Display driving method and device, computer program product and electronic equipment

Abstract

The invention relates to the technical field of display control, and relates to a display driving method and device, a computer program product and electronic equipment. The display driving method comprises the steps that gamma fine tuning data of an adjacent brightness frequency band of a to-be-debugged brightness frequency band in a display panel is acquired, and the adjacent brightness frequency band is a frequency band which is adjacent to the to-be-debugged brightness frequency band and does not need to dynamically debug anode reset voltage; performing feature selection from the gamma fine tuning data to obtain target feature data; inputting the target feature data into a pre-trained prediction model for prediction processing to obtain a target anode reset voltage, the pre-trained prediction model being obtained by training gamma fine tuning data including a brightness frequency band and historical detection sample data of an anode reset voltage value; and resetting an anode of a light-emitting element of the display panel according to the target anode reset voltage. According to the invention, the anode reset voltage can be accurately predicted, and the time required for executing anode reset is shortened.

Description

Technical Field

[0001] This disclosure relates to the field of display control technology, and more specifically, to a display driving method, a display driving device, a computer program product, and an electronic device. Background Technology

[0002] Dynamic anode reset voltage is an important tool for optimizing low grayscale image quality, reducing ghosting and flicker, and balancing contrast in OLED (Organic Light-Emitting Diode) displays. However, current dynamic anode reset voltage strategies require setting a wide scanning range to account for product characteristic fluctuations, which greatly increases the anode reset operation time and results in low accuracy of the reset voltage.

[0003] It should be noted that the information in the background section above is only used to enhance the understanding of the background of this disclosure, and therefore may include information that does not constitute prior art known to those skilled in the art. Summary of the Invention

[0004] The purpose of this disclosure is to provide a display driving method and apparatus, computer program product and electronic device, thereby overcoming at least to some extent the problems existing in the related art, accurately predicting the anode reset voltage and reducing the time required to perform the anode reset.

[0005] Other features and advantages of this disclosure will become apparent from the following detailed description, or may be learned in part from practice of this disclosure.

[0006] According to one aspect of this disclosure, a display driving method is provided, comprising: acquiring gamma fine-tuning data of adjacent brightness frequency bands of a brightness frequency band to be tuned in a display panel, wherein the adjacent brightness frequency bands are frequency bands adjacent to the brightness frequency band to be tuned and do not require dynamic adjustment of the anode reset voltage; performing feature selection from the gamma fine-tuning data to obtain target feature data; inputting the target feature data into a pre-trained prediction model for prediction processing to obtain a target anode reset voltage, wherein the pre-trained prediction model is trained using historical detection sample data containing gamma fine-tuning data of brightness frequency bands and anode reset voltage values; and resetting the anode of the light-emitting element of the display panel according to the target anode reset voltage.

[0007] In one exemplary embodiment of this disclosure, the process of training a pre-trained prediction model includes: acquiring the detected brightness frequency bands, gamma fine-tuning data corresponding to the brightness frequency bands to be detected, and anode reset voltages included in the detected display screen; performing feature selection on the gamma fine-tuning data corresponding to the detected brightness frequency bands to obtain reference feature data required for adjusting the anode reset voltage; inputting the reference feature data into the neural network model to be trained to obtain the predicted anode reset voltage of the detected brightness frequency bands; constructing a loss function based on the predicted anode reset voltage and the anode reset voltage corresponding to the brightness frequency bands to be detected, and adjusting the parameters in the neural network model to be trained based on the loss function to obtain the pre-trained prediction model.

[0008] In one exemplary embodiment of this disclosure, feature selection is performed on the gamma fine-tuning data corresponding to the detected brightness frequency band to obtain reference feature data, including: obtaining a first correlation relationship between R pixels of different gray levels in the gamma fine-tuning data corresponding to the brightness frequency band to be detected and the anode reset voltage corresponding to the brightness frequency band to be detected; determining a second correlation relationship between G pixels of different gray levels in the gamma fine-tuning data corresponding to the brightness frequency band to be detected and the anode reset voltage corresponding to the brightness frequency band to be detected; determining a third correlation relationship between B pixels of different gray levels in the gamma fine-tuning data corresponding to the brightness frequency band to be detected and the anode reset voltage corresponding to the brightness frequency band to be detected; and based on the first correlation relationship, the second correlation relationship, and the third correlation relationship, determining target pixel data with a high linear relationship with the anode reset voltage corresponding to the brightness frequency band to be detected as reference feature data.

[0009] In one exemplary embodiment of this disclosure, the neural network model to be trained includes an input layer network, a residual network, a global pooling network, and a regression output layer. Inputting reference feature data into the neural network model to obtain the predicted anode reset voltage for the detected brightness frequency band includes: preprocessing the reference feature data to obtain processed reference feature data; inputting the processed reference feature data into the input layer network and passing the processed reference feature data to the residual network; extracting features from the processed reference feature data layer by layer through multiple residual blocks in the residual network to obtain reference spatial features; converting the reference spatial features into channel feature vectors using the global pooling network; and mapping the channel feature vectors to the predicted anode reset voltage for the detected brightness frequency band based on the regression output layer.

[0010] In one exemplary embodiment of this disclosure, data preprocessing is performed on the reference feature data to obtain processed reference feature data, including: extracting a preset number of fine-tuning voltage values ​​at a specific grayscale from the reference feature data; normalizing each fine-tuning voltage value according to the display panel dimension to obtain a first processing result; and normalizing each first processing result according to the grayscale dimension to obtain processed reference feature data.

[0011] In one exemplary embodiment of this disclosure, feature selection from gamma-level fine-tuning data to obtain target feature data includes: filtering initial target-type pixel data from the gamma-level fine-tuning data; selecting pixel data covered by key regions from the initial target-type pixel data; and extracting data corresponding to key grayscale points from the selected data to obtain target feature data.

[0012] In one exemplary embodiment of this disclosure, obtaining gamma fine-tuning data of adjacent brightness bands of the brightness band to be adjusted in the display panel includes: obtaining gamma fine-tuning data of the previous adjacent brightness band of the brightness band to be adjusted; or, obtaining gamma fine-tuning data of multiple known brightness bands adjacent to the brightness band to be adjusted, and determining gamma fine-tuning data of adjacent brightness bands based on the gamma fine-tuning data of the multiple known brightness bands.

[0013] In one exemplary embodiment of this disclosure, resetting the anode of the light-emitting element of the display panel according to the target anode reset voltage includes: obtaining the illumination compensation voltage corresponding to the brightness frequency band to be adjusted; determining the anode reset voltage to be adjusted according to the target anode reset voltage and the illumination compensation voltage, and performing gamma calibration according to the anode voltage to be adjusted to obtain a calibration voltage; and resetting the anode of the light-emitting element of the display panel according to the calibration voltage.

[0014] According to one aspect of this disclosure, a display driving device is provided, comprising: a data acquisition module for acquiring gamma fine-tuning data of adjacent brightness frequency bands of a brightness frequency band to be adjusted in a display panel, wherein the adjacent brightness frequency bands are frequency bands adjacent to the brightness frequency band to be adjusted and do not require dynamic adjustment of the anode reset voltage; a feature selection module for performing feature selection from the gamma fine-tuning data to obtain target feature data; a voltage prediction module for inputting the target feature data into a pre-trained prediction model for prediction processing to obtain a target anode reset voltage, wherein the pre-trained prediction model is trained using historical detection sample data containing gamma fine-tuning data of brightness frequency bands and anode reset voltage values; and a display processing module for resetting the anode of the light-emitting element of the display panel according to the target anode reset voltage.

[0015] According to one aspect of this disclosure, a computer program product is provided, comprising a computer program that, when executed by a processor, implements any of the above methods.

[0016] According to one aspect of this disclosure, an electronic device is provided, comprising: a processor; and a memory for storing executable instructions of the processor; wherein the processor is configured to perform any of the above methods by executing the executable instructions.

