Digital photo frame display intelligent adjusting system based on image content identification

Abstract

The invention discloses a digital photo frame display intelligent adjusting system based on image content recognition, and relates to the technical field of digital photo frames. The system comprises a panel characteristic twin modeling module which is used for reversely deducing dynamic characteristic parameters of a digital photo frame display panel through comparative analysis of a standard test image and an actual display image and generating a panel characteristic twin model consistent with the real-time state of the display panel; the pixel demand identification module is used for processing a target image currently displayed by the digital photo frame and generating pixel demand matrixes in one-to-one correspondence with the pixel matrixes of the display panel; and the optimization parameter generation module is used for constructing a multi-target optimization function based on the panel characteristic twinborn model and the pixel demand matrix to calculate a target adjustment parameter of each pixel of the display panel. The problems of adjustment failure, panel damage and the like in the prior art are solved, and display experience, hardware protection and economical efficiency are considered.

Description

TECHNICAL FIELD

[0001] The present application relates to the technical field of digital photo frame intelligent control, in particular to a digital photo frame display intelligent adjustment system based on image content recognition. BACKGROUND

[0002] With the development of digital photo frames towards intelligence, image content recognition technology gradually combines with display adjustment functions. The industry trend is to optimize display parameters such as brightness and color temperature by recognizing scene features such as text, portraits and night scenes in images to improve visual experience and adapt to diversified use requirements, which becomes an important direction for the technical upgrading of digital photo frames.

[0003] Current existing technologies for implementing display adjustment rely on fixed hardware parameters pre-stored by panel manufacturers or additional hardware sensors to collect panel states, and then combine pre-set databases to match adjustment parameters. However, display panels may exhibit dynamic aging phenomena such as pixel attenuation and Gamma curve drift during use, and changes in environmental temperature may further change panel characteristics. The existing technologies cannot accurately perceive these dynamic changes in real time, resulting in a disconnection between adjustment parameters and actual panel states, which may either cause display adjustment failure, such as unclear presentation of dark details, or cause hardware damage, such as static pixel burn-in of OLED panels, due to parameters exceeding the tolerance range of the panel. This problem directly restricts the practicality of intelligent adjustment of digital photo frames and the service life of the panel. SUMMARY

[0004] The present application aims to provide a digital photo frame display intelligent adjustment system based on image content recognition to solve the problems raised in the background.

[0005] To achieve the above-mentioned purpose, the present application provides the following technical solution: a digital photo frame display intelligent adjustment system based on image content recognition, comprising: a panel characteristic twin modeling module for backstepping dynamic characteristic parameters of a digital photo frame display panel through comparative analysis of standard test images and actual display images to generate a panel characteristic twin model consistent with the real-time state of the display panel; the dynamic characteristic parameters include pixel attenuation data, Gamma curve parameters and temperature correlation coefficients; a pixel demand identification module for processing a target image currently displayed by the digital photo frame to generate a pixel demand matrix corresponding one-to-one to the pixel matrix of the display panel, the pixel demand matrix containing display demand information and static pixel markers for each pixel; an optimization parameter generation module for constructing a multi-objective optimization function based on the panel characteristic twin model and the pixel demand matrix to calculate target adjustment parameters for each pixel of the display panel, and additionally generating a static pixel fine-tuning rule if the display panel is an OLED panel, and finally outputting a pixel-level adjustment parameter matrix. The execution feedback module is used to convert the pixel-level adjustment parameter matrix into display control commands and execute the adjustment. The adjustment accuracy is verified by the feedback image. If the deviation exceeds the limit, the parameters are adjusted and optimized and readjusted. At the same time, the panel characteristic twin model is updated periodically to form a closed-loop adjustment.

[0006] Preferably, the specific steps of the panel characteristic twin modeling module to realize dynamic characteristic parameter back-inference and twin model generation include: Input a preset standard test image set into the digital photo frame and acquire the corresponding actual display image. The preset standard test image set includes at least grayscale images and solid color images. Feature points are extracted from the grayscale image and its actual display image using a feature matching algorithm. The brightness deviation rate of the feature points is calculated to determine the pixel attenuation rate, and a pixel attenuation matrix of the display panel is generated. The pixel attenuation matrix is ​​the carrier of the pixel attenuation data. Key gray levels are extracted from the actual displayed image of the grayscale map and substituted into the Gamma curve formula to fit the real-time Gamma value. The real-time Gamma value is the core indicator of the Gamma curve parameters. The Gamma curve formula is: in, This refers to the actual brightness value of the key gray level. The actual brightness is 255 gray levels. This refers to the grayscale value of the current key grayscale level. =255, This is the real-time Gamma value; By comparing the standard color temperature of the solid color image with the actual color temperature of the actual displayed image, the temperature range of the display panel is determined and the temperature correlation coefficient is generated, and a temperature-color temperature drift mapping table is established. By integrating the pixel attenuation matrix, real-time Gamma value, and temperature correlation coefficient, a panel characteristic twin model is generated, and the deviation between the panel characteristic twin model and the actual state of the display panel is controlled within a preset deviation range.

