Self-luminous display device and display correction method

Abstract

A self-luminous display device (1) comprises: an acquisition unit (111) that acquires at least some values extracted from image data pertaining to images displayed on a self-luminous display panel (20) and reference temperature data acquired by a temperature sensor (30) disposed in a predetermined position; and a prediction unit (112) that uses the acquired at least some image data and reference temperature data to predict a panel in-plane temperature of the self-luminous display panel (20).

Description

Self-luminous display device and display correction method

[0001] The present disclosure relates to a self-luminous display device and a display correction method.

[0002] Image display devices include self-luminous display devices that display images using self-luminous display pixels, and non-self-luminous display devices that irradiate non-luminous pixels with a backlight and display images using the transmitted light.

[0003] It has been known that self-luminous display elements deteriorate due to factors such as the current flowing through the display element, the temperature caused by heat generated by the ambient temperature and the current, and the duration of exposure to these factors, resulting in a decrease in the amount of light emitted. Therefore, the amount of light emitted may be corrected according to the degree of deterioration. For example, to correct the decrease in the amount of light emitted due to heat (temperature), which has a significant impact on deterioration, the temperature of a display panel on which display elements are arranged is measured, and the temperature data is accumulated to estimate the degree of deterioration. For example, Patent Document 1 discloses a display device that estimates the temperature distribution of the display portion of a display panel based on the ambient temperature and the elapsed time the display panel has been displayed, and corrects the driving conditions according to the estimated temperature distribution.

[0004] Japanese Patent Publication No. 2020-021032

[0005] The technology disclosed in Patent Document 1 estimates the temperature distribution of the display unit based on one of multiple pre-created temperature distribution tables that corresponds to the acquired environmental temperature and driving time. However, since the temperature distribution table is not related to the type or pattern of the displayed image and the temperature distribution is estimated for each large area of ​​the display panel, it is unclear how accurate the estimation can be.

[0006] An object of the present disclosure is to provide a technique that can predict the temperature within the panel surface of a display panel of a self-luminous display device with greater accuracy than conventional techniques.

[0007] In order to solve the above problems, a self-luminous display device according to one embodiment of the present disclosure includes an acquisition unit that acquires at least a portion of values ​​extracted from image data relating to an image displayed on a self-luminous display panel and reference temperature data acquired by a temperature sensor positioned at a predetermined position, and a prediction unit that predicts the temperature within the panel surface of the self-luminous display panel using at least a portion of the values ​​extracted from the acquired image data and the reference temperature data.

[0008] A display correction method according to one aspect of the present disclosure includes an acquisition step of acquiring at least a portion of values ​​extracted from image data relating to an image displayed on a self-luminous display panel and reference temperature data acquired by a temperature sensor positioned at a predetermined position; a prediction step of predicting an in-plane temperature of the self-luminous display panel using at least a portion of values ​​extracted from the acquired image data and the reference temperature data; and a correction step of correcting a display signal of the image displayed on the self-luminous display panel using the predicted in-plane temperature of the panel.

[0009] The control device 10, 10A and learning device 2 according to each aspect of the present disclosure may be realized by a computer. In this case, the control program of the control device 10, 10A and learning device 2 that realizes the control device 10, 10A and learning device 2 on a computer by operating the computer as each part (software element) of the control device 10, 10A and learning device 2, and the computer-readable recording medium on which it is recorded, also fall within the scope of the present disclosure.

[0010] According to one aspect of the present disclosure, the temperature within the panel surface of a display panel of a self-luminous display device can be predicted more accurately than with conventional techniques.

[0011] 1 is a block diagram showing a configuration of a self-luminous display device 1 according to a first embodiment of the present disclosure. FIG. 1 is a flowchart showing the flow of an image display correction method S1 according to the first embodiment. FIG. 2 is a flowchart showing the flow of a display correction method S2 that takes into account deterioration of a display panel. FIG. 3 is a flowchart showing the flow of a display correction method S3 that takes into account deterioration of a display panel. FIG. 4 is a block diagram showing a configuration of a self-luminous display device 1A in which a prediction unit is provided with a machine learning model. FIG. 5 is a schematic diagram showing an example of explanatory variables input to a machine learning model and an objective variable output from the machine learning model. FIG. 6 is a schematic diagram showing another example of explanatory variables input to a machine learning model and an objective variable output from the machine learning model. FIG. 7 is a flowchart showing the flow of a display correction method S4 that uses a machine learning model. FIG. 8 is a flowchart showing the flow of a machine learning model construction method S5. FIG. 9 is a flowchart showing the flow of a machine learning model construction method S6 that uses time-lapse measurement panel in-plane temperature data corresponding to continuous display time. FIG. 10 is a flowchart showing the flow of a machine learning model construction method S7 that predicts the panel in-plane temperature of a display panel on which moving images are displayed. FIG. 11 is a block diagram showing a configuration of a self-luminous display device according to a fourth embodiment of the present disclosure. FIG. 12 is a schematic diagram showing an example of explanatory variables input to a machine learning model and an objective variable output from the machine learning model. Fig. 1 is a block diagram showing the configuration of a learning device for constructing a machine learning model; Fig. 2 is a flowchart showing the flow of a method for constructing a machine learning model; Fig. 3 is an example of a learning dataset for training a machine learning model; and Fig. 4 is a flowchart showing the flow of a display correction method using a machine learning model.

[0012] [Embodiment 1] (Self-luminous display device) Hereinafter, an embodiment of the present disclosure will be described in detail with reference to the drawings. FIG. 1 is a block diagram showing the configuration of a self-luminous display device 1 according to embodiment 1. The self-luminous display device 1 is a display device including self-luminous elements such as OLEDs (organic light-emitting diodes), QLEDs (quantum dot light-emitting diodes), and micro-LEDs as display elements of a self-luminous display panel 20. The self-luminous display device 1 can predict the temperature within the panel surface of the self-luminous display panel 20. The self-luminous display device 1 may also estimate deterioration of the self-luminous elements and perform image display correction. As shown in the figure, the self-luminous display device 1 includes a control device 10, a self-luminous display panel 20, and a temperature sensor 30.

[0013] (Control Device 10) The control device 10 controls the entire self-luminous display device 1. As shown in FIG. 1 , the control device 10 includes a main control unit 11 and a memory 12. The memory 12 may include multiple types of memory, such as a read-only memory (ROM) and a random access memory (RAM). The memory 12 may also include an internal or external memory, such as a hard disk drive (HDD) or a solid state drive (SSD). The main control unit 11 includes at least one processor. The processor may be configured using a general-purpose processor, such as a micro processing unit (MPU) or a central processing unit (CPU). The processor may also include a dedicated processor, such as an application specific integrated circuit (ASIC), a field programmable gate array (FPGA), or a programmable logic device (PLD).

[0014] The main control unit 11 includes an acquisition unit 111, a prediction unit 112, a display correction unit 114, and a display panel drive unit 116. The main control unit 11 may further include a deterioration estimation unit 113. The main control unit 11 may further include a display time measurement unit 115. As an example, the processor implements the functions of the acquisition unit 111, the prediction unit 112, the deterioration estimation unit 113, the display correction unit 114, the display time measurement unit 115, and the display panel drive unit 116 by loading various control programs recorded in the ROM of the memory 12 into the RAM and executing them.

[0015] (Acquisition Unit 111) The acquisition unit 111 acquires image data (image signals) input from an external device. The acquisition unit 111 sends the acquired image signals to the display panel drive unit 116. The acquisition unit 111 also acquires various information used to estimate deterioration of the display panel 20. For example, the acquisition unit 111 acquires at least a portion of values ​​extracted from image data relating to an image displayed on the self-luminous display panel 20 (hereinafter also simply referred to as the "display panel 20") and sends the acquired values ​​to the prediction unit 112. The value extracted from the image data may be, for example, a current value extracted from a control signal of the display panel drive unit 116 or a pixel gradation value converted from the current value. "At least a portion of the extracted values" may be all of the extracted values, or, for example, an image may be divided into multiple regions and a portion of the values ​​extracted from the image data relating to each image may be selected. Any method for selecting a portion of the extracted values ​​may be used. Using a portion of the extracted values ​​can reduce the amount of data processing, thereby reducing memory requirements and data processing time. Hereinafter, "at least a portion of the extracted value" may be simply referred to as "extracted value." Also, "value extracted from image data" may be simply referred to as "image data."

[0016] The acquisition unit 111 also sends the temperature acquired by the temperature sensor 30 placed at a predetermined position on the self-luminous display device 1 as reference temperature data to the prediction unit 112. The reference temperature is the temperature at a predetermined position, and is the temperature acquired by the temperature sensor 30 placed at the predetermined position. The predetermined position may be inside or outside the self-luminous display device 1, or may be a position spaced apart from the self-luminous display device 1.

[0017] (Prediction Unit 112) The prediction unit 112 predicts the in-plane temperature of the display panel 20 using at least a portion of the values ​​extracted from the image data acquired from the acquisition unit 111 and the reference temperature data. The in-plane temperature is the temperature of the panel surface of the display panel 20. Furthermore, the panel surface temperature may be the temperature of the glass surface if a cover glass or the like is present, or the temperature of the surface of the display element (i.e., inside the cover glass). The panel surface temperature may be the temperature of a location corresponding to the pixel position, or a representative temperature of each region when the panel surface is divided into multiple regions. While it is preferable to predict the temperatures of all locations corresponding to the pixel positions, this increases the amount of data processing. Therefore, taking data processing time into consideration, the panel may be divided into multiple regions and a representative temperature may be predicted for each region. The region division method is arbitrary. For example, the display surface of the display panel 20 may be divided into nine 3x3 regions, 16 4x4 regions, or a greater number of regions. The in-plane temperature may be the temperature of a pixel position within the display panel 20 or a representative temperature of each region. The method by which the prediction unit 112 predicts the temperature within the panel surface of the display panel 20 using at least a part of the image data and the reference temperature data will be described later.