[0017] The display driving method in the exemplary embodiments of this disclosure acquires gamma fine-tuning data of adjacent brightness bands of the brightness band to be calibrated in the display panel, wherein the adjacent brightness bands are frequency bands adjacent to the brightness band to be calibrated and do not require dynamic adjustment of the anode reset voltage; performs feature selection from the gamma fine-tuning data to obtain target feature data; inputs the target feature data into a pre-trained prediction model for prediction processing to obtain the target anode reset voltage, wherein the pre-trained prediction model is trained using historical detection sample data containing gamma fine-tuning data of the brightness band and anode reset voltage values; and resets the anode of the light-emitting element of the display panel according to the target anode reset voltage. On the one hand, for the brightness band to be calibrated, acquiring existing gamma data of adjacent bands and instantaneously calculating the optimal anode reset voltage through the model reduces this step from minutes to seconds, directly compressing the critical path time of module production and improving the efficiency of anode reset. On the other hand, the model can uncover the relationship between gamma data and the optimal anode reset voltage, and the predicted voltage value is theoretically closer to the global optimal solution, thereby improving display uniformity, reducing performance differences between different panels and batches, and improving the accuracy of anode reset.

[0018] It should be understood that the above general description and the following detailed description are exemplary and explanatory only, and are not intended to limit this disclosure. Attached Figure Description

[0019] The above and other objects, features, and advantages of this disclosure will become readily apparent from the following detailed description of exemplary embodiments, taken in conjunction with the accompanying drawings. Several embodiments of this disclosure are illustrated in the drawings by way of example and not limitation.

[0020] Figure 1 A schematic diagram of an OLED pixel driving circuit according to an exemplary embodiment of the present disclosure is shown.

[0021] Figure 2 A schematic diagram illustrating the operating point variation of an OLED DTFT with ghosting, strobe, and hysteresis effect according to an exemplary embodiment of the present disclosure is shown.

[0022] Figure 3 A schematic diagram of the VREFN2 voltage assignment range collected by a production line according to an exemplary embodiment of the present disclosure is shown.

[0023] Figure 4A flowchart of a display driving method according to an exemplary embodiment of the present disclosure is shown.

[0024] Figure 5 A flowchart illustrating feature selection from gamma-trimmed data according to an exemplary embodiment of the present disclosure is shown.

[0025] Figure 6 A linear correlation distribution diagram of gamma-adjusted data G-pixel data values ​​and dynamic anode reset voltage according to an exemplary embodiment of the present disclosure is shown.

[0026] Figure 7 A linear correlation distribution diagram of gamma-adjustment data R pixel data values ​​and dynamic anode reset voltage according to an exemplary embodiment of the present disclosure is shown.

[0027] Figure 8 A linear correlation distribution diagram of gamma-adjusted data B pixel data values ​​and dynamic anode reset voltage according to an exemplary embodiment of the present disclosure is shown.

[0028] Figure 9 A flowchart illustrating training a pre-trained prediction model according to an exemplary embodiment of the present disclosure is shown.

[0029] Figure 10 An example diagram of a neural network model to be trained according to an exemplary embodiment of the present disclosure is shown.

[0030] Figure 11 A schematic diagram of an output unit defined according to an exemplary embodiment of the present disclosure is shown.

[0031] Figure 12 A graph showing the training set accuracy versus the test set accuracy under multiple iterations in accordance with an exemplary embodiment of the present disclosure is shown.

[0032] Figure 13 The diagram illustrates the training set loss curve and the test set loss curve under multiple iterations of an exemplary embodiment of the present disclosure.

[0033] Figure 14 A schematic diagram of dynamic anode reset voltage distribution range prediction under a test set according to an exemplary embodiment of the present disclosure is shown.

[0034] Figure 15 A schematic diagram illustrating the accuracy of anode reset voltage prediction and comparison according to an exemplary embodiment of the present disclosure is shown.

[0035] Figure 16 A flowchart illustrating a data preprocessing of reference feature data according to an exemplary embodiment of the present disclosure is shown.

[0036] Figure 17 A complete flowchart of a display driver according to an exemplary embodiment of the present disclosure is shown.

[0037] Figure 18 A schematic diagram showing a comparison of the Tact time of an anode reset algorithm according to an exemplary embodiment of the present disclosure is illustrated.

[0038] Figure 19 A schematic diagram of the composition of a display driving device according to an exemplary embodiment of the present disclosure is shown.

[0039] Figure 20 A block diagram of an electronic device according to an exemplary embodiment of the present disclosure is shown.

[0040] In the accompanying drawings, the same or corresponding reference numerals indicate the same or corresponding parts. Detailed Implementation

[0041] Exemplary embodiments will now be described more fully with reference to the accompanying drawings. However, exemplary embodiments can be implemented in many forms and should not be construed as limited to the examples set forth herein; rather, these embodiments are provided so that this disclosure will be more comprehensive and complete, and will fully convey the concept of exemplary embodiments to those skilled in the art. The same reference numerals in the drawings denote the same or similar structures, and therefore their detailed description will be omitted.

[0042] Furthermore, the described features, structures, or characteristics can be combined in any suitable manner in one or more embodiments. Numerous specific details are provided in the following description to give a thorough understanding of embodiments of this disclosure. However, those skilled in the art will recognize that the technical solutions of this disclosure can be practiced without one or more of the specific details described, or other methods, components, apparatuses, steps, etc., can be employed. In other instances, well-known structures, methods, apparatuses, implementations, or operations are not shown or described in detail to avoid obscuring various aspects of this disclosure.

[0043] The block diagrams shown in the accompanying drawings are merely functional entities and do not necessarily correspond to physically independent entities. That is, these functional entities can be implemented in software, or in one or more software-hardened modules, or in different network and / or processor devices and / or microcontroller devices.

[0044] First, the terms or concepts involved in this disclosure will be explained.

[0045] OLED: Organic Light-Emitting Diode.

[0046] DTFT: Driver Thin Film Transistor.

[0047] VREFN2 voltage: OLED anode reset voltage during the reset phase.

[0048] Low grayscale image quality: refers to the image quality presented by the display panel in the dark grayscale area that is close to black (usually the brightness range of 0.0005-5 nits).

[0049] Ghosting / flickering: refers to the grayscale reversal phenomenon that occurs when switching between high and low grayscale levels (from 2nit 0gray to 2nit 32gray) of a pixel after displaying a static image for a long time (2nit 0gray). This is caused by hysteresis and changes in the amount of captured charge on the interface.

[0050] Contrast ratio: The ratio of the brightness of the brightest (white) and darkest (black) parts of a displayed image.

[0051] The OLED anode reset voltage algorithm is an important means of optimizing OLED low grayscale image quality, ghosting and flicker optimization, and contrast balance. Its principle is as follows: (1) By adjusting the anode reset voltage, the potential of node N4 in the reset stage is increased, the OLED emission ratio of point N4 is increased, the emission ratio of DTFT driving current is reduced, and the impact of poor DTFT uniformity on low grayscale image quality is weakened; (2) By adjusting the anode reset voltage, the low grayscale emission process is dominated by the emission of OLED at point N4, and the current change caused by hysteresis during DTFT compensation is reduced, so as to improve the poor ghosting and flicker; (3) Under the premise of satisfying the contrast, the anode reset voltage between screens is adapted to low grayscale image quality and ghosting and flicker.

[0052] like Figure 1 The diagram shown is a schematic of an OLED pixel driving circuit according to this disclosure. Figure 1The driving process is illustrated using an example. First, in the VREFN1 initialization phase, VREFN1 serves as the initial voltage for Vg (DTFT gate). T1 is turned on (controlled by Reset P), and T2 is turned on (controlled by Gate N), writing the Vg voltage into VREFN1. Second, in the Vdata pixel voltage writing phase, the Vdata pixel voltage is sequentially written to Vg (DTFT gate) through T4, DTFT, and T2. T4 is turned on (controlled by Gate P), and T2 is turned on (controlled by Gate N). As the Vdata pixel voltage is written, the DTFT gate voltage Vg gradually increases until Vg = Vdata + Vth, completing the pixel voltage writing. Next, in the VREFN2 / VREFN3 initialization phase, the VREFN2 anode reset voltage serves as the reset voltage for the OLED anode. T7 is turned on (controlled by Reset H), and VREFN2 writes the voltage at point N4. VREFN3 serves as the reset voltage for the DTFT source voltage. T8 is turned on (controlled by Reset H), and VREFN3 writes the Vg voltage. Finally, in the emission stage, T5 is turned on (EM control), T6 is turned on (EM control), the source voltage changes at T3, so Vgs < 0, T3 is turned on, thus enabling the OLED device to emit light.

