response relationship. For example, ambient lighting change data over a period of time can be input into a pre-established pupil adaptation curve to determine the pupil adaptation state at the current moment, and this state can be converted into state parameters that the model can recognize.

[0194] After receiving the target brightness contrast and the current pupil adaptation curve, the human eye brightness sensitivity model, based on a pre-established mapping relationship between brightness stimuli and visual responses, jointly processes the visual stimulus intensity corresponding to the target brightness contrast and the current pupil adaptation state to obtain the corresponding model response data. Subsequently, based on a pre-determined foveal sensitive area, the model response data corresponding to the foveal sensitive area is determined from the foveal sensitive area, and the visual response data is weighted and summarized to obtain the target visual perception evaluation result used to characterize the degree of visual perception of the current display effect.

[0195] For example, if the visual response data output by the human eye's luminance sensitivity model is weighted by the foveal sensitive area and yields an evaluation value of 85, then 85 is determined as the target visual perception evaluation result. This evaluation result uses the same evaluation scale as the aforementioned initial visual perception evaluation result in order to determine whether the visual perception has been improved before and after luminance compensation.

[0196] Step S950: If the target visual perception evaluation result meets the preset visual perception evaluation conditions, then output the updated foreground luminance value.

[0197] The target visual perception evaluation result is compared with the preset visual perception evaluation conditions determined in the aforementioned steps. When the target visual perception evaluation result meets the preset visual perception evaluation conditions, it indicates that the display effect after brightness compensation has reached the expected visual perception requirements. At this time, the updated foreground luminance value is determined as the final foreground luminance value and output to the display control module or drive circuit of the transparent LED display screen to drive the transparent LED display screen to display according to the brightness value.

[0198] For example, when the preset visual perception evaluation condition is that the visual perception evaluation result is not lower than 80 points, if the target visual perception evaluation result obtained in step S930 is 85 points, then it is determined that the target visual perception evaluation result meets the preset visual perception evaluation condition, and the corresponding updated foreground luminous brightness value is output.

[0199] When the target visual perception evaluation result does not meet the preset visual perception evaluation conditions, the updated foreground luminance value will not be output. Instead, the brightness compensation and visual perception evaluation will be re-executed according to the preset brightness control process until the brightness adjustment result that meets the preset visual perception evaluation conditions is obtained.

[0200] Furthermore, an exemplary embodiment of this application also provides a brightness control system for a transparent LED display screen based on environmental perception.

[0201] Figure 10 A block diagram illustrating an exemplary embodiment of the ambient-aware transparent LED display brightness control system of this application is shown. (Reference) Figure 10 An environmentally-aware transparent LED display brightness control system 1000 according to an exemplary embodiment of this application may include: an ambient light prediction module 1010, a brightness-dark contrast relationship determination module 1030, an initial visual perception evaluation module 1050, a brightness compensation adjustment module 1070, and a closed-loop evaluation output module 1090.

[0202] Specifically, the ambient lighting prediction module 1010 can be used to acquire the historical ambient lighting sequence of the scene where the transparent LED display is located, and determine the future ambient lighting prediction result based on the historical ambient lighting sequence; the brightness contrast relationship determination module 1030 can be used to determine the background transmission brightness value corresponding to the transparent LED display based on the future ambient lighting prediction result, and acquire the initial foreground luminous brightness value corresponding to the transparent LED display, and determine the brightness contrast relationship between the foreground and the background based on the background transmission brightness value and the initial foreground luminous brightness value; the initial visual perception evaluation module 1050 can be used to determine the initial visual perception evaluation result based on the brightness contrast relationship and visual adaptation parameters characterizing the observer's visual adaptation state; the brightness compensation adjustment module 107 The 0 module can be used to determine the brightness compensation amount based on the future ambient lighting prediction result and the brightness contrast relationship when the initial visual perception evaluation result does not meet the preset visual perception evaluation conditions and the brightness contrast relationship exceeds the preset brightness transition tolerance range. Based on the visual comfort threshold determined based on the background transmission brightness value, the foreground luminous brightness value corresponding to the brightness compensation amount is constrained and adjusted to obtain the updated foreground luminous brightness value. The closed-loop evaluation output module 1090 can be used to redetermine the target brightness contrast relationship based on the updated foreground luminous brightness value and the background transmission brightness value, and perform visual perception evaluation again based on the target brightness contrast relationship. When the visual perception evaluation result determined again meets the preset visual perception evaluation conditions, the updated foreground luminous brightness value is output.