[0018] Here, image data will be described. The image data may be data relating to at least one of the color and gradation of the image displayed on the display panel 20. In other words, the image data is data relating to at least one of the color and gradation of the displayed image itself, and is digital data indicating the color and gradation. The prediction unit 112 may predict the temperature within the panel surface using part or all of this data. When using part of the data, for example, the prediction unit 112 divides the display panel 20 into multiple regions and predicts the temperature within the panel surface using part of the data of each region as data representing that region. Using the color or gradation of the displayed image itself is useful when using a machine learning model, which will be described later.

[0019] The values ​​extracted from the image data may be multiple current values ​​or grayscale values ​​of one or more pixels or one or more regions of the display panel 20. A region refers to each of the multiple sections of the display panel 20. The current value or grayscale value of a region may be the average value of the current values ​​applied to the pixels in that region or a grayscale value corresponding to that average value (hereinafter referred to as the average grayscale value). The average value is the average value of the current values ​​of the R pixels (pixels that emit red light, hereinafter referred to as R), G pixels (pixels that emit green light, hereinafter referred to as G), and B pixels (pixels that emit blue light, hereinafter referred to as B) in that region over a predetermined period of time. The average value may be the average value for each of the R, G, and B colors, or a value equivalent to the assumed gray color. The prediction unit 112 may predict the temperature within the panel surface using some or all of this data.

[0020] In this case, the acquisition unit 111 acquires an average current value or an average gradation value for one or more pixels or one or more regions of the display panel 20. When predicting the temperature of the display panel 20, the prediction may be made on a pixel-by-pixel basis or on a region-by-region basis. The acquisition unit 111 acquires the average current value or average gradation value for each pixel or region and sends it to the prediction unit 112. The prediction unit 112 can predict the temperature for each pixel or region by referring to this information. By using the current value or gradation value in the region, the amount of data processing can be reduced compared to when the current value or gradation value is used for each pixel, and as a result, the required memory amount and prediction processing time can be shortened.

[0021] Here, we will explain the advantages of using a portion (or representative value) of the values ​​extracted from the image data. While using pixel-by-pixel data is preferable for improving prediction accuracy, using pixel-by-pixel data increases the amount of data processing and the processing time. This impact becomes particularly significant as the number of pixels increases. On the other hand, depending on the type of image to be displayed and the characteristics of the display, a sophisticated method that takes pixel-by-pixel data into account may not necessarily be necessary. The temperature within the panel does not necessarily vary significantly from pixel to pixel, and it is believed that in many cases treating multiple pixels together as a single piece of data poses almost no problems and is sufficient.

[0022] Therefore, it is preferable to consider the possibility that inputting all the data for each pixel may have a significant negative impact on the amount of data processing and processing time, and to determine the method for selecting the values ​​to extract from the image by comprehensively considering the type of image, display characteristics, processing amount, processing time, etc.

[0023] There is also the possibility that defects or bright spots exist in specific pixels on the panel. In this case, if data for each pixel is input as an explanatory variable, the defective or bright pixel will become an outlier, and there is a concern that appropriate learning and prediction will not be performed. Therefore, by extracting a single value from multiple pixels, it is expected that the influence of such outliers can be reduced. In other words, by dividing the image into multiple regions and using a portion or representative value of the image data representing each region to predict the representative temperature for each region, the influence of outliers can be reduced.

[0024] (Deterioration Estimation Unit 113) The deterioration estimation unit 113 estimates the degree of deterioration of the display panel 20 by referring to the panel in-plane temperature predicted by the prediction unit 112. Alternatively, the deterioration estimation unit 113 may estimate the degree of deterioration of the display panel 20 by referring to the display time of an image measured by the display time measurement unit 115 (described later) or the cumulative value of the display time, in addition to the predicted panel in-plane temperature. The deterioration estimation unit 113 may estimate the degree of deterioration for each pixel of the display panel 20 or for each pre-defined region. Estimating the degree for each region reduces the amount of information processing and shortens the estimation processing time of the deterioration estimation unit 113. A large amount of data has been previously acquired for each type of display element regarding the degree of deterioration due to temperature, the degree of deterioration due to current, and the degree of deterioration due to the cumulative value of display time at a certain temperature or current value. Therefore, the deterioration estimation unit 113 can estimate the degree of deterioration for each pixel or region using the data previously acquired.

[0025] (Display Correction Unit 114) The display correction unit 114 corrects the image to be displayed by referring to the degree of deterioration estimated by the deterioration estimation unit 113. Specifically, the display correction unit 114 derives the amount of deterioration compensation for the drive signal to be generated by the display panel drive unit 116 by referring to the estimated degree of deterioration. As described above, the degree of deterioration is estimated based on the temperature within the panel surface, the current value (grayscale value), and the cumulative value of the display time under those conditions. The amount of deterioration compensation is the amount of current or grayscale value that is increased to compensate for the decrease in the amount of light emission due to deterioration. Data on the derived amount of deterioration compensation for the drive signal is sent to the display panel drive unit 116. The display panel drive unit 116 generates the drive signal by adding the amount of deterioration compensation to the acquired drive signal.

[0026] The value extracted from the image data acquired by the acquisition unit 111 may be a plurality of current values ​​or grayscale values ​​to which the amount of degradation compensation for the pixels of the display panel 20 is added.

[0027] The display correction unit 114 may correct the image displayed on the display panel 20 by referring to the panel in-plane temperature predicted by the prediction unit 112. The display correction unit 114 may correct the image displayed on the display panel 20 by referring to the degree of deterioration estimated by the deterioration estimation unit 113 and the panel in-plane temperature predicted by the prediction unit 112.

[0028] (Display Panel Driving Unit 116) The display panel driving unit 116 generates a driving signal for driving the display panel 20 based on the image signal received from the acquiring unit 111, and applies a voltage to each pixel of the display panel 20. As a result, an image is displayed on the display panel 20.

[0029] (Display time measurement unit 115) The display time measurement unit 115 measures the display time for which an image is displayed on the display panel 20. The display time is the time from when the power of the display panel 20 is turned on and a certain type of image begins to be displayed. The display time may be the time for which the image is continuously displayed. The measured display time may be acquired by the acquisition unit 111 and recorded in the display history storage unit 121 described below. The acquisition unit 111 may acquire the display time at any time while an image is being displayed on the display panel 20 and record the acquired display time in the display history storage unit 121. Alternatively, the acquisition unit 111 may acquire the display time for which an image was displayed when a certain type of image has finished being displayed on the display panel 20, and record the acquired display time in the display history storage unit 121.

[0030] The prediction unit 112 may predict the temperature within the panel surface using the display time during which an image is displayed on the display panel 20, in addition to the image data and the reference temperature data. The temperature within the panel surface generally follows a curve that gradually rises after an image begins to be displayed and then saturates, so the temperature changes depending on the display time immediately after the image begins to be displayed. Therefore, by taking the display time into consideration, the temperature within the panel surface can be predicted with greater accuracy.

[0031] (Display History Storage Unit 121) The memory 12 may include a display history storage unit 121. The display history storage unit 121 stores information related to the display history of the display panel 20. The information related to the display history may include information related to a displayed image and the display time of the image, a predicted in-plane temperature of the display panel 20, and the duration of the predicted temperature for each pixel or region when an image is displayed on the display panel 20.

[0032] The information related to the display history is also information used to estimate deterioration of the display panel 20. For example, the display history storage unit 121 associates and stores the average current value or average gray level value for one or more pixels or one or more regions of the display panel 20 acquired by the acquisition unit 111, the panel in-plane temperature predicted by the prediction unit 112, and the display time measured by the display time measurement unit 115 under these conditions. The display history storage unit 121 may also store reference temperature data acquired by the acquisition unit 111. The display history storage unit 121 may also store the panel in-plane temperature of the display panel 20 predicted by the prediction unit 112.

[0033] The deterioration estimation unit 113 may have a display history update function that accumulates new data and updates the data stored in the display history storage unit 121. For example, the deterioration estimation unit 113 may acquire the display time of an image measured by the display time measurement unit 115 and add it to the display time stored in the display history storage unit 121 to update the display time. As a result, the display history storage unit 121 stores the cumulative value of the time an image is displayed on the display panel 20. Alternatively, the deterioration estimation unit 113 may acquire the average current value or average gradation value for each pixel or region, and the predicted in-plane temperature of the panel, and update the display time stored in the display history storage unit 121 for each different average current value or average gradation value and in-plane temperature of the panel. As a result, the deterioration estimation unit 113 can estimate the degree of deterioration of the display panel 20 by referring to the cumulative values ​​(accumulated data) of the average current value or average gradation value, the in-plane temperature of the panel, and the display time.

[0034] Alternatively, the deterioration estimation unit 113 may calculate the degree of deterioration each time a display is performed, and store only the accumulated deterioration amount updated as needed in the display history storage unit 121. Specifically, the deterioration estimation unit 113 may calculate the degree of deterioration for a predetermined period of time each time a display is performed on the display panel 20, and add the calculated degree of deterioration to the accumulated deterioration amount recorded in the display history storage unit 121. The predetermined period may be, for example, a predetermined time, or may be the time from when the display panel 20 is powered on to when the power is turned off. The accumulated deterioration amount is data obtained by accumulating and recording the amount of deterioration of the display panel 20 since the display panel 20 began to be used. The accumulated deterioration amount may be the amount of deterioration accumulated for each pixel, or may be the amount of deterioration evaluated and accumulated for each predetermined region. The deterioration estimation unit 113 adds the calculated amount of deterioration of the display panel 20 to the accumulated deterioration amount recorded in the display history storage unit 121. In the case of a format in which only the cumulative deterioration amount is stored in the display history memory unit 121, by updating the cumulative deterioration amount from time to time, there is no need to store data on the average current value or average gradation value, the temperature within the panel surface, and the cumulative value of the display time in the display history memory unit 121, and therefore the amount of data stored in the display history memory unit 121 can be reduced.