[0007] Preferably, in the panel characteristic twin modeling module, the selected feature points are distributed in a uniform grid pattern in the grayscale image; the formula for calculating the pixel attenuation rate is: in, Pixel attenuation rate, The theoretical brightness of the corresponding feature point in the grayscale image. The measured brightness of the corresponding feature point in the actual displayed image; the grid density of the pixel attenuation matrix is ​​adapted to the pixel resolution of the display panel.

[0008] Preferably, in the panel characteristic twin modeling module, the extracted key gray levels include 0, 32, 64, 96, 128, 160, 192, 224, and 255; the least squares method is used when fitting the real-time Gamma value, by adjusting... The value represents the actual brightness value for all key gray levels. The goal is to minimize the sum of squared errors between the calculated value and the Gamma curve formula; simultaneously, the brightness deviation value for each key gray level is output, where the brightness deviation value is the actual brightness value of the key gray level. The difference between the value calculated using the Gamma curve formula.

[0009] Preferably, in the panel characteristic twin modeling module, for an LCD display panel, if the actual color temperature is lower than the standard color temperature and the deviation is ≥150K, then the temperature range is determined to be a low-temperature range, and the corresponding temperature correlation coefficient is... ≤0.95; If the actual color temperature is higher than the standard color temperature and the deviation is ≥100K, then the temperature range is determined to be a high-temperature range, and the corresponding temperature correlation coefficient is... ≥1.05; For OLED display panels, if the actual color temperature is higher than the standard color temperature and the deviation is ≥200K, the temperature range is determined to be a high-temperature range, and the corresponding temperature correlation coefficient is... ≥1.05; If the actual color temperature is lower than the standard color temperature and the deviation is ≥150K, then the temperature range is determined to be a low temperature range, and the corresponding temperature correlation coefficient is... ≤0.95; The temperature-color temperature drift mapping table records the correspondence between different temperature ranges and the amount of color temperature drift, which is used for subsequent color temperature compensation.

[0010] Preferably, the pixel demand identification module generates the pixel demand matrix in the following manner: The acquired target image is preprocessed with noise removal and resolution normalization to make the resolution of the target image consistent with the pixel matrix resolution of the display panel; A pre-trained semantic segmentation model is used to perform pixel-level annotation on the preprocessed target image, classifying it into at least the following pixel types: text pixels, human skin pixels, dark area pixels of night scene, bright area pixels of landscape, and background pixels. Ideal parameters are set for each pixel type, including ideal brightness range, ideal color temperature, and display priority. The display priority is set based on the degree of influence of the pixel on the user's visual experience. The inter-frame difference algorithm is used to compare the target image for a consecutive preset number of frames, calculate the brightness change rate of the same pixel, and if the brightness change rate is less than a preset change threshold, it is marked as a static pixel. The static duration of the static pixel is then calculated. The pixel demand matrix is ​​generated by integrating the ideal parameters for each pixel type and the static duration of static pixels. Each entry in the pixel demand matrix corresponds one-to-one with a single pixel of the display panel.

[0011] Preferably, in the pixel requirement recognition module, the training dataset of the pre-trained semantic segmentation model covers sample images containing text, portraits, night scenes, landscapes, and illustrations. Each sample image is labeled with pixel type and corresponding ideal parameters; the ideal brightness range for text pixels is set to 280~320 nits, the ideal color temperature is set to 4300~4700K, and the display priority is set to the highest level; the ideal brightness range for portrait skin pixels is set to 260~300 nits, and the ideal color temperature is set to 4000~4400K. K, Display priority is high; for dark pixels in night scenes, the ideal brightness range is 150~200 nits, the ideal color temperature is 2800~3200K, and the display priority is medium; for bright pixels in landscapes, the ideal brightness range is 300~340 nits, the ideal color temperature is 4800~5200K, and the display priority is medium-high; for background pixels, the ideal brightness range is 200~240 nits, the ideal color temperature is 4600~5000K, and the display priority is low.