[0053] Therefore, it can be seen that the OLED display module optimizes low grayscale image quality by adjusting the OLED anode reset voltage (VREFN2) to raise the potential of point N4 during the reset phase, thereby increasing the OLED light emission ratio at point N4 and reducing the light emission ratio of the DTFT driving current, thus mitigating the impact of poor DTFT uniformity on low grayscale image quality. For example, Figure 2 The diagram shows the operating point change of an OLED DTFT with hysteresis effect in motion blur and strobe. The optimization action for low grayscale motion blur and strobe is to adjust the OLED anode reset voltage (VREFN2) so that the low grayscale light emission process is dominated by the OLED at point N4, thereby reducing the current change (Delta) caused by the hysteresis effect (the current generated by the negative bias of Vth when switching from 0gray to 32gray) during DTFT compensation, so as to improve the poor motion blur and strobe effect.

[0054] However, during the debugging process of OLED display modules, a dilemma often arises: balancing black screen brightness (million contrast ratio) with low grayscale image quality optimization / flicker reduction. Due to fluctuations in product manufacturing processes, it's impossible to set a universal anode reset voltage that suits all products. To address this issue, an existing dynamic anode reset voltage algorithm scans the brightness curve of each product to assign a suitable dynamic anode voltage, ensuring a balance between contrast and low grayscale image quality / flicker reduction. Considering the impact of production line gamma capacity, the dynamic anode reset voltage algorithm, due to the need to consider the characteristic distribution of all products within its existing scan range, typically sets a relatively wide scan range, resulting in a longer Tact time (the time required for the anode reset voltage of a single brightness band to be debugged). Figure 3 The diagram shows a schematic of the VREFN2 voltage assignment range collected on a production line. The VREFN2-ELVSS scan range needs to be set to 0.5V ~ 2.0V (tep=16). Furthermore, due to the different PWM Duty (Pulse Width Modulation Duty) of the high-brightness band and the low-brightness band, their VREFN2-ELVSS start-up curves also differ. Therefore, it is usually necessary to collect the VREFN2-ELVSS start-up curves of 2-3 bands, increasing the adjustment time.

[0055] To address one or more of the aforementioned issues, this disclosure provides a display driving method that enables precise prediction of the anode reset voltage using only the screen's Gamma Tuning Data during the OLED display module production process, thereby reducing Tact Time.

[0056] It should be noted that the display driver method disclosed herein can be applied to terminals and servers. The server can be a standalone physical server, a server cluster or distributed system composed of multiple physical servers, or a cloud server providing basic cloud computing services such as cloud services, cloud computing, cloud functions, and big data and artificial intelligence platforms. The terminal can be a tablet computer, laptop computer, desktop computer, IoT device, etc. Furthermore, the technical solutions of the exemplary embodiments of this disclosure can also be executed collaboratively by the terminal and the server. This disclosure does not impose any special limitations herein.

[0057] refer to Figure 4 The diagram shown is a flowchart of a display driving method according to an exemplary embodiment of this disclosure. Figure 4 As shown, the display driving method includes steps S410 to S440, as detailed below: Step S410: Obtain the gamma fine-tuning data of the adjacent brightness bands of the brightness band to be adjusted in the display panel, wherein the adjacent brightness bands are the frequency bands adjacent to the brightness band to be adjusted and do not require dynamic adjustment of the anode reset voltage.

[0058] Step S420: Select features from the gamma fine-tuning data to obtain target feature data.

[0059] Step S430: Input the target feature data into the pre-trained prediction model for prediction processing to obtain the target anode reset voltage. The pre-trained prediction model is trained using gamma fine-tuning data including the brightness frequency band and historical detection sample data of the anode reset voltage value.

[0060] Step S440: Reset the anode of the light-emitting element of the display panel according to the target anode reset voltage.

[0061] The display driving method in the exemplary embodiments of this disclosure, on the one hand, acquires existing gamma data of adjacent frequency bands for the brightness frequency band to be debugged, and instantly calculates the optimal anode reset voltage through a model, reducing this step from minutes to seconds, directly compressing the critical path time of module production and improving the efficiency of anode reset. On the other hand, the model can discover the relationship between gamma data and the optimal anode reset voltage, and the predicted voltage value is theoretically closer to the global optimal solution, thereby improving display uniformity, reducing performance differences between different panels and different batches, and improving the quality of anode reset.

[0062] Steps S410 to S440 will be described in more detail below.

[0063] In step S410, the gamma fine-tuning data of the adjacent brightness bands of the brightness band to be adjusted in the display panel is obtained. The adjacent brightness bands are the frequency bands that are adjacent to the brightness band to be adjusted and do not require dynamic adjustment of the anode reset voltage.

[0064] In the exemplary embodiments of this disclosure, the brightness band, also known as the band, is used during production debugging to divide the full brightness range of the display panel from the lowest to the highest brightness into several representative brightness intervals or levels for segmented debugging. Gamma fine-tuning data is a set of driving voltage compensation values ​​applied to each sub-pixel at different grayscale points (e.g., 0, 1, 2, ..., 255) to ensure that the input image grayscale signal and the actual output brightness of the display panel conform to a preset gamma curve within a specific brightness band. Essentially, it is a set of calibration parameters used to compensate for non-uniformity caused by OLED backplane process deviations, reflecting the electrical input characteristics required to drive the specific panel to achieve an ideal optical response under a specific brightness target.

[0065] Among them, the brightness band to be adjusted (Band(n)) is the frequency band where the anode reset voltage needs to be adjusted. The adjacent brightness bands of the brightness band to be adjusted are the frequency bands in the adjacent Band(n) that do not need to adjust the anode reset voltage. The number of them can be one or more.

[0066] Optionally, obtaining the gamma fine-tuning data of the adjacent brightness bands of the brightness band to be adjusted in the display panel can be achieved by obtaining the gamma fine-tuning data of the previous adjacent brightness band of the brightness band to be adjusted. That is, the frequency band in the adjacent band (n) that does not require adjustment of the anode reset voltage, such as band (n-1). This previous adjacent brightness band is the frequency band that is numerically lower than and closest to the brightness band to be adjusted in the preset brightness band sequence. For example, if the sequence is [2nit, 50nit, 300nit, 800nit], and the brightness band to be adjusted is 300 nit, then the previous adjacent brightness band refers to 50 nit.

[0067] Optionally, gamma fine-tuning data of multiple known brightness frequency bands adjacent to the brightness frequency band to be tuned can be obtained, and gamma fine-tuning data of adjacent brightness frequency bands can be determined based on the gamma fine-tuning data of the multiple known brightness frequency bands. The frequency bands that are adjacent to each other and do not require dynamic adjustment of the anode reset voltage refer to multiple frequency bands that have been tuned (or do not require adjustment) selected on the side with lower and higher values ​​of the brightness frequency band to be tuned, centered on it. The number of selected bands can be flexibly set according to the actual scenario requirements, such as 5, 7, etc., without any special limitation.

[0068] One method is to determine the gamma fine-tuning data for adjacent frequency bands by interpolating the gamma fine-tuning data from known frequency bands. Specifically, gamma fine-tuning data from at least two known frequency bands before and after the frequency band to be tuned can be obtained. The brightness value of each frequency band is used as the independent variable X, and its complete gamma data matrix is ​​used as the dependent variable Y. For each specific parameter in the gamma fine-tuning data (such as the compensation voltage value of pixel G at grayscale 128), linear interpolation is performed between two points in the at least two known frequency bands before and after, calculating the parameter value under the target virtual brightness. Then, the interpolation operation is repeated for all key parameters in the gamma fine-tuning data (such as the voltages of R, G, and B at each characteristic grayscale point), ultimately integrating them into a complete gamma fine-tuning data matrix corresponding to the virtual brightness frequency band, which serves as the gamma fine-tuning data for adjacent brightness frequency bands. Alternatively, the initial feature vectors extracted from the gamma data of each known frequency band can be weighted and fused according to the correlation weight between their corresponding brightness frequency band and the frequency band to be tuned to obtain the gamma fine-tuning data for adjacent brightness frequency bands. The correlation weight is inversely proportional to the absolute value of the brightness difference.