[0035] (Self-luminous display panel 20) The self-luminous display panel 20 is a display panel in which self-luminous elements are arranged as pixels, and is driven by the display panel drive unit 116. A self-luminous element may be provided for each of the R, G, and B elements of each pixel. The type of self-luminous element is not limited, and may be, for example, a self-luminous element such as an OLED, QLED, or micro LED, as described above. In other words, the self-luminous display panel 20 may be a panel including any of organic light-emitting diodes, quantum dot light-emitting diodes, and micro light-emitting diodes.

[0036] (Temperature Sensor 30) The temperature sensor 30 is disposed at a predetermined arbitrary position. The temperature sensor 30 measures the temperature when an image is displayed on the display panel 20 and sends the measured temperature to the acquisition unit 111. The acquisition unit 111 acquires this temperature as reference temperature data and sends it to the prediction unit 112. The prediction unit 112 predicts the in-plane temperature of the display panel 20 using the reference temperature data and at least a portion of the image data. There may be multiple temperature sensors 30. In other words, there may be multiple reference temperature data.

[0037] The reference temperature data may be temperature data acquired by temperature sensors disposed at one or more locations on the display panel. Alternatively, the reference temperature data may be environmental temperature data acquired by temperature sensors disposed around the display panel. The environmental temperature data may be room temperature or outdoor temperature.

[0038] According to the self-luminous display device 1 having the above-described configuration, the temperature within the panel surface of the display panel 20 is predicted using at least a portion of the information in the image data of the image displayed on the display panel 20, so that the temperature within the panel surface can be predicted more accurately than with conventional technology.

[0039] (Image Display Correction Method) Next, an image display correction method performed by the self-luminous display device 1 will be described with reference to the drawings. Fig. 2 is a flowchart showing the flow of the image display correction method S1 according to this embodiment. As shown in Fig. 2, the display correction method S1 includes the following steps:

[0040] Step S11 is an acquisition step in which the acquisition unit 111 acquires at least a portion of values ​​extracted from image data relating to an image displayed on the display panel 20 and reference temperature data acquired by the temperature sensor 30 disposed at a predetermined position. The image data and the reference temperature are as described above.

[0041] Step S12 is a prediction step in which the prediction unit 112 predicts the in-plane temperature of the display panel 20 using at least a part of the values ​​extracted from the acquired image data and the reference temperature data. The method by which the prediction unit 112 predicts the in-plane temperature of the display panel 20 will be described later.

[0042] Step S13 is a correction step in which the display correction unit 114 corrects the display signal of the image to be displayed on the display panel 20, using the panel in-plane temperature predicted by the prediction unit 112. The method of correcting the display signal performed by the display correction unit 114 is as described above.

[0043] Steps S11 and S12 are performed in parallel. The predicted temperature within the panel surface changes as the image data and reference temperature data change. When the temperature within the panel surface changes, the amount of correction in the correction step of step S13 also changes.

[0044] According to the above-described display correction method S1, the temperature within the panel surface of the display panel 20 is predicted using at least a portion of the information in the image data of the image displayed on the display panel 20, so that the temperature within the panel surface of the display panel 20 can be predicted more accurately, and the display correction of the image can also be performed more accurately.

[0045] In step S13 described above, the display correction unit 114 corrects the display signal of the image displayed on the display panel 20 using the panel in-plane temperature. However, a display correction method S2 that further takes into account the deterioration of the display panel 20 may also be used. FIG. 3 is a flowchart showing the flow of the display correction method S2 that takes into account the deterioration of the display panel 20. As shown in FIG. 3, the display correction method S2 includes a step of estimating the deterioration of the display panel 20.

[0046] In step S21, the acquiring unit 111 further acquires the display time during which the image was displayed on the display panel 20. That is, this is an acquiring step in which the acquiring unit 111 acquires at least a portion of the values ​​extracted from the image data related to the image displayed on the display panel 20, the reference temperature data acquired by the temperature sensor disposed at a predetermined position, and the display time during which the image was displayed on the display panel 20.

[0047] Step S22 is a prediction step in which the prediction unit 112 predicts the in-plane temperature of the display panel 20 using at least a part of the values ​​extracted from the acquired image data and the reference temperature data. This step is similar to step S12 in the display correction method S1.

[0048] Step S23 is a degradation estimation step in which the degradation estimation unit 113 estimates the degree of degradation of the display panel 20 using display image data (current amount, more specifically, average current value or average gradation value for one or more pixels or one or more regions), the temperature within the panel surface, and the display time.

[0049] Step S24 is a correction step in which display correction unit 114 further refers to the degree of deterioration and corrects the display signal of the image displayed on display panel 20. According to the display correction method S2 described above, by taking into consideration the deterioration of display panel 20 due to display image data (amount of current), temperature within the panel surface, and display time, image display correction can also be performed with higher accuracy.

[0050] The acquisition unit 111 may acquire, as image data, an average current value or an average gray level value for one or more pixels or one or more regions of the display panel 20, a predicted in-plane temperature of the display panel 20, and cumulative data of the display time under those conditions. Then, the deterioration estimation unit 113 may estimate deterioration of the display panel 20 using the cumulative data.

[0051] 4 is a flowchart showing the flow of a display correction method S3 for estimating deterioration of the display panel 20 and correcting the display using an average current value or average gray level value for one or more pixels or one or more regions of the display panel 20, a predicted in-plane temperature of the display panel 20, and cumulative data on the display time under those conditions. As shown in FIG. 4, the display correction method S3 includes a step of estimating the degree of deterioration of the display panel 20 using the cumulative data.

[0052] Step S31 is a step in which the acquisition unit 111 acquires, as values ​​extracted from image data, an average current value or an average gradation value for one or more pixels or one or more regions of the display panel 20, a predicted in-plane temperature of the panel, and cumulative data on the display time during which an image is displayed on the display panel 20 under those conditions. In other words, step S31 is an acquisition step in which the acquisition unit 111 acquires, as values ​​extracted from image data, an average current value or an average gradation value for one or more pixels or one or more regions of the display panel 20, a predicted in-plane temperature of the panel from reference temperature data acquired by the temperature sensor 30 disposed at a predetermined position, and cumulative data on the display time during which an image is displayed on the display panel 20 under those conditions.

[0053] Step S32 is an estimation step in which the deterioration estimation unit 113 estimates the degree of deterioration of the display panel 20 using the accumulated data.

[0054] Step S33 is a correction step in which the display correction unit 114 corrects the display signal of the image to be displayed on the display panel 20 by further referring to the degree of deterioration.

[0055] According to the display correction method S3 described above, it is possible to refer to the amount of deterioration that takes into consideration the overall deterioration due to past lighting, and therefore it is possible to perform more accurate correction of the image display.

[0056] As explained in the display correction methods S2 and S3, when the deterioration estimation unit 113 estimates the degree of cumulative deterioration of the display panel 20 using cumulative data, it is not necessary to use the panel in-plane temperature predicted at the time of correction for image display correction. In other words, the cumulative deterioration amount is derived taking into consideration the panel in-plane temperature, display time, and also the average current value or average gradation value, but when correcting image display using the cumulative deterioration amount, it is not necessary to use the panel in-plane temperature at the time of correction.

[0057] For example, a display correction method S2A that does not use the panel in-plane temperature at the time of correction when correcting the display of an image may include an acquisition step S21A in which the acquisition unit 111 further acquires the display time during which the image is displayed on the display panel 20 in the acquisition step; a prediction step S22A in which the prediction unit 112 predicts the panel in-plane temperature of the display panel 20 using at least a portion of the values ​​extracted from the acquired image data and reference temperature data; a deterioration estimation step S23A in which the deterioration estimation unit 113 estimates the degree of deterioration of the display panel 20 using the display image data (amount of current), the panel in-plane temperature, and the display time; and a correction step S24A in which the display correction unit 114 corrects the display signal of the image displayed on the display panel 20 by referring only to the estimated degree of deterioration.

[0058] Furthermore, a display correction method S3A that does not use the panel in-plane temperature at the time of correction when correcting the display of an image may include an acquisition step S31A in which the acquisition unit 111 acquires an average current value or average gradation value for one or more pixels or one or more regions of the display panel 20 as values ​​extracted from image data, an in-plane temperature of the panel predicted from reference temperature data acquired by a temperature sensor 30 arranged at a predetermined position, and cumulative data on the display time during which an image is displayed on the display panel 20 under these conditions; an estimation step S32A in which the deterioration estimation unit 113 estimates the degree of deterioration of the display panel 20 using the cumulative data; and a correction step S33A in which the display correction unit 114 corrects the display signal of the image displayed on the display panel 20 by referring only to the estimated degree of deterioration.

[0059] (Method for Predicting Panel In-Plane Temperature) Next, a method for predicting the panel in-plane temperature of the display panel 20 by the prediction unit 112 using at least a part of the image data and the reference temperature data will be described.

[0060] The in-plane temperature of the display panel 20 can be predicted using at least a portion of the image data of the image displayed on the display panel 20 and reference temperature data acquired by one or more temperature sensors 30 arranged at predetermined positions. Specifically, relationship information between the image displayed on the display panel, the in-plane temperature when the image is displayed, and the reference temperature is acquired in advance. To predict the in-plane temperature of the display panel, the image displayed on the display panel and the measured reference temperature are compared with the previously acquired relationship information. The in-plane temperature of the display panel 20 may also be predicted using a machine learning model that uses at least a portion of the values ​​extracted from the image data of the image displayed on the display panel 20 and the reference temperature data acquired by the temperature sensors 30 arranged at predetermined positions as explanatory variables, and the predicted in-plane temperature of the display panel 20 as a target variable. A prediction method using a machine learning model will be described later.

[0061] Each step will be described in detail below. In step 1, a predetermined type of image is displayed on an arbitrary display panel. In parallel with the display of the image, a temperature (reference temperature) at a predetermined position is acquired. This temperature may be acquired over time. The reference temperature may be the temperature at a predetermined position on the display panel, or may be the room temperature or outdoor air temperature at a predetermined position around the display panel. Furthermore, in parallel with the display of the image, the panel surface temperature of the display panel is acquired. The panel surface temperature may be acquired over time. The panel surface temperature may be the temperature of the surface of the display panel. The surface temperature of the display panel can be measured by capturing an image of the surface of the display panel with, for example, a thermographic camera. The reference temperature and the panel surface temperature may each be measured until they are saturated.