[0012] Preferably, the specific steps of the optimization parameter generation module in calculating the target adjustment parameters and generating a pixel-level adjustment parameter matrix include: For each pixel of the display panel, a target brightness optimization function and a target color temperature optimization function are constructed; the target brightness optimization function is: in, For target brightness, The display priority weight is equal to the display priority in the pixel demand matrix. This represents the average value of the ideal brightness range for the corresponding pixel type. This represents the pixel attenuation rate of the region corresponding to the pixel and is derived from the panel characteristic twin model. The attenuation influence coefficient is initially set at 0.15~0.2 for OLED panels and 0.1~0.15 for LCD panels. The temperature correlation coefficient is derived from the panel characteristic twin model. The initial value of the temperature influence coefficient is 0.04~0.06 for the low temperature range and 0.07~0.09 for the high temperature range. The target color temperature optimization function is: in, For the target color temperature, The ideal color temperature for the corresponding pixel type and derived from the pixel requirement matrix. The color temperature compensation value is determined based on the temperature-color temperature drift mapping table. The color temperature compensation value for LCD in the low-temperature range is 250~350K, and the color temperature compensation value for OLED in the high-temperature range is -150~250K. If the pixel decay rate of the corresponding pixel area If the attenuation effect coefficient β is greater than 30%, then the attenuation effect coefficient β will be increased by 0.04~0.06 based on the initial value; if the static duration of the corresponding pixel is greater than 10 minutes, then the display priority weight α will be decreased by 0.08~0.12 based on the initial value. When the display panel is an OLED panel, if the static duration of the static pixels is greater than 5 minutes, a dynamic fine-tuning rule is generated for it. The dynamic fine-tuning rule includes a fine-tuning time interval of 3 to 5 minutes, and a fine-tuning amplitude of ±2% to ±3% when the static duration is 5 to 8 minutes, ±3% to ±4% when the static duration is 8 to 12 minutes, and ±4% to ±5% when the static duration is greater than 12 minutes. The direction of brightness change is random each time. The pixel-level adjustment parameter matrix is ​​generated by integrating the target brightness, target color temperature of each pixel, and the dynamic fine-tuning rules of the OLED panel.

[0013] Preferably, in the optimization parameter generation module, the average value of the ideal brightness range The pixel attenuation rate is the arithmetic mean of the upper and lower limits of the ideal brightness range for the corresponding pixel type. The value is the average attenuation rate of the corresponding pixel's grid; if the pixel is located at the grid edge, the average attenuation rate of the two adjacent grids is used; the temperature influence coefficient... The value of the correlation coefficient with the temperature The degree of deviation from 1 increases as the value increases.

[0014] Preferably, the specific steps for the execution feedback module to implement closed-loop regulation include: The pixel-level adjustment parameter matrix is ​​converted into pixel-level display control instructions according to the display driving protocol. The display control instructions include the brightness control signal and color temperature control signal of each pixel of the display panel, and are transmitted to the display driving module of the digital photo frame to perform pixel-level adjustment. Within 1-2 seconds after the adjustment is executed, a feedback image is acquired. The actual brightness and actual color temperature of each pixel in the feedback image are extracted, and the deviation rate is calculated using the following formula: in, The deviation rate, These are the actual pixel parameter values. The target parameter value is set for each pixel. If the deviation rate of any pixel is greater than 8%, the optimization function coefficients are readjusted and the target adjustment parameters are recalculated until the deviation rate of all pixels on the display panel is less than or equal to 8%. The panel characteristic twin model is updated according to a preset cycle. If the deviation rate of pixels in the same area of ​​the display panel is greater than 10% in three consecutive feedbacks, an emergency update is triggered and the panel characteristic twin model is immediately regenerated.

[0015] Compared with the prior art, the beneficial effects of the present invention are: By employing panel characteristic twin modeling, precise perception of panel dynamic characteristics is achieved, avoiding adjustment failures caused by dynamic aging and temperature changes. Pixel demand identification enables precise matching of display requirements, prioritizing the display effect of key pixels. Optimized parameters generate multi-objective optimization functions and OLED static pixel fine-tuning rules, balancing display effect and panel lifespan while reducing the risk of static pixel damage. Execution of feedback loops verifies adjustment accuracy and updates the panel characteristic twin model, ensuring long-term adjustment accuracy. Simultaneously, hardware sensors are eliminated, achieving multi-panel compatibility and cost reduction. This approach solves problems such as adjustment failures and panel damage in existing technologies, balancing display experience, hardware protection, and economy. Attached Figure Description

[0016] Figure 1 This is a schematic diagram of the structure of a digital photo frame display intelligent adjustment system based on image content recognition, provided in an embodiment of the present invention. Detailed Implementation

[0017] The technical solutions of the embodiments of the present invention will be clearly and completely described below with reference to the accompanying drawings. Obviously, the described embodiments are only some embodiments of the present invention, and not all embodiments. Based on the embodiments of the present invention, all other embodiments obtained by those skilled in the art without creative effort are within the scope of protection of the present invention.