[0069] Both methods ensure a high correlation between the input data and the frequency band to be debugged in terms of physical characteristics, providing a data foundation for subsequent predictions. In the first method, no intermediate calculations or fusion steps are required; the raw data is directly fed into the prediction model, maximizing the overall speed of the prediction process, minimizing the computational requirements of the production line control system, and ensuring a clear prediction chain. If a prediction deviation occurs, the problem tracing path is very short, allowing for quick identification of whether the issue stems from abnormal data in the previous frequency band, feature selection problems, or a problem with the model itself, facilitating process maintenance and troubleshooting. In the second method, by fusing data from multiple brightness points, a local model of panel characteristics in the brightness dimension can be constructed, thereby more accurately inferring the characteristics that should exist in the target frequency band neighborhood, improving the accuracy of the basic data.

[0070] In step S420, feature selection is performed from the gamma fine-tuning data to obtain target feature data.

[0071] In an exemplary embodiment of this disclosure, feature selection is the process of systematically and purposefully filtering out the most relevant and effective subset of the anode reset voltage prediction target from the aforementioned high-dimensional, raw gamma fine-tuning data. That is, the target feature data is formatted and vectorized data obtained after feature selection and used to directly input the prediction model.

[0072] In the case where the target class pixel data has been determined during the model training stage, and the model is trained based on this data to obtain a pre-trained prediction model, features of that class of pixels can be directly selected from the gamma fine-tuning data as target feature data.

[0073] In one exemplary embodiment, such as Figure 5 As shown, a method for feature selection through correlation analysis during the model training phase is provided. Feature selection is performed on the gamma fine-tuning data corresponding to the detected brightness frequency band to obtain reference feature data, including step S510: obtaining the first correlation relationship between R pixels of different gray levels in the gamma fine-tuning data and the anode reset voltage corresponding to the detected brightness frequency band.

[0074] Step S520: Determine the second correlation relationship between G pixels of different gray levels in the gamma fine-tuning data and the anode reset voltage corresponding to the detected brightness frequency band.

[0075] Step S530: Determine the third correlation relationship between B pixels of different gray levels in the gamma fine-tuning data and the anode reset voltage corresponding to the detected brightness frequency band.

[0076] Step S540: Based on the first correlation relationship, the second correlation relationship, and the third correlation relationship, determine the target pixel data that has a high linear relationship with the anode reset voltage corresponding to the brightness frequency band to be debugged as reference feature data.

[0077] The first, second, and third correlation relationships are statistical measures of the correlation strength between the gamma adjustment data (values ​​at different gray levels) of each of the three types of sub-pixels (R, G, and B) within the same detected brightness frequency band and the optimal anode reset voltage determined for that frequency band. The quantification method can use the Pearson correlation coefficient. The closer the absolute value of this coefficient is to 1, the stronger the linear correlation; the closer it is to 0, the weaker the linear relationship. Of course, other correlation determination methods can also be used. A high linear relationship means that, after statistical analysis of a large number of historical samples, the correlation coefficient between a certain type of pixel data (such as G pixel data in the gray level range of 32-128) and the anode reset voltage is consistently and stably higher than a preset threshold (e.g., |r|>0.7), and passes the statistical significance test (e.g., p-value <0.01). Therefore, based on the above correlation analysis, the gamma adjustment data of the sub-pixels is determined to have the most significant and stable linear relationship with the anode reset voltage.

[0078] The process of determining correlation is illustrated below using a specific type of pixel data (such as G pixels) as an example.

[0079] Specifically, effective features strongly correlated with the anode reset voltage can be extracted from batch data (20k~25k) during the trial production phase. This involves multiple data sets containing gamma fine-tuning data corresponding to the detected brightness frequency bands. Each data set includes gamma voltage values ​​for 256 gray levels each for R, G, and B, as well as the optimal anode reset voltage value obtained through dynamic tuning. For each gray level, the correlation coefficient between the gamma data and the anode reset voltage is calculated, and a corresponding correlation curve is plotted based on this coefficient. The correlation relationship is then obtained from the correlation coefficient curve. For example,... Figure 6-8 The figures show the linear correlation distribution between G / R / B pixel data values ​​in the gamma fine-tuning data and the dynamic anode reset voltage. Multiple features from the high, medium, and low gray levels of the band to be tuned were selected to analyze the linear relationship with the anode reset voltage value. The tuning data values ​​(fine-tuning data) from the production line showed that the linear correlation of G pixels (from gray level 1 to 255, 27 binding points) was significantly higher than that of R pixels and B pixels. Since features with high linear correlation indicate that the model is easier to train and converge, and does not require the use of complex neural networks, the G pixel data values ​​of the corresponding band were determined to be used as the training features of the input layer, i.e., the target class pixel data. Furthermore, in practical applications, G pixel data can be directly selected as the target feature data.

[0080] This disclosure calculates and compares the statistical correlation between the gamma data of three sub-pixels (R, G, and B) and the anode reset voltage, replacing empirical judgment with quantitative evidence to select the pixel category with the highest predictive value as the feature source. This ensures the strongest linear correlation between the input data and the prediction target, improving the information quality of the prediction model from the source, thereby improving the accuracy of the final prediction of the anode reset voltage. At the same time, it avoids processing more pixel data during actual prediction implementation, further improving processing efficiency.

[0081] In one exemplary embodiment, feature selection from gamma-ray fine-tuning data to obtain target feature data may include: First, filter the initial target class pixel data from the gamma fine-tuning data; then select the pixel data covered by the key region from the initial target class pixel data, and extract the data corresponding to the key gray level points from the selected data to obtain the target feature data.

[0082] The initial target pixel data is the grayscale data of a certain type of sub-pixel (e.g., G) initially selected from the complete gamma fine-tuning data according to preset rules (correlation analysis conclusions). For example, after selecting the G pixel, what is obtained is the set of fine-tuning voltage values ​​of all G pixels in all 256 grayscale points under that brightness frequency band. The key area refers to a specific area on the physical plane of the display panel that has the best process uniformity, best represents the overall electro-optical characteristics of the panel, and is least affected by edge effects. For example, it could be the geometric center area of ​​the panel, or the set of pixels covered by a specific adjustment pattern (e.g., checkerboard, window pattern). Key grayscale points are several specific grayscale points within the full grayscale range of 0-255 that are most closely related to the anode reset voltage and best reflect changes in the driving state. They can be concentrated in the low to medium grayscale range (e.g., grayscales 32, 64, 96, 128). Since the anode reset voltage mainly affects charge initialization under low current conditions, the driving conditions of low to medium grayscale levels are directly related to this. Of course, several specific grayscale data can be selected according to the actual panel type or driving requirements.

[0083] This disclosure ensures feature quality from the source by screening target pixels (such as G pixels) that are most physically correlated with the anode reset voltage and eliminating interference noise from R / B pixels. By selecting key areas for process uniformity, it effectively shields abnormal data caused by edge effects and random process defects, making the features represent the intrinsic characteristics of the panel.

[0084] In step S430, the target feature data is input into the pre-trained prediction model for prediction processing to obtain the target anode reset voltage. The pre-trained prediction model is trained using gamma fine-tuning data including the brightness frequency band and historical detection sample data of the anode reset voltage value.

[0085] In an exemplary embodiment of this disclosure, each sample of the historical detection sample data is a data pair: the input is reference feature data extracted from the gamma fine-tuning data of the historical panel in a certain brightness band (usually an adjacent band), and the output (label) is the true value of the anode reset voltage that is finally determined by the traditional dynamic adjustment method in the brightness band to be adjusted of the historical panel and verified as optimal.

[0086] In one exemplary embodiment, such as Figure 9 As shown, the process of training a pre-trained prediction model includes: Step S910: Obtain the detected brightness frequency band, the gamma fine-tuning data corresponding to the brightness frequency band to be detected, and the anode reset voltage included in the detected display screen.