[0062] The predetermined type of image may be either a still image or a moving image. In either case, still image data or moving image data is acquired using multiple patterns. In the case of a still image, for example, the image may be a uniform image of a certain color, an image with a certain color gradation, an image divided into multiple areas like a chessboard with various colors arranged in each area, an image with a specific design, or an input image for work. In the case of a moving image, data is acquired for each type of moving image to be displayed. The type of moving image may be, for example, news, movies, animation, drama, etc.

[0063] Next, in step 2, a regression equation is found between the image data, the reference temperature data, and the temperature within the panel surface over time. A plurality of regression equations may be set in response to various conditions. The type and number of regression equations are not limited.

[0064] The above steps 1 and 2 are performed before predicting the panel in-plane temperature. Then, in step 3, a reference temperature when a certain type of image is displayed on a certain display panel is obtained, and the image data and the reference temperature are applied to the relational expression obtained in step 2, and the derived panel in-plane temperature is used as the predicted panel in-plane temperature for that display panel. The temperature sensor for measuring the reference temperature on a certain display panel may be placed in the same position as the sensor that obtained the reference temperature in step 1.

[0065] Furthermore, when it is desired to predict the in-plane temperature of a display device of a specific model (structure), a regression equation between the in-plane temperature of the panel and image data and reference temperature data acquired using a display device of the same model as the display panel may be used. By using a regression equation between the in-plane temperature of the panel and image data and reference temperature data acquired using a display device of the same model as the display device to be predicted, it is possible to more accurately predict the in-plane temperature of the panel for that model of display device.

[0066] [Embodiment 2] (Prediction of In-Plane Temperature Using Machine Learning Model) Next, as Embodiment 2 of the present disclosure, a self-luminous display device that predicts an in-plane temperature of a panel using a machine learning model will be described with reference to the drawings. Fig. 5 is a block diagram showing the configuration of a self-luminous display device 1A according to Embodiment 2. The self-luminous display device 1A according to this embodiment differs from the self-luminous display device 1 of Embodiment 1 in that a prediction unit 112A is different.

[0067] As shown in the figure, the self-luminous display device 1A includes a control device 10A, a self-luminous display panel 20, a temperature sensor 30, and a communication unit 40. The self-luminous display panel 20 and the temperature sensor 30 have the same functions as the self-luminous display panel 20 and the temperature sensor 30 described in embodiment 1, and therefore a description thereof will be omitted here. Only the parts that differ from the configuration of the self-luminous display device 1 described in embodiment 1 will be described below.

[0068] The control device 10A includes a main control unit 11A and a memory 12. The main control unit 11A includes an acquisition unit 111, a prediction unit 112A, a display correction unit 114, and a display panel drive unit 116. The main control unit 11A may further include a deterioration estimation unit 113 and a display time measurement unit 115. The prediction unit 112A may include a machine learning model 122 and a model update unit 1121. The prediction unit 112A may include the machine learning model 122, for example, if the memory 12 stores a trained machine learning model 122, and the prediction unit 112A predicts the in-plane temperature of the display panel 20 using the machine learning model 122. The machine learning model 122 predicts the in-plane temperature of the display panel 20 using at least a portion of the image data and reference temperature data. The image data, reference temperature data, and in-plane temperature are as described in the first embodiment. The memory 12 may include a display history storage unit 121. The acquisition unit 111, the display correction unit 114, the display panel drive unit 116, the deterioration estimation unit 113, and the display history storage unit 121 are the same as those described in the first embodiment, and therefore description thereof will be omitted.

[0069] 6 and 7 are schematic diagrams showing an example of explanatory variables input to the machine learning model 122 and a response variable output from the machine learning model 122. As shown in FIG. 6 , the machine learning model 122 is a trained machine learning model that uses at least a portion of the values ​​extracted from image data 71 and reference temperature data 72 acquired by a temperature sensor placed at a predetermined position as explanatory variables, and a predicted in-panel temperature 91 of the display panel 20 as a response variable. The explanatory variables are data that a user inputs to the machine learning model 122. The response variable is data that is output by the machine learning model 122 to which the explanatory variables have been input. A method for constructing (learning) the machine learning model 122 will be described later.

[0070] The value extracted from the image data as the explanatory variable may be a plurality of current values ​​or grayscale values ​​of one or more pixels or one or more regions of the display panel 20 corresponding to the image data. Furthermore, the value extracted from the image data may be a value based on a plurality of current values ​​or grayscale values ​​to which a degradation compensation amount for the pixels of the display panel 20 is added. The reference temperature data as the explanatory variable may be a temperature acquired by a temperature sensor 30 disposed at one or more locations on the display panel 20. Furthermore, the reference temperature data may be environmental temperature data acquired by a temperature sensor 30 disposed around the display panel 20. As shown in FIG. 6 , the explanatory variable of the machine learning model 122 may further include display time data for an image displayed on the display panel 20. The display time may be a continuous display time.

[0071] The objective variable output by the machine learning model 122 may be a predicted in-plane panel temperature for each of one or more pixels or one or more regions of the display panel 20 .

[0072] The machine learning model 122 may include a still image model 122A and a moving image model 122B. The still image model 122A is a machine learning model that outputs a predicted panel in-plane temperature of the display panel 20 when the image displayed on the display panel 20 is a still image. The moving image model 122B is a machine learning model that outputs a predicted panel in-plane temperature of the display panel 20 when the image displayed on the display panel 20 is a moving image. In other words, when the image displayed on the display panel 20 is a still image, the prediction unit 112A inputs the image data of the image to the still image model 122A to output a predicted panel in-plane temperature. Furthermore, when the image displayed on the display panel 20 is a moving image, the prediction unit 112A inputs the image data of the image to the moving image model 122B to output a predicted panel in-plane temperature.

[0073] The image patterns displayed in moving images are not consistent. Therefore, even models trained using many moving images may not be able to accurately predict the panel in-plane temperature depending on the trained patterns. Therefore, as shown in FIG. 7 , the moving image model 122B may be a model that uses at least a portion of values ​​extracted from image data 71 related to the displayed image and predicted panel in-plane temperature data 74 at a first time T1 as explanatory variables, and outputs a predicted panel in-plane temperature 92 at a second time (T1 + Δt) that is Δt after the first time T1 as a dependent variable. Here, Δt represents a predetermined period of time. The moving image model 122B accepts image data based on the average R, G, and B current values ​​or gradation values ​​corresponding to the average R, G, and B current values ​​for a pixel or region over Δt and the predicted panel in-plane temperature data at the first time T1 as explanatory variables, and outputs the predicted panel in-plane temperature of the pixel or region at the second time (T1 + Δt) as a dependent variable. The dynamic image model 122B repeats this process every Δt, thereby outputting the predicted panel in-plane temperature 92 for each Δt. Here, the "predicted panel in-plane temperature data at the first time T1" used initially in this process may be the result of a prediction made by the still image model 122A. For example, the still image model 122A may predict the panel in-plane temperature data at time T1 using image data and reference temperature data as explanatory variables, which are based on the average current values ​​of R, G, and B in a pixel or region during Δt up to time T1, or on the gradation values ​​corresponding to the average current values.

[0074] Such a moving image model 122B can be constructed by learning the actual measured value of the temperature within the panel surface at an arbitrary time T1, which is designated as a first time, and the actual measured value of the temperature within the panel surface at a second time (T1+Δt) when a certain type of moving image is continuously displayed. As shown in FIG. 7 , the explanatory variables of the moving image model 122B may further include display time data for which the image is displayed on the display panel 20. Furthermore, the explanatory variables of the moving image model 122B may further include reference temperature data.

[0075] The machine learning model 122 may be a model that is trained using values ​​extracted from image data of an image displayed on a self-luminous display panel, reference temperature data, and a measurement panel in-plane temperature, and that uses values ​​extracted from image data of an image displayed on a specific self-luminous display panel and the reference temperature data as explanatory variables and outputs a predicted panel in-plane temperature of the specific self-luminous display panel as a target variable. The measurement panel in-plane temperature is one of the training data of the machine learning model 122 and is actual measurement data of the panel in-plane temperature of the display panel 20 when an image is displayed. The measurement panel in-plane temperature data may be, for example, temperature data on the surface of the display panel 20 measured by a thermal camera or a temperature sensor, or internal temperature data acquired by a temperature sensor disposed inside the display panel 20. The measurement panel in-plane temperature data may be temperature data at a position corresponding to the position of a pixel, or may be a representative temperature of a region obtained by dividing the display panel 20. The representative temperature may be, for example, average temperature data of the region.

[0076] Alternatively, the machine learning model 122 may be a model that is trained using at least a portion of the values ​​extracted from image data of an image displayed on a self-luminous display panel of a predetermined type, reference temperature data, and the measured panel in-plane temperature, and that outputs the predicted panel in-plane temperature of the self-luminous display panel of that type as an explanatory variable, using at least a portion of the values ​​extracted from the image data of an image displayed on the self-luminous display panel of that type. By using the machine learning model 122 trained using a display panel of a type whose panel in-plane temperature is desired to be predicted, it is possible to output a more accurate predicted panel in-plane temperature.

[0077] The machine learning model 122 may include only one of the still image model 122A and the moving image model 122B. The algorithm used by the machine learning model 122 is not limited, and may include, for example, a multi-layer neural network that performs deep learning, a random forest, a support vector machine, or the like.

[0078] The prediction unit 112A may include a model update unit 1121. The model update unit 1121 obtains an improved machine learning model from an external database 100 via the communication unit 40 and updates the existing machine learning model 122. The database 100 stores machine learning models that have been improved by a user or a third party using new learning data. By updating the machine learning model 122 to this improved machine learning model, the user can make more accurate predictions of the panel in-plane temperature.