[0018] Please see Figure 1 This invention provides an intelligent adjustment system for digital photo frame display based on image content recognition, comprising: The panel characteristic twin modeling module is used to reverse-engineer the dynamic characteristic parameters of the digital photo frame display panel by comparing and analyzing standard test images with actual display images, and to generate a panel characteristic twin model that is consistent with the real-time state of the display panel; the dynamic characteristic parameters include pixel attenuation data, Gamma curve parameters, and temperature correlation coefficient. The pixel demand recognition module is used to process the target image currently displayed by the digital photo frame and generate a pixel demand matrix that corresponds one-to-one with the pixel matrix of the display panel. The pixel demand matrix contains the display demand information and static pixel markers for each pixel. The optimization parameter generation module is used to construct a multi-objective optimization function based on the panel characteristic twin model and the pixel demand matrix to calculate the target adjustment parameters of each pixel of the display panel. If the display panel is an OLED panel, static pixel fine-tuning rules are generated additionally, and finally a pixel-level adjustment parameter matrix is ​​output. The execution feedback module is used to convert the pixel-level adjustment parameter matrix into display control commands and execute the adjustment. The adjustment accuracy is verified by the feedback image. If the deviation exceeds the limit, the parameters are adjusted and optimized and readjusted. At the same time, the panel characteristic twin model is updated periodically to form a closed-loop adjustment.

[0019] In an optional embodiment, the specific steps of the panel characteristic twin modeling module to realize dynamic characteristic parameter back-calculation and twin model generation include: Input a preset standard test image set into the digital photo frame and acquire the corresponding actual display image. The preset standard test image set includes at least grayscale images and solid color images. Feature points are extracted from the grayscale image and its actual display image using a feature matching algorithm. The brightness deviation rate of the feature points is calculated to determine the pixel attenuation rate, and a pixel attenuation matrix of the display panel is generated. The pixel attenuation matrix is ​​the carrier of the pixel attenuation data. Key gray levels are extracted from the actual displayed image of the grayscale map and substituted into the Gamma curve formula to fit the real-time Gamma value. The real-time Gamma value is the core indicator of the Gamma curve parameters. The Gamma curve formula is: in, This refers to the actual brightness value of the key gray level. The actual brightness is 255 gray levels. This refers to the grayscale value of the current key grayscale level. =255, This is the real-time Gamma value; By comparing the standard color temperature of the solid color image with the actual color temperature of the actual displayed image, the temperature range of the display panel is determined and the temperature correlation coefficient is generated, and a temperature-color temperature drift mapping table is established. By integrating the pixel attenuation matrix, real-time Gamma value, and temperature correlation coefficient, a panel characteristic twin model is generated, and the deviation between the panel characteristic twin model and the actual state of the display panel is controlled within a preset deviation range.

[0020] In this embodiment, the "preset standard test image set" is an image set that provides a unified reference benchmark for back-calculating panel characteristics. Grayscale images are used to obtain brightness characteristics at different grayscale levels to calculate attenuation rates and fit Gamma curves, while solid color images are used to infer the temperature effect through color temperature changes. Together, they comprehensively cover the panel's brightness, grayscale, and color temperature characteristics. The "feature matching algorithm" is used to accurately locate feature points corresponding to the positions in the grayscale images and the actual displayed images, ensuring the accuracy of brightness deviation calculations and avoiding attenuation rate errors caused by feature point misalignment. The "pixel attenuation matrix" divides the panel into multiple regions through grid partitioning, with each region corresponding to an average attenuation rate. Its function is to transform discrete feature point attenuation data into a continuous panel region attenuation distribution, providing a regional attenuation basis for subsequent pixel-level adjustment. The "temperature-color temperature drift mapping table" records the correspondence between different temperature ranges and the amount of color temperature drift, serving as the core reference for subsequent color temperature compensation and ensuring that color temperature adjustment can accurately offset the drift effect when the temperature changes. In one possible implementation, the preset standard test image set may further include a gradient brightness map, which contains a continuous brightness gradient from dark to bright, which can further improve the density of feature point extraction and make the pixel attenuation matrix more accurate.