[0203] Since the functional modules of the transparent LED display brightness control system based on environmental perception in this application are the same as those in the above method implementation, they will not be described again here.

[0204] It should be noted that, in this document, the terms "comprising," "including," and any other variations are intended to cover non-exclusive inclusion, such that a process, method, article, or apparatus that comprises a list of elements includes not only those elements but also other elements not expressly listed, or elements inherent to such a process, method, article, or apparatus. Specific examples have been used in this document to illustrate the principles and implementation methods of the technical solutions of this application. The above examples are only for the purpose of helping to understand the methods and core ideas of this application. The above are merely preferred embodiments of this application. It should be pointed out that, due to the limitations of written expression and the objective existence of infinite specific structures, those skilled in the art can make several improvements, modifications, or changes without departing from the principles of this application, and can also combine the above technical features in an appropriate manner; these improvements, modifications, changes, or combinations, or the direct application of the concept and technical solutions of this application to other situations without modification, should all be considered within the scope of protection of this application.

Claims

1. A method for adjusting the brightness of a transparent LED display screen based on environmental perception, characterized in that, The method includes: Obtain the historical ambient lighting sequence of the scene where the transparent LED display is located, and determine the future ambient lighting prediction result based on the historical ambient lighting sequence; The background transmission brightness value corresponding to the transparent LED display screen is determined based on the future ambient light prediction result, and the initial foreground luminous brightness value corresponding to the transparent LED display screen is obtained. The contrast relationship between the foreground and the background is determined based on the background transmission brightness value and the initial foreground luminous brightness value. Based on the aforementioned light-dark contrast relationship and the visual adaptation parameters characterizing the observer's visual adaptation state, the initial visual perception evaluation result is determined. When the initial visual perception evaluation result does not meet the preset visual perception evaluation conditions and the brightness contrast relationship exceeds the preset brightness transition tolerance range, the brightness compensation amount is determined according to the future ambient lighting prediction result and the brightness contrast relationship, and the foreground luminous brightness value corresponding to the brightness compensation amount is constrained and adjusted according to the visual comfort threshold determined based on the background transmission brightness value, so as to obtain the updated foreground luminous brightness value. The target brightness contrast is re-determined based on the updated foreground luminance value and the background transmittance value, and a visual perception evaluation is performed again based on the target brightness contrast. When the visual perception evaluation result is satisfied with the preset visual perception evaluation conditions, the updated foreground luminance value is output.

2. The method according to claim 1, characterized in that, The determination of future ambient lighting prediction results based on the historical ambient lighting sequence includes: Temporal features are extracted from the historical ambient lighting sequence to obtain ambient lighting variation features; The ambient light change characteristics are input into a long short-term memory network for prediction, resulting in an ambient light prediction sequence covering the future prediction period, which is used as the future ambient light prediction result.

3. The method according to claim 1, characterized in that, The step of determining the background transmission brightness value corresponding to the transparent LED display screen based on the future ambient lighting prediction result, and obtaining the initial foreground luminous brightness value corresponding to the transparent LED display screen, includes: Obtain the optical transmission characteristic parameters of the transparent LED display screen, and determine the background transmission brightness value corresponding to the transparent LED display screen based on the optical transmission characteristic parameters and the future ambient light prediction results; The total equivalent brightness value is determined based on the predicted future ambient light. After subtracting the background transmission brightness value from the total equivalent brightness value, the initial foreground luminous brightness value is separated by combining the grayscale distribution matrix of the current image to be displayed.

4. The method according to claim 1, characterized in that, The step of determining the contrast relationship between the foreground and the background based on the background transmittance brightness value and the initial foreground luminance value includes: The initial image contrast is calculated using the relative difference calculation method defined by the Weber fraction threshold, between the background transmittance brightness value and the initial foreground luminance brightness value. The retinal illuminance value is obtained, and the corresponding visual weight value is calculated based on the retinal illuminance value. The initial image brightness contrast is multiplied and weighted using the visual weight value to form a weighted brightness contrast feature, which serves as the brightness contrast relationship.

5. The method according to claim 1, characterized in that, The determination of the initial visual perception evaluation result based on the brightness contrast relationship and the visual adaptation parameters characterizing the observer's visual adaptation state includes: The light-dark contrast relationship, along with the temporal adaptation state characterized by the pupil adaptation curve and the dark adaptation time constant, are input into a pre-established human eye light-dark sensitivity model to obtain model response data. The response data within the cone cell response range is extracted from the model response data, and the response data is mapped to the foveal sensitive area using a spatial weighting algorithm to obtain the initial visual perception evaluation result.