[0079] In order to predict the in-plane temperature of each pixel or each region using a regression equation between image data, reference temperature data, and the in-plane temperature over time, it is necessary to use a large number of regression equations as the number of pixels or regions increases. Alternatively, it is necessary to construct a very complex regression equation. Therefore, formulating such a regression equation is difficult, and the error is likely to be large. By using the machine learning model 122, it is possible to predict the in-plane temperature of the panel with higher accuracy than a method of predicting the in-plane temperature by calculating a regression equation.

[0080] In particular, a multilayer neural network that performs deep learning can learn the current value, grayscale value, or average value thereof for each pixel or region of the display panel 20 as numerical data or image data, and use these as explanatory variables. Therefore, the multilayer neural network model is a model that has a convenient and accurate algorithm for outputting a predicted panel in-plane temperature.

[0081] (Display Correction Method Using Machine Learning Model) A description will now be given of the display correction method S4 using the machine learning model 122 having the above configuration. Fig. 8 is a flowchart showing the flow of the display correction method S4 using the machine learning model 122. As shown in the figure, the display correction method S4 includes a step of predicting the temperature within the panel surface using the machine learning model.

[0082] Step S41 is an acquisition step in which at least a portion of the values ​​extracted from image data relating to the image displayed on the display panel 20 and reference temperature data acquired by a temperature sensor 30 placed at a predetermined position are acquired.

[0083] Step S42 is a prediction step that predicts the in-plane temperature of the display panel 20 using a machine learning model that uses at least a portion of the values ​​extracted from the acquired image data and reference temperature data as explanatory variables and the predicted in-plane temperature of the display panel 20 as the objective variable.

[0084] Step S43 is a correction step in which the predicted panel in-plane temperature is used to correct the display signal of the image displayed on the display panel 20. The method for correcting the display signal of the image is as described above.

[0085] By using the machine learning model 122, the following effects can be obtained. Depending on the type of image displayed on the display panel 20, the display gradation for each pixel or region differs. In other words, the amount of current varies depending on the location on the display panel surface, and the amount of heat generated by the current also differs. As a result, the local temperature within the panel surface also differs. Therefore, it is very difficult to accurately predict the temperature within the panel surface for each pixel or region. However, by performing machine learning using image data, it becomes possible to predict the temperature within the panel surface while taking into account the influence of the current, leading to more accurate temperature prediction than conventional methods.

[0086] For example, temperature gradients occur within the panel surface due to factors such as panel design and integrated circuit layout, such as higher temperatures near the TCON (timing controller). However, it is believed that this effect can also be taken into account by performing machine learning using image data.

[0087] Furthermore, it is possible to realize predictions that take into account the thermal effects of surrounding areas on the panel surface. The temperature directly above a specific pixel is affected not only by the current and heat generated by the corresponding pixel itself, but also by heat generated in surrounding areas, so these effects would be ignored if temperature predictions were made independently for each pixel. However, by including the image pattern itself as an explanatory variable, it becomes possible to take into account the distribution of displayed grayscales on the panel surface.

[0088] [Embodiment 3] (Method for constructing a machine learning model) Next, as a third embodiment of the present disclosure, a method for constructing a machine learning model 122 will be described with reference to the drawings. Fig. 9 is a flowchart showing the flow of a method S5 for constructing a machine learning model. As shown in the figure, the method S5 includes steps S51 and S52.

[0089] Step S51 is an acquisition step for acquiring multiple sets of training data, each set consisting of at least a portion of values ​​extracted from image data relating to the displayed image, temperature data within the measurement panel surface of the display panel 20 displaying the image, and reference temperature data acquired by a temperature sensor 30 positioned at a predetermined position when the image was displayed. To create a machine learning model that can be widely applied to situations expected in actual use, it is desirable to create training data that is as broad as possible within the range of image patterns and environmental temperatures expected in actual display panel use (i.e., including various cases expected in actual use). Furthermore, it is preferable to create as many sets of training data as possible. The more sets, the more accurate the machine learning model can be expected to be constructed. The temperature data within the measurement panel surface and the reference temperature data are as described above.

[0090] Step S52 is a creation step of creating a machine learning model that outputs a predicted panel in-plane temperature when at least some of the values ​​extracted from the image data and reference temperature data are input by training multiple sets of training data. The predicted panel in-plane temperature may be a predicted panel in-plane temperature for one or more pixels or one or more regions of the display panel 20.

[0091] The temperature data within the measurement panel of the learning data is temperature data within the panel when various types of images are displayed. The temperature data within the measurement panel may also be time-varying temperature data within the measurement panel corresponding to the continuous display time of the image. The temperature data within the measurement panel may also be data measured under different room or outdoor temperature conditions. FIG. 10 is a flowchart showing the flow of a method S6 for building a machine learning model using time-varying temperature data within the measurement panel corresponding to the continuous display time. As shown in the figure, step S61 is an acquisition step for acquiring multiple sets of learning data, each set including at least a portion of values ​​extracted from image data relating to the displayed image, the continuous display time of the image, time-varying temperature data within the measurement panel corresponding to the continuous display time of the image, and reference temperature data acquired by a temperature sensor 30 placed at a predetermined position when the image was displayed.

[0092] Step S62 is a creation step in which, by training multiple sets of training data, a machine learning model is created that outputs a predicted panel in-plane temperature over time according to the continuous display time when at least some of the values ​​extracted from the image data, the reference temperature data, and the continuous display time are input. By using the construction method S6, a machine learning model can be created that takes into account changes in the panel in-plane temperature over time according to the continuous display time.

[0093] Furthermore, if the machine learning model is a moving image model 122B, a different construction method S7 may be used. FIG. 11 is a flowchart showing the flow of the construction method S7 of a machine learning model for predicting the in-plane temperature of the display panel 20 on which a moving image is displayed. First, in an acquisition step S71, learning data is acquired, including temperature data on the measurement panel at a first time T1 and temperature data on the measurement panel at a second time (T1 + Δt) that is Δt after the first time T1. That is, step S71 is an acquisition step for acquiring multiple sets of learning data, each set including at least a portion of values ​​extracted from image data relating to the displayed image and temperature data on the measurement panel of the display panel on which the image is displayed, the temperature data on the measurement panel at time T1 and the temperature data on the measurement panel at time (T1 + Δt). Note that step S71 may also acquire the display time Δt (the difference between the first time T1 and the second time (T1 + Δt)) during which the image is displayed. The image data may be image data based on average current values ​​of R, G, and B for the pixel or region during Δt, or on grayscale values ​​corresponding to the average current values.

[0094] Then, in a creation step S72, a machine learning model is created so that, when at least a portion of the values ​​extracted from the image data and the measurement panel in-plane temperature data at the first time T1 are input, a predicted panel in-plane temperature at the second time (T1 + Δt) is output. In other words, step S72 is a creation step in which a machine learning model is created so that, when image data and the measurement panel in-plane temperature data at the first time T1 are input, a predicted panel in-plane temperature at the second time (T1 + Δt) is output by training multiple sets of training data. In the construction method S7, the training data may also include a continuous display time of the image. Since the continuous display time can be obtained as the elapsed time Δt from the first time T1 to the second time (T1 + Δt), the first time T1 and the second time (T1 + Δt) may be included in the training data. Furthermore, the training data may also include reference temperature data acquired by a temperature sensor located at a predetermined position. In this case, the reference temperature data may be input as an explanatory variable to the created machine learning model. This allows the machine learning model to output a more accurate predicted panel in-plane temperature. Note that when the moving image model 122B first predicts the panel in-plane temperature at time T1, the still image model (machine learning model 122) may be used. In this case, reference temperature data may be used as an explanatory variable to be input to the machine learning model 122.

[0095] Construction method S7 makes it possible to create a machine learning model that accurately predicts the temperature within the panel surface of the display panel 20 on which a moving image is displayed, even when the displayed image pattern is not constant.

[0096] (Learning Device) The machine learning model construction method S5 to S7 can be executed by a learning device 2 described below. Fig. 12 is a block diagram of the learning device 2 that executes learning of a machine learning model. The learning device 2 includes a control device 50, a memory 60, and a machine learning model 70. The machine learning model 70 is the model to be learned.

[0097] The control device 50 includes an acquisition unit 51 and a learning unit 52. The acquisition unit 51 acquires learning data from an external database 200 via a communication unit (not shown). The acquisition unit 51 records the acquired learning data in a memory 60. The learning data includes image data 61, temperature data 62 within the measurement panel surface, and reference temperature data 63. These are referred to as a learning data set. The learning data set may further include display time data 64. The display time data 64 may be continuous display time data. The contents of each of these data have already been described. The acquisition unit 51 acquires a plurality of learning data sets.

[0098] The acquisition unit 51 reads a learning data set from the memory 60 and inputs it to the learning unit 52. Specifically, for example, image data, temperature data within the measurement panel, and reference temperature data are input. The learning unit 52 then creates and constructs a machine learning model 70 so that the temperature within the measurement panel is output when the image data and reference temperature data are input as learning data. For example, the learning unit 52 updates various parameters and weights in the machine learning model 70 so that the output of the machine learning model 70 approaches the temperature within the measurement panel. By repeating this process for each learning data set, the output of the machine learning model 70 approaches the actual temperature within the measurement panel. In this manner, the learning unit 52 can construct the machine learning model 70.

[0099] [Embodiment 4] (Self-luminous display device that estimates cumulative deterioration amount) Next, as embodiment 4 of the present disclosure, a self-luminous display device 1B that estimates cumulative deterioration amount will be described in detail with reference to the drawings. Fig. 13 is a block diagram showing the configuration of the self-luminous display device 1B according to embodiment 4. The self-luminous display device 1B according to this embodiment differs from the self-luminous display device 1 of embodiment 1 or the self-luminous display device 1A of embodiment 2 in that the deterioration estimation unit 113 is different.