[0087] Step S920: Perform feature selection on the gamma fine-tuning data corresponding to the detected brightness frequency band to obtain the reference feature data required for adjusting the anode reset voltage.

[0088] Step S930: Input the reference feature data into the neural network model to be trained to obtain the predicted anode reset voltage of the detected brightness frequency band.

[0089] Step S940: Construct a loss function based on the predicted anode reset voltage and the anode reset voltage corresponding to the brightness frequency band to be detected, and adjust the parameters in the neural network model to be trained based on the loss function to obtain the pre-trained prediction model.

[0090] Among them, the gamma fine-tuning data corresponding to the detected brightness frequency band is used for feature selection to obtain the reference feature data required for adjusting the anode reset voltage. This has already been explained in the process of determining the target feature data above, and will not be repeated here.

[0091] Specifically, the neural network model to be trained includes an input layer network, a residual network, a global pooling network, and a regression output layer. Inputting reference feature data into the neural network model to be trained yields the predicted anode reset voltage for the detected brightness frequency band, which may include: First, the reference feature data is preprocessed to obtain processed reference feature data. Then, the processed reference feature data is input into the input layer network and passed to the residual network. Next, multiple residual blocks in the residual network extract features from the processed reference feature data layer by layer to obtain reference spatial features. Finally, a global pooling network is used to convert the reference spatial features into channel feature vectors, and the channel feature vectors are mapped to the predicted anode reset voltage of the detected brightness frequency band based on the regression output layer.

[0092] The input layer network receives and initially transforms the preprocessed feature vectors, projecting the dimension of the input data onto the feature space dimension required by the hidden layers inside the model, preparing for subsequent deep feature extraction. The basic component of the residual network is the residual block. This structure greatly alleviates the gradient vanishing / exploding problem in deep networks, making it possible to build deeper networks to extract complex features while ensuring training stability. The reference space features are high-order, abstract feature representations learned after layer-by-layer nonlinear transformation by multiple residual blocks in the residual network. The global pooling network is used for dimensionality compression and feature aggregation operations, such as averaging all spatial locations for each channel, ultimately compressing each channel into a single scalar. Its output is a one-dimensional channel feature vector, the length of which is equal to the number of channels in the input features. The regression output layer is the last layer of the model, used to map the high-dimensional channel feature vectors to the final, continuous numerical prediction target, i.e., the predicted anode reset voltage value.

[0093] like Figure 10 The diagram shows an example of a neural network model to be trained. This model is a residual network model, consisting of three main parts: an input layer, hidden layers, and an output layer. Each layer contains multiple neurons, and these neurons are connected to all neurons in the next layer. The input layer receives the raw data input; the hidden layers are one or more intermediate layers used to extract features and learn the nonlinear relationship between the data; and the output layer produces the final prediction result. The residual network model introduces residual learning to address the degradation problem in deep networks. For a desired mapping H(x), residual learning does not directly learn H(x), but rather learns the residual mapping F(x) such that H(x) = F(x) + x.

[0094] For a residual block, given an input x, the output of the staggered block is: Formula 1 Formula 2 Formula 3 in, is the residual function (a combination of multiple convolutional layers), x is the identity mapping passed by the skip connection, W1 and W2 are the weights of the convolutional kernel (e.g., 3x3), which are the weights of the two convolutional layers respectively; BN: represents the batch normalization operation.

[0095] When the input and output dimensions do not match, a 1×1 convolution can be introduced to adjust the dimensions: Formula 4 In the formula, The parameters for the 1x1 convolution kernel (used to adjust the number of channels).

[0096] In the training principle of residual networks, each neuron in a layer has associated weights and biases. The weights determine the influence of the input factor, and the biases determine the activation threshold of the neuron. Each neuron is introduced with a non-linearity through an activation function. Backpropagation between layers calculates a cost function to measure the difference between the predicted output and the true value, thus iterating through gradient descent to update the neuron's weights and biases. The training process involves the following formulas: Formula 5 Formula 6 Formula 7

[0097] Formula 8 Formula 9 Among the various types, This refers to the K-dimensional vector output by the neural network, where K represents the number of units in the output layer. This represents the i-th output, indicating the selection of the i-th element of the neural network output vector. Equation 7 represents... As a K-dimensional vector, it can be represented by multiplying the weight parameters of the weight matrix of each layer in the neural network with the input matrix of each layer and then adding them together. The cost function is defined as follows: m is the number of samples in the training set; K is the number of units in the output layer; L is the total number of layers in the neural network (including the input and output layers); sl is the number of units (neurons) in the l-th layer; Θ(l) is the weight matrix of the l-th layer, connecting the l-th and l+1-th layers; Θji(l) is the weight from the i-th unit in the l-th layer to the j-th unit in the l+1-th layer in the l-th weight matrix; λ is the regularization coefficient, controlling the strength of the regularization term; the regularization term is the sum of the squares of all weight parameters (L2 regularization), penalizing the large range of weight summation; l is the number of layers to be traversed (from the l-1-1-th layer); i is the number of units to be traversed in the l-th layer; j is the number of units to be traversed in the l+1-th layer; yk(i) is the true label of the i-th sample in the k categories (i.e., the anode reset voltage corresponding to the brightness frequency band to be detected). If the i-th sample belongs to class 2, then y2(i) = 1; otherwise, it is 0. hΘ(x(i))) is the model's prediction of the i-th sample in class k (i.e., the predicted anode reset voltage), which can be obtained using ReLU with a range of (0, 1). Compared to the sigmoid activation function, ReLU has the advantage of high computational speed and alleviates gradient vanishing. Therefore, the ReLU activation function is used in this training model, and the classification result is output through a fully connected layer.

[0098] In the OLED dynamic anode reset voltage algorithm, 16 anode reset voltage ranges were selected by collecting batch data (20k~25k) from the trial production stage (obvious abnormal values ​​were removed). Therefore, the output unit was defined as 16, and a multivariate classification method was used for prediction. Figure 11 As shown. Optionally, 80% of the production line batch data can be extracted as the training set and 20% as the test set in this training model to better verify the accuracy of the trained model. Furthermore, it can be trained using backpropagation and gradient descent, as shown... Figure 12 The graph shows the training set accuracy and test set accuracy after 500 iterations, demonstrating that the accuracy gradually converges to over 97%. Figure 13 The image shows the training set loss curve and the test set loss curve after 500 iterations. It can be seen that the loss gradually converges to a value below 0.1. Figure 14 The figure shows the predicted distribution range of the dynamic anode reset voltage under the test set, which is basically consistent with the true value. Therefore, it is proved that the training model can effectively establish a regression relationship between the gamma fine-tuning data and the dynamic anode reset voltage, thereby effectively predicting the set value of the dynamic anode reset voltage.

[0099] Furthermore, the prediction accuracy of the residual network model disclosed herein is compared with that of a fully connected feedforward neural network, such as... Figure 15 This diagram illustrates the prediction accuracy of the OLED dynamic anode reset voltage based on the residual model of this disclosure, and compares the prediction accuracy of the residual network model based on this disclosure with that of the fully connected feedforward neural network. The comparison shows that the prediction accuracy of the residual neural network is higher than that of the fully connected feedforward neural network. The training model of this disclosure can effectively establish a regression relationship between gamma fine-tuning data and dynamic anode reset voltage through residual learning, thereby accurately predicting the anode reset voltage.

[0100] In one exemplary embodiment, such as Figure 16 As shown, data preprocessing of the reference feature data yields processed reference feature data, which may include: Step S1610: Extract a preset number of fine-tuning voltage values ​​at a specific gray level from the reference feature data.

[0101] Step S1620: Normalize each fine-tuning voltage value according to the dimensions of the display panel to obtain the first processing result.

[0102] Step S1630: Normalize each of the first processing results according to the grayscale dimension to obtain the processed reference feature data.