[0021] For example, in one feasible implementation, for a 10.1-inch OLED digital photo frame, the panel characteristic twin modeling module first inputs a preset standard test image set (including a 256-level grayscale image, a pure red image, and a gradient brightness image), and acquires the actual display image through the CMOS image acquisition module; an improved SIFT feature matching algorithm is used to extract 120 uniformly distributed feature points from the 256-level grayscale image and its actual display image, calculates the brightness deviation rate of each feature point to determine the pixel attenuation rate, and generates a 16×16 grid pixel attenuation matrix (20% attenuation rate in the central area and 25% attenuation rate in the corner areas); from the actual display of the grayscale image... Nine key gray levels (0, 32, 64, etc.) were extracted from the image. The actual brightness value of each key gray level was obtained and substituted into the Gamma curve formula. The least squares method was used to fit the real-time Gamma value γ=2.6. By comparing the standard color temperature (4500K) of the red pure color image with the actual display color temperature (4800K), the temperature range was determined to be a high temperature range. A temperature correlation coefficient T=1.08 was generated and a temperature-color temperature drift mapping table was established (for every 5℃ increase in the high temperature range, the color temperature drift is +50K). Finally, the above data were integrated to generate a panel characteristic twin model that is consistent with the real-time state of the OLED panel. The deviation between the model and the actual state of the panel was controlled within 4%.

[0022] In an optional embodiment, in the panel characteristic twin modeling module, the selected feature points are distributed in a uniform grid pattern in the grayscale image; the formula for calculating the pixel attenuation rate is: in, Pixel attenuation rate, The theoretical brightness of the corresponding feature point in the grayscale image. The measured brightness of the corresponding feature point in the actual displayed image; the grid density of the pixel attenuation matrix is ​​adapted to the pixel resolution of the display panel.

[0023] In this embodiment, the purpose of "uniform grid distribution of feature points" is to ensure that feature points cover the entire image area of ​​the grayscale image, avoiding attenuation rate calculation deviations caused by dense feature points in some areas and sparse feature points in others, and ensuring that the attenuation of each area of ​​the panel can be accurately reflected. "The grid density of the pixel attenuation matrix is ​​adapted to the panel pixel resolution" means that the number of grids is proportional to the panel resolution. For example, a 1920×1080 resolution panel can use a 32×32 grid, so that each grid corresponds to approximately 60×33 pixels, which ensures the consistency of pixel attenuation characteristics within each grid and avoids data redundancy caused by an excessive number of grids. In the pixel attenuation rate calculation formula, the degree of pixel attenuation can be intuitively quantified by the percentage of the difference between theoretical brightness and measured brightness relative to the theoretical brightness. This calculation method is not affected by the absolute value of brightness and is applicable to display panels of different models and brightness ranges. In one possible implementation, if the feature points are located in the edge area of ​​the panel (pixel attenuation is usually more obvious in the edge area), the feature point density in that area can be appropriately increased. For example, the spacing between feature points in the edge area can be half that in the center area to more accurately capture the difference in edge attenuation.

[0024] For example, in one feasible implementation, for a 1920×1080 resolution LCD display panel, the panel characteristic twin modeling module selects 270 feature points (distributed in a uniform grid, with an additional 30 feature points in the edge region) at an 80×45 pixel spacing in a 256-level grayscale image; when calculating a certain feature point, 120 nits It is 96 nits, substituting it into the formula gives... =(120-96) / 120×100%=20%; Based on the panel resolution, the pixel attenuation matrix is ​​divided into a 32×32 grid, with each grid corresponding to 60×33 pixels. The attenuation rate of adjacent feature points is averaged and then filled into the corresponding grid to finally generate a pixel attenuation matrix covering the entire panel. The deviation between this matrix and the actual attenuation state of the panel is less than 5%.

[0025] In an optional embodiment, the key gray levels extracted in the panel characteristic twin modeling module include 0, 32, 64, 96, 128, 160, 192, 224, and 255; the least squares method is used when fitting the real-time Gamma value, by adjusting... The value represents the actual brightness value for all key gray levels. The goal is to minimize the sum of squared errors between the calculated value and the Gamma curve formula; simultaneously, the brightness deviation value for each key gray level is output, where the brightness deviation value is the actual brightness value of the key gray level. The difference between the value calculated using the Gamma curve formula.