[0100] As shown in the figure, the self-luminous display device 1B includes a control device 10B, a self-luminous display panel (display panel) 20, a temperature sensor 30, and a communication unit 40. The display panel 20 and the temperature sensor 30 have the same functions as the display panel 20 and the temperature sensor 30 of the self-luminous display device 1 described in embodiment 1, and the communication unit 40 has the same function as the communication unit 40 described in the self-luminous display device 1A of embodiment 2, so a description thereof will be omitted here. Only the parts that differ from the configuration of the self-luminous display device 1 or 1A described in embodiment 1 or 2 will be described below.

[0101] The control device 10B of the self-luminous display device 1B includes a main control unit 11B and a memory 12B. The main control unit 11B includes an acquisition unit 111, a deterioration estimation unit 113B, a display correction unit 114, a display time measurement unit 115, a display panel drive unit 116, and an accumulated deterioration amount recording unit 117. The main control unit 11B may also include a prediction unit 112. The acquisition unit 111, the prediction unit 112, the display correction unit 114, the display time measurement unit 115, and the display panel drive unit 116 have the same functions as those of the respective units described in the first to third embodiments, and therefore description thereof will be omitted.

[0102] The memory 12B may include a display history storage unit 121, a machine learning model 123, or an accumulated deterioration amount storage unit 124. The display history storage unit 121 stores image data related to an image. In this embodiment, the image data related to an image is at least one of data representing the image itself displayed by the display panel 20 and a current value or a gradation value extracted from the image data. The data representing the image itself refers to data of the image itself, rather than a current value or a gradation value. Hereinafter, the data representing the image itself and / or the current value or the gradation value extracted from the image data will be collectively referred to as "image data." The accumulated deterioration amount storage unit 124 stores the estimated accumulated deterioration amount output by the machine learning model 123.

[0103] The degradation estimation unit 113B estimates the cumulative degradation amount of the display panel 20 using a machine learning model 123. The machine learning model 123 is stored in, for example, the memory 12B. FIG. 14 is a schematic diagram showing an example of explanatory variables input to the machine learning model 123 and a response variable output from the machine learning model. As shown in the figure, the machine learning model 123 is a trained machine learning model that uses as explanatory variables a cumulative degradation amount 65 of the display panel 20 before display, image data 61 related to an image displayed on the display panel 20, and display time data 64 (hereinafter, the display time data will also be simply referred to as "display time") during which the image is displayed on the display panel 20, and as a response variable an estimated cumulative degradation amount 80 of the display panel 20 after display. "Before display" refers to before the display panel 20 is powered on. Note that, as described below, temperature data of the display panel 20 may be further added as an explanatory variable (not shown).

[0104] The cumulative deterioration amount may be expressed, for example, based on the percentage decrease in luminance value when a predetermined gradation (e.g., V255, V127, etc.) is displayed after deterioration of the display panel 20, compared to the luminance value when displayed in the initial state. Specifically, for example, if the luminance of a predetermined gradation display in the initial state of the display panel 20 is 100 and the luminance of the gradation display after deterioration is X, the cumulative deterioration amount can be defined as (100-X)%. A method for constructing the machine learning model 123 will be described later.

[0105] The deterioration estimation unit 113B inputs the cumulative deterioration amount of the display panel 20 before display stored in the cumulative deterioration amount storage unit 124, image data related to the image displayed on the display panel 20 acquired by the acquisition unit 111, and the display time measured by the display time measurement unit 115 for displaying the image on the display panel 20 into the machine learning model 123, and acquires an estimated cumulative deterioration amount of the display panel 20 after display output by the machine learning model 123. The acquired estimated cumulative deterioration amount of the display panel 20 after display is recorded by the cumulative deterioration amount recording unit 117 in the cumulative deterioration amount storage unit 124 of the memory 12B. The memory 12B may further include a display history storage unit 121. The display history storage unit 121 may store image data related to the image displayed on the display panel 20.

[0106] In this embodiment, the display time is not limited to a continuous display time. The deterioration estimation unit 113B may divide the time during which an image is continuously displayed into multiple segments and estimate the cumulative deterioration amount for each segmented display time. For example, suppose a continuously displayed image is divided into n segments at arbitrary intervals along the way, and the total display time until the power is turned off is n segments, with the segments being x1 hours, x2 hours, ..., xn hours. Each segmented time may be different. In this case, the deterioration estimation unit 113B acquires the cumulative deterioration amount of the display panel 20 up until the image corresponding to the segmented time is displayed, image data of the image corresponding to the segmented time into x1 hours, and the data for that segmented time, x1 hours, and inputs them into the machine learning model 123. The output estimated cumulative deterioration amount is then updated and recorded. The "image data of the image corresponding to the segmented time into x1 hours" may be, for example, image data based on the average current values ​​of R, G, and B in a pixel or region during x1 hours, or on the gradation values ​​corresponding to the average current values. Next, the deterioration estimation unit 113B acquires the recorded estimated cumulative deterioration amount, image data of images corresponding to the times divided into x2 hours, and x2 hours which is the data for the divided times, and inputs these to the machine learning model 123, and acquires and updates the output estimated cumulative deterioration amount. In this way, the deterioration estimation unit 113B may sequentially acquire and update the cumulative deterioration amount for each divided display time.

[0107] This method of dividing and estimating the amount of accumulated degradation can be applied to cases where the displayed image changes over time, such as a moving image. For example, the amount of accumulated degradation can be estimated and updated by dividing the image into scenes where the displayed content is somewhat similar or scenes where the displayed content does not change significantly. This improves the accuracy of estimating the amount of accumulated degradation when displaying a moving image.

[0108] Furthermore, the deterioration estimation unit 113B may include a model update unit 1131. The model update unit 1131 updates the machine learning model 123 by training the machine learning model 123 using new training data acquired from, for example, the database 100. Alternatively, the model update unit 1131 may acquire a new machine learning model 123 from the database 100 and replace the machine learning model 123 stored in the memory 12B.

[0109] The content displayed on the display panel 20 is not specified. Many factors, including the display content, are thought to be involved in the deterioration of the display panel 20, but it is not easy for a user to devise a specific method for estimating the degree of deterioration while taking these many factors into consideration. The machine learning model 123 is a pre-trained machine learning model that estimates the degree of cumulative deterioration of the display panel 20 when various types of images are displayed on the display panel 20 under various conditions. By using such a machine learning model 123, it is possible to mechanically estimate the cumulative deterioration amount over time of the display panel 20, whose display content is not specified.

[0110] (Method for Estimating the Accumulative Deterioration Amount) Next, a method for estimating the cumulative deterioration amount of the display panel 20 using the machine learning model 123 will be described with reference to the drawings. Fig. 15 is a flowchart showing the flow of the method S8 for estimating the cumulative deterioration amount of the display panel 20 using the machine learning model 123. As shown in the figure, the estimation method S8 includes steps S81 to S83.

[0111] Step SS81 is a step in which at least one processor acquires the cumulative deterioration amount of the display panel 20 before display, image data related to the image displayed on the display panel 20, and display time data for displaying the image on the self-luminous display panel. The cumulative deterioration amount of the display panel 20 before display can be acquired by, for example, the degradation estimation unit 113B from information stored in the cumulative deterioration amount storage unit 124. The image data may be acquired by, for example, the degradation estimation unit 113B from information acquired by the acquisition unit 111 and recorded in the display history storage unit 121. The display time may be acquired by, for example, the degradation estimation unit 113B from information acquired by the display time measurement unit 115 and recorded in the display history storage unit 121.

[0112] Step S82 is a step in which at least one processor inputs the acquired cumulative deterioration amount, image data, and display time data into the machine learning model 123. Step S82 may be executed by, for example, the deterioration estimation unit 113B.

[0113] In step S83, at least one processor acquires the estimated cumulative deterioration amount output by the machine learning model. Step S83 may be executed by, for example, the deterioration estimation unit 113B. The acquired estimated cumulative deterioration amount is overwritten and recorded in the cumulative deterioration amount storage unit 124 by the cumulative deterioration amount recording unit 117, and the recorded cumulative deterioration amount is used for estimating the cumulative deterioration amount for the next display.

[0114] The image data may be at least a portion of the current value or gradation value for each pixel. In this case, the cumulative deterioration amount stored in the cumulative deterioration amount storage unit 124 and the estimated cumulative deterioration amount output by the machine learning model 123 are expressed as the cumulative deterioration amount and the estimated cumulative deterioration amount for each pixel. The machine learning model 123 estimates the cumulative deterioration amount of the display panel 20 using at least a portion of the current value or gradation value for each pixel of the display panel 20 based on the displayed image.

[0115] Alternatively, the image data may be at least a portion of the current value or gradation value for each region including multiple pixels. In this case, the cumulative deterioration amount stored in the cumulative deterioration amount storage unit 124 and the estimated cumulative deterioration amount output by the machine learning model 123 are expressed as the cumulative deterioration amount and estimated cumulative deterioration amount for each region including multiple pixels. The machine learning model 123 estimates the cumulative deterioration amount of the display panel 20 using at least a portion of the current value or gradation value for each region including multiple pixels based on the displayed image. The size of the region including multiple pixels is not limited. It may be an arbitrary number of regions obtained by dividing the display panel 20.

[0116] Alternatively, the image data may be data representing the displayed image itself. In this case, the accumulated deterioration amount stored in the accumulated deterioration amount storage unit 124 and the estimated accumulated deterioration amount output by the machine learning model 123 are expressed as panel in-plane distribution data of the accumulated deterioration amount and panel in-plane distribution data of the estimated accumulated deterioration amount. The machine learning model 123 estimates the accumulated deterioration amount of the display panel 20 using data representing the displayed image itself. In this case, the machine learning model 123 may be a multilayer neural network model suitable for image processing. The image data may be a part of, or the entire, displayed image itself.

[0117] As described above, the machine learning model 123 is a machine learning model that uses as explanatory variables the cumulative deterioration amount of the display panel 20 before display, image data related to the image displayed on the display panel 20, and the display time for which the image is displayed on the display panel 20. However, the machine learning model 123 may also be a machine learning model that uses temperature data of the display panel 20 as an explanatory variable. In that case, the method S8 for estimating the cumulative deterioration amount described above further inputs temperature data of the display panel 20 in input step S82.