[0103] Specifically, from the reference feature data, according to predetermined rules (such as a defined list of key grayscale points), specific values ​​at corresponding positions can be extracted to obtain a preset number of fine-tuning voltage values ​​at a specific grayscale. The preset number can be flexibly set according to actual needs and is not specifically limited. Normalization processing for the display panel dimension refers to different display panel samples, such as simultaneously processing reference feature data from a batch (N) of panels to eliminate data amplitude differences caused by absolute characteristic differences between panels (such as overall shifts in luminous efficiency or overall drifts in TFT backplane threshold voltage). Specifically, the mean and standard deviation of all N panel feature values ​​at that specific grayscale point can be calculated, and then the Z-score standardization can be applied to the value of each panel at that point. Normalization processing for the grayscale dimension refers to different grayscale points within the same display panel, eliminating inherent voltage range differences caused by the grayscale level itself; for example, the absolute voltage value of a high grayscale is naturally greater than that of a low grayscale. Specifically, for a single panel, the mean and standard deviation of its feature values ​​at all preset key gray levels are calculated, and then the values ​​of the panel at each gray level are standardized using Z-score.

[0104] For example, the following formula is the feature scaling formula when normalizing the fine-tuning voltage value: Formula 10 in, is the mean of the j-th feature (such as the j-th feature value of all N panels at a specific gray level or the j-th feature value at all preset key gray levels), where m is the number of samples. It is the value of the i-th sample on the j-th feature.

[0105] Formula 11

[0106] in, It is the standard deviation of the j-th feature.

[0107] Formula 12

[0108] Equation 12 is a standardized formula.

[0109] This disclosure normalizes the display panel dimension, filtering out overall voltage level deviations caused by batch material differences and process baseline drift between different panels. This allows the model to shift its focus from absolute voltage values ​​to learning relative characteristic patterns within the panel. Furthermore, by normalizing the grayscale dimension, the inherent magnitude of higher grayscale voltage compared to lower grayscale voltage is eliminated. This allows the model to focus on the shape of the voltage change curve between different grayscale points, which is the essential characteristic most closely related to the physical relationship of anode reset voltage, providing the crucial data foundation for subsequent high-precision predictions.

[0110] In step S440: The anode of the light-emitting element of the display panel is reset according to the target anode reset voltage.

[0111] In an exemplary embodiment of this disclosure, after obtaining the target anode reset voltage, the anode of the light-emitting element of the display panel can be reset using the target anode reset voltage.

[0112] In one exemplary embodiment, resetting the anode of the light-emitting element of the display panel according to the target anode reset voltage further includes: First, obtain the illumination compensation voltage corresponding to the brightness frequency band to be adjusted; then, determine the anode reset voltage to be adjusted based on the target anode reset voltage and the illumination compensation voltage, and perform gamma calibration based on the anode voltage to be adjusted to obtain the calibration voltage; finally, reset the anode of the light-emitting element of the display panel according to the calibration voltage.

[0113] The brightness compensation voltage is a specific offset or proportional coefficient applied to the driving voltage during the production and debugging of display panels to compensate for brightness deviations caused by factors such as differences in OLED luminescent material efficiency, TFT backplane threshold voltage (Vth) non-uniformity, or optical crosstalk. This voltage value can be independently measured and stored for each panel and each brightness frequency band during the initial optical inspection, or it can be preset based on experience to ensure that the panel can achieve the target brightness in that brightness frequency band. The intermediate voltage value obtained by fusing the target anode reset voltage and the brightness compensation voltage through specific rules (such as addition, weighted combination, or table lookup) is the anode reset voltage to be adjusted, which can be understood as the optimal anode reset voltage.

[0114] Gamma calibration is the process of calculating the actual driving voltage applied to the pixel circuit based on the anode reset voltage to be adjusted. Because the relationship between OLED luminance and driving voltage is highly non-linear (typically a power function), directly using linearly superimposed voltages may not produce the desired brightness effect. Gamma calibration takes this non-linearity into account to ensure that the voltage command is accurately converted into the target brightness. For example, a voltage-brightness gamma lookup table (LUT) pre-built for this panel model is used. This LUT defines a non-linear mapping from digital voltage command to actual brightness. The optimal anode reset voltage is used as the theoretical voltage corresponding to the target brightness. A reverse lookup or interpolation calculation is performed in the LUT to obtain the calibration voltage that the driver IC actually needs to output to produce the brightness corresponding to the optimal anode reset voltage.

[0115] The publicly disclosed brightness compensation voltage, as a fine-tuning quantity based on actual measurements, can correct for potential systematic biases or individual anomalies in the prediction model, thereby improving safety. Furthermore, the gamma calibration step eliminates execution errors caused by the electro-optical nonlinearity of OLEDs, ensuring that the optimal voltage calculated in the digital world can be executed accurately in the physical world.

[0116] like Figure 17 Here is a complete flowchart of a display driver, combined with... Figure 17 The display driving method of this disclosure is described.

[0117] First, power on the device.

[0118] Then, obtain the gamma fine-tuning data of the previous adjacent brightness band (Band(n-1)) of the brightness band to be adjusted (Band(n)) in the display panel.

[0119] Next, feature selection is performed from the gamma fine-tuning data to obtain target feature data. Then, the residual model of the pre-trained prediction model is loaded, and the target feature data is input into the pre-trained prediction model for prediction processing to obtain the target anode reset voltage.

[0120] Furthermore, the illumination compensation voltage corresponding to the brightness frequency band to be adjusted is obtained; based on the target anode reset voltage and the illumination compensation voltage, the anode reset voltage to be adjusted is determined, and gamma calibration is performed based on the anode voltage to be adjusted to obtain the calibration voltage; the anode of the light-emitting element of the display panel is reset based on the calibration voltage.

[0121] Finally, if all the frequency bands requiring anode reset for brightness indication have been adjusted, then power off the device.

[0122] It should be noted that the specific implementation of each step has already been described in the above embodiments and will not be repeated here. Figure 18 The diagram shows a comparison between the time required for the anode reset voltage using the display driving method of this disclosure and the Tact time based on the OLED dynamic anode reset algorithm. The method of this disclosure can significantly shorten the Tact time.

[0123] After obtaining the trained model, the pre-trained prediction model can be deployed based on the production line environment. For example, if the production line environment is Lua / C, the pre-trained prediction model can also be saved as a cross-platform general (supporting C++ / Python) middleware file, so that the Lua environment can call the pre-trained prediction model through the corresponding compilation method to realize the display driving process of the display panel.

[0124] The display driving method in the exemplary embodiments of this disclosure acquires gamma fine-tuning data of adjacent brightness bands of the brightness band to be calibrated in the display panel, wherein the adjacent brightness bands are frequency bands adjacent to the brightness band to be calibrated and do not require dynamic adjustment of the anode reset voltage; performs feature selection from the gamma fine-tuning data to obtain target feature data; inputs the target feature data into a pre-trained prediction model for prediction processing to obtain the target anode reset voltage, wherein the pre-trained prediction model is trained using historical detection sample data containing gamma fine-tuning data of the brightness band and anode reset voltage values; and resets the anode of the light-emitting element of the display panel according to the target anode reset voltage. On the one hand, for the brightness band to be calibrated, acquiring existing gamma data of adjacent bands and instantaneously calculating the optimal anode reset voltage through the model reduces this step from minutes to seconds, directly compressing the critical path time of module production and improving the efficiency of anode reset. On the other hand, the model can uncover the relationship between gamma data and the optimal anode reset voltage. The predicted voltage value is theoretically closer to the global optimal solution, thereby improving display uniformity, reducing performance differences between different panels and batches, and improving the quality of anode reset.

[0125] In an exemplary embodiment of this disclosure, a display driving device is also provided. (See reference...) Figure 19 As shown, the display driving device 1900 may include a data acquisition module 1910, a feature selection module 1920, a voltage prediction module 1930, and a display processing module 1940. Specifically: The data acquisition module 1910 is used to acquire gamma fine-tuning data of adjacent brightness frequency bands of the brightness frequency band to be adjusted in the display panel, wherein the adjacent brightness frequency bands are those adjacent to the brightness frequency band to be adjusted and do not require dynamic adjustment of the anode reset voltage; the feature selection module 1920 is used to select features from the gamma fine-tuning data to obtain target feature data; the voltage prediction module 1930 is used to input the target feature data into a pre-trained prediction model for prediction processing to obtain the target anode reset voltage, wherein the pre-trained prediction model is trained using historical detection sample data containing gamma fine-tuning data of brightness frequency bands and anode reset voltage values; the display processing module 1940 is used to reset the anode of the light-emitting element of the display panel according to the target anode reset voltage.