[0026] In this embodiment, nine key gray levels, including 0, 32, and 64, are selected because these gray levels uniformly cover the complete gray range of 0-255, comprehensively reflecting the brightness response characteristics of the panel at different gray levels and avoiding Gamma curve fitting deviations caused by too few key gray levels. The "least squares method" determines the optimal value by minimizing the sum of squared errors. The mathematical method used to calculate the value effectively reduces the impact of individual abnormal key gray levels on the overall fitting result, making the real-time Gamma value of the fit closer to the actual gray-level-brightness response relationship of the panel. The "brightness deviation value" is used to quantify the difference between the actual brightness of each key gray level and the calculated value of the fitted curve. The larger the absolute value, the more significantly the panel brightness deviates from the ideal state at that gray level. Targeted parameter adjustments can then be made for gray levels with large deviations. In one possible implementation, if the absolute value of the brightness deviation value of a key gray level exceeds 10 nits, the number of key gray levels near that gray level can be increased in the next model update (e.g., adding 80 gray levels between 64 and 96) to improve the fitting accuracy of that area.

[0027] For example, in one feasible implementation, for a 10.1-inch OLED digital photo frame, the panel characteristic twin modeling module extracts nine key gray levels (0, 32, 64, 96, 128, 160, 192, 224, and 255) from the actual grayscale image, and measures the grayscale values ​​of each key gray level. The values ​​are 0 nits, 40 nits, 100 nits, 160 nits, 220 nits, 280 nits, 310 nits, 350 nits, and 400 nits, respectively; substitute these values ​​into the Gamma curve formula. Adjustment using the least squares method Value, when At a value of 2.6, the sum of squared errors for all key gray levels is minimized (total sum of squared errors is 85 nits). 2 Simultaneously, the brightness deviation value for each key gray level is calculated, where the deviation value for the 128 gray levels is 220 nits - [400× ]≈220-212=8 nits, the deviation value for 64 gray levels is 100 nits-[400× The brightness deviation of all key gray levels is less than 10 nits, which meets the fitting accuracy requirements. (Approximately 100-95=5 nits) In an optional embodiment, in the panel characteristic twin modeling module, for an LCD display panel, if the actual color temperature is lower than the standard color temperature and the deviation is ≥150K, then the temperature range is determined to be a low-temperature range, and the corresponding temperature correlation coefficient is... ≤0.95; If the actual color temperature is higher than the standard color temperature and the deviation is ≥100K, then the temperature range is determined to be a high-temperature range, and the corresponding temperature correlation coefficient is... ≥1.05; For OLED display panels, if the actual color temperature is higher than the standard color temperature and the deviation is ≥200K, the temperature range is determined to be a high-temperature range, and the corresponding temperature correlation coefficient is... ≥1.05; If the actual color temperature is lower than the standard color temperature and the deviation is ≥150K, then the temperature range is determined to be a low temperature range, and the corresponding temperature correlation coefficient is... ≤0.95; The temperature-color temperature drift mapping table records the correspondence between different temperature ranges and the amount of color temperature drift, which is used for subsequent color temperature compensation.

[0028] In this embodiment, different color temperature deviation thresholds are used to determine the temperature range for different display panels (LCD and OLED) because their color temperatures are less sensitive to temperature. LCD panels are prone to warm-tonal drift (color temperature decrease) at low temperatures, and a drift of more than 150K significantly impacts display performance. At high temperatures, cool-tonal drift (color temperature increase) requires an intervention of 100K. OLED panels exhibit more significant cool-tonal drift at high temperatures, requiring a deviation of 200K to be considered a high-temperature range. At low temperatures, warm-tonal drift is consistent with LCD (150K threshold). This differentiated threshold setting ensures the accuracy of temperature range determination. The "temperature correlation coefficient T" is a parameter that quantifies the degree of temperature's influence on panel characteristics. The greater T deviates from 1, the greater the impact of temperature on panel brightness and color temperature, requiring more significant parameter corrections in subsequent adjustments. The "temperature-color temperature drift mapping table" binds temperature ranges to specific color temperature drift amounts. For example, the LCD low-temperature range (<5℃) corresponds to a color temperature drift of -200K, which is used in the subsequent target color temperature optimization function. Provide specific numerical data to ensure accurate color temperature compensation. In one possible implementation, the threshold for color temperature deviation of different panel models can be fine-tuned based on the temperature-color temperature characteristic data provided by the panel manufacturer. For example, if a color temperature drift of 130K at low temperatures affects the display effect of a certain brand of LCD panel, its low-temperature range judgment threshold can be adjusted to 130K.