[0118] The temperature data of the display panel 20 can be temperature data obtained by various methods. For example, the temperature data may be an actual measured temperature or a predicted temperature. As described in the above embodiments, the actual measured temperature data may be reference temperature data actually measured by a temperature sensor placed at a predetermined position. Alternatively, the actual measured temperature data may be environmental temperature data actually measured by an environmental temperature sensor placed at a predetermined position. Alternatively, the predicted temperature of the display panel 20 may be predicted in-plane temperature data of the self-luminous display panel predicted by the prediction unit 112.

[0119] The temperature data may be the above-mentioned temperature data, and may be average temperature data during the period when the display panel 20 displayed the temperature, or temperature data after the display panel 20 ended displaying the temperature. That is, the temperature data may be the average temperature of the actually measured temperatures during the period when the display panel 20 displayed the temperature, or the actually measured temperature after the display panel 20 ended displaying the temperature. Alternatively, the temperature data may be the average temperature of the predicted temperatures during the period when the display panel 20 displayed the temperature, or the predicted temperature after the display panel 20 ended displaying the temperature. The temperature after the display ended may be the temperature immediately after the power to the display panel 20 was turned off.

[0120] (Dividing Display Time) The above-described acquisition step S81 may be performed for each divided display time by dividing the time during which images are continuously displayed into multiple parts. That is, the step may acquire the accumulated deterioration amount of the self-luminous display panel up until the image corresponding to the divided time is displayed, image data of the image corresponding to the divided time, and data for the divided time. In the subsequent input step S82, the acquired information is input to the machine learning model 123. In the next acquisition step S83, the estimated accumulated deterioration amount output by the machine learning model 123 is acquired.

[0121] By executing this acquisition step S81, it is possible to divide a continuous display time (image) into several parts and estimate the cumulative deterioration amount for each divided display time. This method of dividing and estimating the cumulative deterioration amount can be applied when a moving image in which the display image changes over time is displayed on the display panel 20. Alternatively, it may be applied to a display panel 20 that displays a still image whose display content is changed periodically.

[0122] (Image data with degradation compensation added) When a current value or a gradation value is used as image data, the current value or gradation value may be a value based on a current value or a gradation value to which a degradation compensation amount for the pixels of the display panel 20 has been added. Some self-luminous display devices have a function of supplying a current value (gradation value) to the display panel 20 with a degradation compensation amount added, taking into account the degradation of the display panel 20. In such cases, the cumulative degradation amount can be estimated more accurately by estimating it with consideration of the current value (gradation value) supplied to the display panel 20.

[0123] (Estimation of cumulative deterioration amount for each RGB pixel) The display panel 20 is generally composed of three types of pixels: R, G, and B. In such a case, the estimated cumulative deterioration amount may be estimated for each of the three types of pixels: R, G, and B of the display panel 20. In other words, the machine learning model 123 may estimate the cumulative deterioration amount for each of the three types of pixels: R, G, and B. Because the deterioration characteristics of each of R, G, and B are generally different, such a configuration enables more accurate estimation.

[0124] According to the method for estimating the cumulative deterioration amount using the machine learning model 123 described above, it is possible to mechanically estimate the cumulative deterioration amount over time of the display panel 20 whose display content is not specified.

[0125] (Learning device for constructing machine learning model 123) Next, a learning device for constructing machine learning model 123 will be described with reference to the drawings. Fig. 16 is a block diagram showing the configuration of a learning device 2A for constructing machine learning model 123. As shown in the figure, learning device 2A includes a control device 50A and a memory 60A.

[0126] The control device 50A includes an acquisition unit 51A and a learning unit 52A. The acquisition unit 51A acquires learning data from an external database 200A via a communication unit (not shown). The acquisition unit 51A records the acquired learning data in a memory 60A. The learning data includes a cumulative deterioration amount 65 before display, image data 61, display time data 64, and a cumulative deterioration amount 66 after display. These are referred to as a learning data set. The learning data set may further include temperature data (not shown). The display time data 64 is a continuous display time, but it does not have to be the time from power-on to power-off, but may be a continuous display time obtained by dividing that time. The content of each of these data is as described above. The acquisition unit 51A acquires multiple learning data sets. Temperature data may also be added to the learning data. By adding temperature data and performing learning, a more accurate machine learning model 123 can be created.

[0127] The learning unit 52A reads the learning data set from the memory 60B and performs learning. For example, the learning unit 52A constructs the machine learning model 70A so that when the pre-display cumulative degradation amount 65, image data 61, and display time data 64 are input as explanatory variables to the machine learning model 70A, the post-display cumulative degradation amount 66 is output as the objective variable (estimated post-display cumulative degradation amount). Specifically, the learning unit 52A updates various parameters and weights in the machine learning model 70A so that the output of the machine learning model 70A approaches the learning data (post-display cumulative degradation amount 66). This process is repeated each time the learning data set is added or changed, so that the output of the machine learning model 70A approaches the learning data cumulative degradation amount 66. In this manner, the learning unit 52A can create and construct a trained machine learning model 70A (machine learning model 123).

[0128] 17 is an example of a flowchart showing the flow of a method S9 for constructing the machine learning model 123. As shown in the figure, the construction method S9 includes steps S91 and S92.

[0129] Step S91 is a step in which at least one processor acquires multiple sets of learning data, each set including cumulative deterioration amount 65 before display on display panel 20, image data 61 related to the image displayed on display panel 20, display time data 64 during which the image was displayed on display panel 20, and cumulative deterioration amount 66 after display on display panel 20. Temperature data may also be added to the learning data.

[0130] Step S92 is a step in which at least one processor learns a plurality of sets of learning data to create a machine learning model that outputs an estimated cumulative deterioration amount of the display panel 20 after display when inputting the cumulative deterioration amount before display on the display panel 20, image data related to the image displayed on the display panel 20, and display time data for the image displayed on the display panel 20. By the above-described construction method S9, it is possible to construct a machine learning model 123 that estimates the cumulative deterioration amount of the display panel 20.

[0131] (Creating Learning Data) The learning data to be acquired can be created, for example, as follows. First, the cumulative deterioration amount in the pre-display state of the self-luminous display panel is measured to obtain the cumulative deterioration amount before display. Next, the panel is turned on for a certain period of time under predetermined temperature and display gradation conditions (current conditions), and the cumulative deterioration amount is measured to obtain the cumulative deterioration amount after display. This series of operations is repeated while changing each parameter. Note that it is desirable to change each parameter so as to roughly cover the range expected in actual panel use. This results in, for example, the learning data set shown in FIG. 18. Temperature data is not necessarily required. Note that the cumulative deterioration amount can be determined using luminance values, as described above. Therefore, the cumulative deterioration amount can be determined by performing a luminance evaluation in advance in the panel's initial state and measuring the luminance before and after lighting.

[0132] The luminance evaluation for calculating the cumulative deterioration amount may be performed for each pixel. Alternatively, the luminance evaluation may be substituted with the average luminance of a specific region (e.g., a circle with a diameter of 1 cm) including multiple pixels. In this case, however, it is preferable that the lighting gradation and lighting time are the same for the entire specific region. Alternatively, the display screen of the display panel 20 may be divided into multiple regions, and the average luminance of each region may be used.

[0133] The user may manually execute the above learning data acquisition steps to create the learning data. Alternatively, the user may use a learning data creation program to automatically execute the above steps to create the learning data. Using the program, it is possible to automatically execute the following operations: display a pre-stored image on any self-luminous display panel for a predetermined period of time, then display an image with a predetermined gradation value, acquire its luminance value, and calculate the cumulative deterioration amount. By repeating these operations, the learning data can be created.

[0134] (Display Correction Method) Next, a display correction method for correcting a display image using the estimated cumulative degradation amount will be described with reference to the drawings. Fig. 19 is an example of a flowchart showing the flow of a display correction method S10 using a machine learning model. As shown in the figure, the display correction method S10 includes steps S101 and S102.

[0135] Step S101 is a step in which at least one processor estimates the cumulative deterioration amount of the display panel 20 using a machine learning model in which the cumulative deterioration amount before display on the display panel 20, image data related to the image displayed on the display panel 20, and the display time for which the image is displayed on the display panel 20 are used as explanatory variables, and the estimated cumulative deterioration amount after display on the display panel 20 is used as the objective variable.

[0136] Step S102 is a step in which at least one processor corrects a display signal of an image displayed on the display panel 20 using the estimated cumulative deterioration amount.

[0137] The correction step of step S102 may be a step of correcting a display signal of an image displayed on the display panel 20 by further referring to the temperature within the panel surface of the display panel 20.

[0138] As described above, according to the method of correcting the display of an image using the cumulative deterioration amount estimated using the machine learning model 123, the cumulative deterioration amount is estimated and corrected after comprehensively inputting factors that affect the deterioration, thereby enabling more accurate display correction of the image.

[0139] [Example of implementation using software] The functions of the control devices 10, 10A, 10B and the learning devices 2, 2A (hereinafter referred to as "devices") can be realized by a program that causes a computer to function as the device, and a program that causes a computer to function as each control block of the device (particularly the main control units 11, 11A, 11B of the control devices 10, 10A, 10B and each unit included in the control devices 50, 50A of the learning devices 2, 2A).

[0140] In this case, the device includes a computer having at least one control device (e.g., a processor) and at least one storage device (e.g., a memory) as hardware for executing the program. The functions described in each of the above embodiments are realized by executing the program using the control device and storage device.

[0141] The program may be non-transitory and may be recorded on one or more computer-readable recording media. The recording media may or may not be included in the device. In the latter case, the program may be supplied to the device via any wired or wireless transmission medium.

[0142] In addition, some or all of the functions of each of the control blocks can be realized by logic circuits. For example, integrated circuits in which logic circuits that function as each of the control blocks are formed are also included in the scope of the present disclosure. In addition, the functions of each of the control blocks can also be realized by, for example, a quantum computer.