6. The method according to claim 1, characterized in that, The step of determining the brightness compensation amount based on the predicted future ambient lighting and the brightness contrast relationship includes: The predicted future ambient lighting results and the contrast relationship between light and dark are used to construct the input feature vector of a support vector machine; The input feature vector is classified using the hyperplane segmentation boundary obtained by training with radial basis kernel function and soft margin relaxation factor to obtain initial classification results; Based on the initial classification results, a multi-classification voting mechanism is used to conduct voting statistics in a preset candidate compensation interval, and the compensation value corresponding to the candidate compensation interval with the highest number of votes is determined as the brightness compensation amount.

7. The method according to claim 1, characterized in that, The visual comfort threshold determined based on the background transmittance value includes: The background transmittance brightness value is input into a preset visual model, and the visual model is used to calculate the range of human eye visual comfort brightness corresponding to the background transmittance brightness value. The upper and lower limits of the visual comfort brightness range are used as the visual comfort threshold.

8. The method according to claim 1, characterized in that, The step of constraining and adjusting the foreground luminance value corresponding to the luminance compensation amount to obtain the updated foreground luminance value includes: Obtain the initial foreground luminance value and the luminance compensation amount, and sum the initial foreground luminance value and the luminance compensation amount to obtain the theoretical required luminance value; Obtain the upper and lower limits of the visual comfort threshold, and compare the theoretical required brightness value with the upper and lower limits. If the theoretical required brightness value is greater than the upper limit value, the theoretical required brightness value is adjusted to the upper limit value; if the theoretical required brightness value is less than the lower limit value, the theoretical required brightness value is adjusted to the lower limit value; otherwise, the theoretical required brightness value is used as the updated foreground luminous brightness value.

9. The method according to claim 5, characterized in that, The step of re-determining the target brightness contrast relationship based on the updated foreground luminance value and the background transmittance value, and performing a visual perception evaluation again based on the target brightness contrast relationship, and outputting the updated foreground luminance value when the re-determined visual perception evaluation result meets the preset visual perception evaluation conditions, includes: The relative difference calculation method defined by the Weber fraction threshold is used to calculate the target image brightness contrast between the updated foreground luminance value and the background transmittance value, which is used as the target brightness contrast relationship. The target brightness contrast relationship, together with the current pupil adaptation curve state, is input again into the human eye brightness sensitivity model. After processing by the human eye brightness sensitivity model, the target visual perception evaluation result is obtained in the foveal sensitive area. If the target visual perception evaluation result meets the preset visual perception evaluation conditions, then the updated foreground luminance value is output.

10. A brightness control system for a transparent LED display screen based on environmental perception, characterized in that, The system includes: An ambient lighting prediction module is used to acquire the historical ambient lighting sequence of the scene where the transparent LED display is located, and to determine the future ambient lighting prediction result based on the historical ambient lighting sequence. The brightness contrast relationship determination module is used to determine the background transmission brightness value corresponding to the transparent LED display screen based on the future ambient light prediction result, and to obtain the initial foreground luminous brightness value corresponding to the transparent LED display screen, and to determine the brightness contrast relationship between the foreground and the background based on the background transmission brightness value and the initial foreground luminous brightness value. The initial visual perception evaluation module is used to determine the initial visual perception evaluation result based on the brightness contrast relationship and the visual adaptation parameters that characterize the observer's visual adaptation state. The brightness compensation adjustment module is used to determine the brightness compensation amount based on the future ambient light prediction result and the brightness contrast relationship when the initial visual perception evaluation result does not meet the preset visual perception evaluation conditions and the brightness contrast relationship exceeds the preset brightness transition tolerance range. It also uses the module to constrain and adjust the foreground luminous brightness value corresponding to the brightness compensation amount based on the visual comfort threshold determined based on the background transmittance brightness value, so as to obtain the updated foreground luminous brightness value. The closed-loop evaluation output module is used to redetermine the target brightness contrast relationship based on the updated foreground luminance value and the background transmission luminance value, and to perform visual perception evaluation again based on the target brightness contrast relationship. When the visual perception evaluation result determined again meets the preset visual perception evaluation conditions, the module outputs the updated foreground luminance value.

Images