[0126] In one exemplary embodiment of this disclosure, the process of training a pre-trained prediction model includes: acquiring the detected brightness frequency bands, gamma fine-tuning data corresponding to the brightness frequency bands to be detected, and anode reset voltages included in the detected display screen; performing feature selection on the gamma fine-tuning data corresponding to the detected brightness frequency bands to obtain reference feature data required for adjusting the anode reset voltage; inputting the reference feature data into the neural network model to be trained to obtain the predicted anode reset voltage of the detected brightness frequency bands; constructing a loss function based on the predicted anode reset voltage and the anode reset voltage corresponding to the brightness frequency bands to be detected, and adjusting the parameters in the neural network model to be trained based on the loss function to obtain the pre-trained prediction model.

[0127] In one exemplary embodiment of this disclosure, feature selection is performed on the gamma fine-tuning data corresponding to the detected brightness frequency band to obtain reference feature data, including: obtaining a first correlation relationship between R pixels of different gray levels in the gamma fine-tuning data corresponding to the brightness frequency band to be detected and the anode reset voltage corresponding to the brightness frequency band to be detected; determining a second correlation relationship between G pixels of different gray levels in the gamma fine-tuning data corresponding to the brightness frequency band to be detected and the anode reset voltage corresponding to the brightness frequency band to be detected; determining a third correlation relationship between B pixels of different gray levels in the gamma fine-tuning data corresponding to the brightness frequency band to be detected and the anode reset voltage corresponding to the brightness frequency band to be detected; and based on the first correlation relationship, the second correlation relationship, and the third correlation relationship, determining target pixel data with a high linear relationship with the anode reset voltage corresponding to the brightness frequency band to be detected as reference feature data.

[0128] In one exemplary embodiment of this disclosure, the neural network model to be trained includes an input layer network, a residual network, a global pooling network, and a regression output layer. Inputting reference feature data into the neural network model to obtain the predicted anode reset voltage for the detected brightness frequency band includes: preprocessing the reference feature data to obtain processed reference feature data; inputting the processed reference feature data into the input layer network and passing the processed reference feature data to the residual network; extracting features from the processed reference feature data layer by layer through multiple residual blocks in the residual network to obtain reference spatial features; converting the reference spatial features into channel feature vectors using the global pooling network; and mapping the channel feature vectors to the predicted anode reset voltage for the detected brightness frequency band based on the regression output layer.

[0129] In one exemplary embodiment of this disclosure, data preprocessing is performed on the reference feature data to obtain processed reference feature data, including: extracting a preset number of fine-tuning voltage values ​​at a specific grayscale from the reference feature data; normalizing each fine-tuning voltage value according to the display panel dimension to obtain a first processing result; and normalizing each first processing result according to the grayscale dimension to obtain processed reference feature data.

[0130] In one exemplary embodiment of this disclosure, feature selection from gamma-level fine-tuning data to obtain target feature data includes: filtering initial target-type pixel data from the gamma-level fine-tuning data; selecting pixel data covered by key regions from the initial target-type pixel data; and extracting data corresponding to key grayscale points from the selected data to obtain target feature data.

[0131] In one exemplary embodiment of this disclosure, obtaining gamma fine-tuning data of adjacent brightness bands of the brightness band to be adjusted in the display panel includes: obtaining gamma fine-tuning data of the previous adjacent brightness band of the brightness band to be adjusted; or, obtaining gamma fine-tuning data of multiple known brightness bands adjacent to the brightness band to be adjusted, and determining gamma fine-tuning data of adjacent brightness bands based on the gamma fine-tuning data of the multiple known brightness bands.

[0132] In one exemplary embodiment of this disclosure, resetting the anode of the light-emitting element of the display panel according to the target anode reset voltage includes: obtaining the illumination compensation voltage corresponding to the brightness frequency band to be adjusted; determining the anode reset voltage to be adjusted according to the target anode reset voltage and the illumination compensation voltage, and performing gamma calibration according to the anode voltage to be adjusted to obtain a calibration voltage; and resetting the anode of the light-emitting element of the display panel according to the calibration voltage.

[0133] Since the details of each functional module of the display driving device of the exemplary embodiments of this disclosure have been described in the exemplary embodiments of the display driving method described above, they will not be repeated here.

[0134] It should be noted that although several modules or units of the display driving device have been mentioned in the detailed description above, this division is not mandatory. In fact, according to embodiments of this disclosure, the features and functions of two or more modules or units described above can be embodied in one module or unit. Conversely, the features and functions of one module or unit described above can be further divided and embodied by multiple modules or units.

[0135] Exemplary embodiments of this disclosure also provide a computer program product. The computer program product includes a computer program that, when executed by a processor, implements the display driving method described above.

[0136] In one implementation, the computer program product can be a tangible product containing a computer program, such as a computer-readable storage medium storing the computer program. The readable storage medium can be a storage medium based on electrical, magnetic, optical, electromagnetic, infrared, or other signals, including but not limited to: random access memory (RAM), read-only memory (ROM), magnetic tape, floppy disk, flash memory, hard disk drive (HDD), solid-state drive (SSD), etc. For example, the computer program product can be implemented as a non-volatile storage medium storing a computer program, such as read-only memory, NAND flash memory, etc.

[0137] In one implementation, the computer program product can be an intangible product containing a computer program. For example, the computer program product can be implemented as a virtual digital product, such as an executable file, installation package, or other digital file storing the computer program.

[0138] Computer program code can be written in one or more programming languages. Examples of programming languages ​​include C, Java, and C++. Program code can execute entirely on the user's computing device, partially on the user's computing device, or as a standalone software package. It can also execute partially on the user's computing device and partially on a remote computing device, or entirely on a remote computing device or server. In cases involving remote computing devices, the remote computing device can be connected to the user's computing device via any type of network, such as a local area network (LAN) or a wide area network (WAN), or it can be connected to an external computing device (e.g., via an internet connection provided by a mobile network operator).

[0139] Computer programs can be carried or transmitted via signals such as electricity, magnetism, light, electromagnetic fields, and infrared radiation. Electronic devices can convert the signals carrying computer programs into digital signals, thereby running the computer programs. When a computer program runs on an electronic device, its code is used to cause the electronic device to execute (more specifically, to be executed by the processor of the electronic device) the method steps of various exemplary embodiments of this disclosure, such as the display driving method described above.

[0140] Furthermore, in exemplary embodiments of this disclosure, an electronic device capable of implementing the above-described methods is also provided. Those skilled in the art will understand that various aspects of this disclosure can be implemented as systems, methods, or program products. Therefore, various aspects of this disclosure can be specifically implemented as: entirely hardware embodiments, entirely software embodiments (including firmware, microcode, etc.), or embodiments combining hardware and software aspects, collectively referred to herein as "circuit," "module," or "system."

[0141] The following reference Figure 20To describe an electronic device 2000 according to such an embodiment of the present disclosure. Figure 20 The electronic device 2000 shown is merely an example and should not impose any limitation on the functionality and scope of use of the embodiments disclosed herein.

[0142] like Figure 20 As shown, the electronic device 2000 is manifested in the form of a general-purpose computing device. The components of the electronic device 2000 may include, but are not limited to: at least one processing unit 2010, at least one storage unit 2020, a bus 2030 connecting different system components (including storage unit 2020 and processing unit 2010), and a display unit 2040.

[0143] The storage unit stores program code that can be executed by the processing unit 2010, causing the processing unit 2010 to perform the steps described in the "Exemplary Methods" section above according to various exemplary embodiments of this disclosure.

[0144] Storage unit 2020 may include readable media in the form of volatile storage units, such as random access memory (RAM) 2021 and / or cache memory 2022, and may further include read-only memory (ROM) 2023.

[0145] The storage unit 2020 may also include a program / utility 2024 having a set (at least one) program module 2025, such program module 2025 including but not limited to: an operating system, one or more applications, other program modules, and program data, each or some combination of these examples may include an implementation of a network environment.