[0143] Furthermore, each process described in each of the above embodiments may be automatically executed. For example, the process may be executed by AI (Artificial Intelligence). In this case, the AI ​​may run on the control device or on another device (for example, an edge computer or a cloud server).

[0144] The present disclosure is not limited to the above-described embodiments, and various modifications are possible within the scope of the claims. Embodiments obtained by appropriately combining the technical means disclosed in different embodiments are also included in the technical scope of the present disclosure. Furthermore, new technical features can be formed by combining the technical means disclosed in each embodiment.

[0145] 1, 1A, 1B Self-luminous display device 2, 2A Learning device 10, 10A, 10B, 50, 50A Control device 11, 11A, 11B Main control unit 111, 51, 51A Acquisition unit 112, 112A Prediction unit 122, 123, 70, 70A Machine learning model 122A Still image model 122B Moving image model 1121, 1131 Model update unit 113, 113B Deterioration estimation unit 114 Display correction unit 115 Display time measurement unit 116 Display panel drive unit 117 Accumulated deterioration amount recording unit 12, 12B, 60, 60A Memory 121 Display history storage unit 124 Accumulated deterioration amount storage unit 20 Self-luminous display panel 30 Temperature sensor 40 Communication unit 52, 52A Learning unit 61, 71 Image data 62 Temperature data inside the measurement panel 63, 72 Reference temperature data 64, 73 Display time data 65 Accumulated deterioration amount before display 66 Accumulated deterioration amount after display 74 Predicted temperature data inside the panel at time T1 80 Estimated cumulative deterioration amount 91 Predicted temperature inside the panel 92 Predicted temperature inside the panel at time (T1 + Δt) 100, 200, 200A Database

Claims

1. A self-luminous display device comprising: an acquisition unit that acquires at least a portion of a value extracted from image data relating to an image displayed on a self-luminous display panel and reference temperature data acquired by a temperature sensor arranged at a predetermined position; and a prediction unit that predicts a temperature within the panel surface of the self-luminous display panel using at least a portion of the value extracted from the acquired image data and the reference temperature data.

2. The self-luminous display device according to claim 1, further comprising: a degradation estimation unit that estimates a degree of degradation of the self-luminous display panel by referring to the panel in-plane temperature predicted by the prediction unit; and a display correction unit that corrects the image to be displayed by referring to the degree of degradation estimated by the degradation estimation unit.

3. The self-luminous display device according to claim 2, further comprising a display time measurement unit that measures a display time during which the image is displayed on the self-luminous display panel, and the acquisition unit acquires the display time.

4. A self-luminous display device as described in claim 3, further comprising a display history memory unit in which the display history of the image for each of one or more pixels or one or more regions of the self-luminous display panel is recorded, and the deterioration estimation unit estimates the degree of deterioration of the self-luminous display panel by referring to the recorded display history of the image.

5. The self-luminous display device according to claim 4, wherein the display history of the image is a display history in which an average current value or average gradation value for one or more pixels or one or more regions of the self-luminous display panel, the predicted temperature within the panel surface, and the display time under the relevant conditions are associated with each other, and the deterioration estimation unit estimates the degree of deterioration of the self-luminous display panel by referring to the recorded average current value or average gradation value, the predicted temperature within the panel surface, and the display time under the relevant conditions.

6. The self-luminous display device according to claim 5, wherein the deterioration estimation unit has a display history update function for updating the data stored in the display history memory unit by accumulating new data, and estimates the degree of deterioration by referring to the accumulated data of the average current value or average gradation value, the temperature within the panel surface, and the display time.

7. A self-luminous display device as described in claim 4 or 5, wherein the deterioration estimation unit has a display history update function that accumulates new data and updates the data stored in the display history memory unit, calculates the degree of deterioration for a predetermined period each time a display is performed on the self-luminous display panel, and adds the calculated degree of deterioration to the accumulated amount of deterioration recorded in the display history memory unit.

8. The self-luminous display device according to claim 2, wherein the display correction section corrects the image to be displayed by referring to the panel in-plane temperature predicted by the prediction section.

9. A self-luminous display device according to any one of claims 1 to 8, wherein the values ​​extracted from the image data are a plurality of current values ​​or gradation values ​​of one or more pixels or one or more regions of the self-luminous display panel.

10. The self-luminous display device according to claim 9, wherein the values ​​extracted from the image data are the plurality of current values ​​or gradation values ​​to which a deterioration compensation amount for a pixel of the self-luminous display panel is added.

11. A self-luminous display device according to any one of claims 1 to 10, wherein the reference temperature data is temperature data acquired by a temperature sensor arranged at one or more locations on the self-luminous display panel.

12. A self-luminous display device according to any one of claims 1 to 10, wherein the reference temperature data is environmental temperature data acquired by the temperature sensor disposed around the self-luminous display panel.

13. A self-luminous display device as described in any one of claims 1 to 12, wherein the prediction unit is provided with a machine learning model that uses at least a portion of the values ​​extracted from the image data and the reference temperature data acquired by the temperature sensor arranged at a predetermined position as explanatory variables, and the predicted in-plane temperature of the self-luminous display panel as an objective variable.

14. The self-luminous display device according to claim 13, wherein the objective variable is the predicted in-plane panel temperature for one or more pixels or one or more regions.

15. The self-luminous display device according to claim 13 or 14, wherein the prediction unit includes a model update unit that updates the machine learning model.

16. A self-luminous display device described in any one of claims 13 to 15, wherein the machine learning model is a model that uses at least a portion of the values ​​extracted from the image data and the predicted panel in-plane temperature at a first time as explanatory variables, and outputs the predicted panel in-plane temperature at a second time that is a period Δt that has elapsed from the first time as a dependent variable.

17. A self-luminous display device described in any one of claims 13 to 16, wherein the machine learning model is a still image model in which still image data is the image data, or a moving image model in which moving image data is the image data.

18. A self-luminous display device as described in any one of claims 13 to 17, wherein the machine learning model is trained using at least a portion of the values ​​extracted from the image data of an image displayed on a self-luminous display panel of a predetermined type, the reference temperature data, and the measured panel in-plane temperature, and is a model that uses at least a portion of the values ​​extracted from the image data of an image displayed on a self-luminous display panel of the said type and the reference temperature data as explanatory variables, and outputs a predicted panel in-plane temperature of the self-luminous display panel of the said type as a target variable.

19. The self-luminous display device according to any one of claims 13 to 18, wherein the machine learning model is a multi-layer neural network model.

20. The self-luminous display device according to any one of claims 1 to 19, wherein the self-luminous display panel is a panel including any one of organic light-emitting diodes, quantum dot light-emitting diodes, and micro light-emitting diodes.

21. A self-luminous display device as described in any one of claims 1 to 12, wherein the prediction unit is provided with a machine learning model that uses at least a portion of the values ​​extracted from the image data and the predicted panel in-plane temperature at a first time as explanatory variables, and uses the predicted panel in-plane temperature at a second time that is Δt after the first time as a dependent variable.

22. A self-luminous display device comprising: an acquisition unit that acquires at least a portion of a value extracted from image data relating to an image displayed on a self-luminous display panel; and a prediction unit that predicts a panel in-plane temperature of the self-luminous display panel using at least a portion of the value extracted from the acquired image data, wherein the prediction unit predicts the panel in-plane temperature of the self-luminous display panel using a machine learning model that uses at least a portion of the value extracted from the image data and the predicted panel in-plane temperature at a first time as explanatory variables, and outputs a predicted panel in-plane temperature at a second time that is Δt after the first time as an objective variable.

23. A display correction method comprising: an acquisition step of acquiring at least a portion of a value extracted from image data relating to an image displayed on a self-luminous display panel and reference temperature data acquired by a temperature sensor arranged at a predetermined position; a prediction step of predicting an in-plane temperature of the self-luminous display panel using at least a portion of the value extracted from the acquired image data and the reference temperature data; and a correction step of correcting a display signal of the image displayed on the self-luminous display panel using the predicted in-plane temperature.

24. A display correction method according to claim 23, further comprising a degradation estimation step of estimating a degree of degradation of the self-luminous display panel, wherein in the acquisition step, a display time during which the image was displayed on the self-luminous display panel is further acquired, in the degradation estimation step, a degree of degradation of the self-luminous display panel is estimated using at least a part of a value extracted from the image data, the temperature within the panel surface, and the display time, and in the correction step, a display signal of the image is corrected by further referring to the degree of degradation.

25. A display correction method as described in claim 24, wherein in the acquisition step, an average current value or average gray scale value for one or more pixels or one or more regions of the self-luminous display panel is acquired as a value extracted from the image data, and cumulative data of the display time under the conditions is acquired; in the deterioration estimation step, a degree of deterioration is estimated using the cumulative data; and in the correction step, a display signal of the image is corrected by further referring to the degree of deterioration.

26. The display correction method according to claim 24, wherein in the degradation estimation step, the degree of degradation for a predetermined period of time is calculated each time a display is performed on the self-luminous display panel, and the calculated degree of degradation is added to an accumulated amount of degradation.

27. A display correction method according to any one of claims 23 to 26, wherein in the prediction step, the temperature within the panel surface is predicted using a trained machine learning model.

28. A display correction method according to any one of claims 23 to 27, wherein the self-luminous display panel is a panel including any one of organic light-emitting diodes, quantum dot light-emitting diodes, and micro light-emitting diodes.

29. A display correction method comprising: an acquisition step of acquiring at least a portion of a value extracted from image data relating to an image displayed on a self-luminous display panel; a prediction step of predicting an in-plane temperature of the self-luminous display panel using at least a portion of the value extracted from the acquired image data; and a correction step of correcting a display signal of the image displayed on the self-luminous display panel using the predicted in-plane temperature, wherein the prediction step is a step of predicting the in-plane temperature of the panel using a machine learning model that uses at least a portion of the value extracted from the image data and the predicted in-plane temperature of the panel at a first time as explanatory variables, and outputs a predicted in-plane temperature of the panel at a second time that is Δt after the first time as a dependent variable.

Images