[0029] For example, in one feasible implementation, for a 15.6-inch LCD digital photo frame, the panel characteristic twin modeling module inputs a pure blue color image (standard color temperature 5000K), acquires the actual displayed image and measures the actual color temperature to be 4800K, with a color temperature deviation of -200K (lower than the standard color temperature but ≥150K). The temperature range is determined to be a low-temperature range (actual ambient temperature 3℃), generating a temperature correlation coefficient T=0.93. Simultaneously, a temperature-color temperature drift mapping table is established, recording "Low-temperature range (<5℃): color temperature drift -200K, normal temperature range (5℃-35℃): color temperature drift 0K, high-temperature range (>35℃): color temperature drift +100K". The target color temperature can then be determined based on this mapping table. =+200K to counteract the warm tone shift caused by low temperature, so that the final displayed color temperature is close to the ideal value. In an optional embodiment, the pixel demand identification module generates the pixel demand matrix in the following manner: The acquired target image is preprocessed with noise removal and resolution normalization to make the resolution of the target image consistent with the pixel matrix resolution of the display panel; A pre-trained semantic segmentation model is used to perform pixel-level annotation on the preprocessed target image, classifying it into at least the following pixel types: text pixels, human skin pixels, dark area pixels of night scene, bright area pixels of landscape, and background pixels. Ideal parameters are set for each pixel type, including ideal brightness range, ideal color temperature, and display priority. The display priority is set based on the degree of influence of the pixel on the user's visual experience. The inter-frame difference algorithm is used to compare the target image for a consecutive preset number of frames, calculate the brightness change rate of the same pixel, and if the brightness change rate is less than a preset change threshold, it is marked as a static pixel. The static duration of the static pixel is then calculated. The pixel demand matrix is ​​generated by integrating the ideal parameters for each pixel type and the static duration of static pixels. Each entry in the pixel demand matrix corresponds one-to-one with a single pixel of the display panel.

[0030] It should be noted that in this embodiment, "noise removal processing" is used to eliminate random noise in the target image (such as noise generated during shooting or transmission) to avoid noise interfering with the semantic segmentation model's judgment of pixel type. In this embodiment, a Gaussian filtering algorithm is used, which can preserve image edge details well while removing noise. "Resolution normalization processing" scales the resolution of the target image to be consistent with the pixel matrix of the display panel, ensuring that subsequent pixel-level annotation results can directly correspond to the panel pixels, avoiding misalignment of the requirement matrix due to resolution mismatch. "Pre-trained semantic segmentation model" is an image segmentation model trained with a large number of labeled samples. Its function is to automatically identify and label different types of pixels in the target image without manual intervention, improving the efficiency and accuracy of pixel type classification. "Inter-frame difference algorithm" determines whether the pixel is in a static state by comparing the brightness changes of the same pixel in multiple consecutive frames of the target image, providing precise positioning for static pixel protection of the OLED panel. The setting of "display priority" is based on the impact of pixels on the user's visual experience—text pixels and portrait skin pixels are the focus of user viewing, so they have high priority, while background pixels have less impact on the experience, so they have low priority. This setting can ensure that the display needs of key pixels are prioritized during adjustment. In one possible implementation, the number of consecutive preset frames can be adjusted according to the target image type. For static images (such as photos), the preset number of frames is set to 5 frames, and for dynamic images (such as animated illustrations), the preset number of frames is set to 3 frames, in order to adapt to the dynamic characteristics of different images. For example, in one feasible implementation, for a 10.1-inch OLED digital photo frame, the pixel demand recognition module first acquires the currently displayed family gathering text notification image (initial resolution 2048×1536), performs noise removal using a 3×3 Gaussian filter, and then normalizes the resolution to 1920×1080 (consistent with the panel pixel matrix); a pre-trained improved U-Net semantic segmentation model is used to label the preprocessed image, labeling the "family gathering notification" text area as text pixels (ideal brightness range 280-320 nits, ideal color temperature 4500K, display priority 1.0), and labeling the portrait area as... The portrait skin pixels (ideal brightness range 260-300 nits, ideal color temperature 4200K, display priority 0.9) and the background area pixels (ideal brightness range 200-240 nits, ideal color temperature 4800K, display priority 0.6) are used. An inter-frame difference algorithm is used to compare five consecutive frames. If the brightness change rate of text pixels is less than 3% (preset change threshold is 5%), it is marked as a static pixel and the static duration is calculated to be 3 minutes. Finally, the above data is integrated to generate a 1920×1080 pixel requirement matrix. Each matrix entry corresponds one-to-one with a single pixel on the panel, including the pixel type, ideal parameters, and static label. In an optional embodiment, the pixel requirement recognition module uses a training dataset for a pre-trained semantic segmentation model that includes sample images of text, portraits, night scenes, landscapes, and illustrations. Each sample image is labeled with its pixel type and corresponding ideal parameters. The ideal brightness range for text pixels is set to 280-320 nits, the ideal color temperature to 4300-4700K, and the highest display priority. The ideal brightness range for portrait skin pixels is set to 260-300 nits, and the ideal color temperature to 4000-4700K. 400K, display priority is high; the ideal brightness range for dark pixels in night scenes is 150~200 nits, ideal color temperature is 2800~3200K, display priority is medium; the ideal brightness range for bright pixels in landscapes is 300~340 nits, ideal color temperature is 4800~5200K, display priority is medium-high; the ideal brightness range for background pixels is 200~240 nits, ideal color temperature is 4600~5000K, display priority is low.