[0146] Bus 2030 can represent one or more of several types of bus structures, including memory cell bus or memory cell controller, peripheral bus, graphics acceleration port, processing unit, or local bus using any of the multiple bus structures.

[0147] Electronic device 2000 can also communicate with one or more external devices 2100 (e.g., keyboard, pointing device, Bluetooth device, etc.), one or more devices that enable a user to interact with electronic device 2000, and / or any device that enables electronic device 2000 to communicate with one or more other computing devices (e.g., router, modem, etc.). This communication can be performed via input / output (I / O) interface 2050. Furthermore, electronic device 2000 can also communicate with one or more networks (e.g., local area network (LAN), wide area network (WAN), and / or public networks, such as the Internet) via network adapter 2060. As shown, network adapter 2060 communicates with other modules of electronic device 2000 via bus 2030. It should be understood that, although not shown in the figures, other hardware and / or software modules can be used in conjunction with electronic device 2000, including but not limited to: microcode, device drivers, redundant processing units, external disk drive arrays, RAID systems, tape drives, and data backup storage systems.

[0148] From the above description of the embodiments, those skilled in the art will readily understand that the exemplary embodiments described herein can be implemented by software or by combining software with necessary hardware. Therefore, the technical solutions according to the embodiments of this disclosure can be embodied in the form of a software product, which can be stored in a non-volatile storage medium (such as a CD-ROM, USB flash drive, external hard drive, etc.) or on a network, including several instructions to cause a computing device (such as a personal computer, server, terminal device, or network device, etc.) to execute the methods according to the embodiments of this disclosure.

[0149] Furthermore, the above figures are merely illustrative of the processes included in the method according to exemplary embodiments of this disclosure and are not intended to be limiting. It is readily understood that the processes shown in the above figures do not indicate or limit the temporal order of these processes. Additionally, it is readily understood that these processes may be executed synchronously or asynchronously, for example, in multiple modules.

[0150] Other embodiments of this disclosure will readily occur to those skilled in the art upon consideration of the specification and practice of the invention disclosed herein. This disclosure is intended to cover any variations, uses, or adaptations of this disclosure that follow the general principles of this disclosure and include common knowledge or customary techniques in the art not disclosed herein. The specification and embodiments are to be considered exemplary only, and the true scope and spirit of this disclosure are indicated by the claims.

Claims

1. A display driving method, characterized in that, include: Obtain gamma fine-tuning data of the adjacent brightness bands of the brightness band to be adjusted in the display panel, wherein the adjacent brightness bands are the frequency bands that are adjacent to the brightness band to be adjusted and do not require dynamic adjustment of the anode reset voltage; Feature selection is performed on the gamma-ray fine-tuning data to obtain target feature data; The target feature data is input into a pre-trained prediction model for prediction processing to obtain the target anode reset voltage. The pre-trained prediction model is trained using gamma fine-tuning data including the brightness frequency band and historical detection sample data of anode reset voltage values. The anode of the light-emitting element of the display panel is reset according to the target anode reset voltage.

2. The method according to claim 1, characterized in that, The process of training the pre-trained prediction model includes: Acquire the detected brightness frequency bands, the gamma fine-tuning data corresponding to the brightness frequency bands to be detected, and the anode reset voltage from the tested display screen; Feature selection is performed on the gamma fine-tuning data corresponding to the detected brightness frequency band to obtain reference feature data required for adjusting the anode reset voltage; The reference feature data is input into the neural network model to be trained to obtain the predicted anode reset voltage of the detected brightness frequency band; A loss function is constructed based on the predicted anode reset voltage and the anode reset voltage corresponding to the brightness frequency band to be detected, and the parameters in the neural network model to be trained are adjusted based on the loss function to obtain the pre-trained prediction model.

3. The method according to claim 2, characterized in that, The step of performing feature selection on the gamma fine-tuning data corresponding to the detected brightness frequency band to obtain reference feature data includes: Obtain the first correlation relationship between R pixels of different gray levels in the gamma fine-tuning data corresponding to the brightness frequency band to be detected and the anode reset voltage corresponding to the brightness frequency band to be detected; Determine a second correlation relationship between G pixels of different gray levels in the gamma fine-tuning data corresponding to the brightness frequency band to be detected and the anode reset voltage corresponding to the brightness frequency band to be detected; A third correlation relationship is determined between B pixels of different gray levels in the gamma fine-tuning data corresponding to the brightness frequency band to be detected and the anode reset voltage corresponding to the brightness frequency band to be detected; Based on the first correlation relationship, the second correlation relationship, and the third correlation relationship, target pixel data that has a high linear relationship with the anode reset voltage corresponding to the brightness frequency band to be detected is determined as the reference feature data.

4. The method according to claim 2, characterized in that, The neural network model to be trained includes an input layer network, a residual network, a global pooling network, and a regression output layer; the step of inputting the reference feature data into the neural network model to be trained to obtain the predicted anode reset voltage of the detected brightness frequency band includes: The reference feature data is preprocessed to obtain the processed reference feature data; The processed reference feature data is input into the input layer network, and then transmitted to the residual network through the input layer network. The reference space features are obtained by extracting features from the processed reference feature data layer by layer through multiple residual blocks in the residual network. The reference spatial features are converted into channel feature vectors using the global pooling network, and the channel feature vectors are mapped to the predicted anode reset voltage of the detected brightness band based on the regression output layer.

5. The method according to claim 4, characterized in that, The step of preprocessing the reference feature data to obtain processed reference feature data includes: Extract a preset number of fine-tuning voltage values ​​at a specific gray level from the reference feature data; The fine-tuning voltage values ​​are normalized according to the dimensions of the display panel to obtain the first processing result; The first processing results are normalized according to the grayscale dimension to obtain the processed reference feature data.

6. The method according to claim 1, characterized in that, The step of selecting features from the gamma-ray fine-tuning data to obtain target feature data includes: Filter the initial target class pixel data from the gamma fine-tuning data; The target feature data is obtained by selecting the pixel data covered by the key region from the initial target pixel data and extracting the data corresponding to the key grayscale points from the selected data.

7. The method according to claim 1, characterized in that, The acquisition of gamma fine-tuning data of adjacent brightness frequency bands of the brightness frequency band to be adjusted in the display panel includes: Obtain the gamma fine-tuning data of the previous adjacent brightness frequency band of the brightness frequency band to be adjusted; Alternatively, obtain gamma fine-tuning data of multiple known brightness frequency bands adjacent to the brightness frequency band to be adjusted, and determine the gamma fine-tuning data of the adjacent brightness frequency band based on the gamma fine-tuning data of the multiple known brightness frequency bands.

8. The method according to any one of claims 1 to 7, characterized in that, The step of resetting the anode of the light-emitting element of the display panel according to the target anode reset voltage includes: Obtain the illumination compensation voltage corresponding to the brightness frequency band to be adjusted; Based on the target anode reset voltage and the luminescence compensation voltage, the anode reset voltage to be adjusted is determined, and gamma calibration is performed based on the anode position voltage to be adjusted to obtain the calibration voltage. The anode of the light-emitting element of the display panel is reset according to the calibration voltage.

9. A display driving device, characterized in that, include: The data acquisition module is used to acquire the gamma fine-tuning data of the adjacent brightness frequency bands of the brightness frequency band to be adjusted in the display panel, wherein the adjacent brightness frequency bands are the frequency bands that are adjacent to the brightness frequency band to be adjusted and do not require dynamic adjustment of the anode reset voltage. The feature selection module is used to select features from the gamma fine-tuning data to obtain target feature data; The voltage prediction module is used to input the target feature data into a pre-trained prediction model for prediction processing to obtain the target anode reset voltage. The pre-trained prediction model is trained using gamma fine-tuning data including the brightness frequency band and historical detection sample data of the anode reset voltage value. The display processing module is used to reset the anode of the light-emitting element of the display panel according to the target anode reset voltage.

10. A computer program product, comprising a computer program, characterized in that, When the computer program is executed by a processor, it implements the method described in any one of claims 1 to 8.

11. An electronic device, characterized in that, include: processor; as well as Memory for storing the executable instructions of the processor; The processor is configured to perform the method of any one of claims 1 to 8 by executing the executable instructions.

Images