[0031] It should be noted that in this embodiment, "the training dataset covers a variety of sample images" to enable the pre-trained semantic segmentation model to recognize common image types in digital photo frames, avoiding recognition bias caused by a single sample. For example, including night scene image samples can improve the model's recognition accuracy for dark pixels in night scenes, and including illustration samples can improve the accuracy of labeling special color pixels in dynamic illustrations. The "ideal parameters for each pixel type" are set based on user visual experience and image display rules—text pixels need higher brightness and moderate color temperature to ensure readability, portrait skin pixels need a color temperature close to natural light to restore skin tone, dark pixels in night scenes need lower brightness to avoid overexposure and highlight dark details, bright pixels in landscapes need higher brightness to show landscape details, and background pixels need moderate brightness to avoid interfering with foreground pixels. These parameter settings ensure that different types of pixels can present optimal display effects. In one possible implementation, users can fine-tune the ideal brightness range and ideal color temperature for each pixel type by ±10% through the digital photo frame's settings interface to adapt to different users' visual preferences. For example, in one feasible implementation, the training dataset of the pre-trained semantic segmentation model of the pixel requirement recognition module contains 150,000 sample images (including 30,000 text images, 40,000 portrait images, 20,000 night scene images, 30,000 landscape images, and 30,000 illustration images). Each sample image is labeled with pixel type and ideal parameters by professionals. When recognizing a night scene portrait photo, the model labels the face of the person in the photo as portrait skin pixels (with ideal parameters set at 260-300 nits brightness and 4000-4400K color temperature), labels the night sky area as night scene dark area pixels (with ideal parameters set at 150-200 nits brightness and 2800-3200K color temperature), and labels the text in the photo as text pixels (with ideal parameters set at 280-320 nits brightness and 4300-4700K color temperature). The labeling accuracy is over 92%, which can accurately match the display requirements of different pixels.

[0032] In an optional embodiment, the specific steps of the optimization parameter generation module in calculating the target adjustment parameters and generating a pixel-level adjustment parameter matrix include: For each pixel of the display panel, a target brightness optimization function and a target color temperature optimization function are constructed; the target brightness optimization function is: in, For target brightness, The display priority weight is equal to the display priority in the pixel demand matrix. This represents the average value of the ideal brightness range for the corresponding pixel type. This represents the pixel attenuation rate of the region corresponding to the pixel and is derived from the panel characteristic twin model. The attenuation influence coefficient is initially set at 0.15~0.2 for OLED panels and 0.1~0.15 for LCD panels. The temperature correlation coefficient is derived from the panel characteristic twin model. The initial value of the temperature influence coefficient is 0.04~0.06 for the low temperature range and 0.07~0.09 for the high temperature range. The target color temperature optimization function is: in, For the target color temperature, The ideal color temperature for the corresponding pixel type and derived from the pixel requirement matrix. The color temperature compensation value is determined based on the temperature-color temperature drift mapping table. The color temperature compensation value for LCD in the low-temperature range is 250~350K, and the color temperature compensation value for OLED in the high-temperature range is -150~250K. If the pixel decay rate of the corresponding pixel area If the attenuation effect coefficient β is greater than 30%, then the attenuation effect coefficient β will be increased by 0.04~0.06 based on the initial value; if the static duration of the corresponding pixel is greater than 10 minutes, then the display priority weight α will be decreased by 0.08~0.12 based on the initial value. When the display panel is an OLED panel, if the static duration of the static pixels is greater than 5 minutes, a dynamic fine-tuning rule is generated for it. The dynamic fine-tuning rule includes a fine-tuning time interval of 3 to 5 minutes, and a fine-tuning amplitude of ±2% to ±3% when the static duration is 5 to 8 minutes, ±3% to ±4% when the static duration is 8 to 12 minutes, and ±4% to ±5% when the static duration is greater than 12 minutes. The direction of brightness change is random each time. The pixel-level adjustment parameter matrix is ​​generated by integrating the target brightness, target color temperature of each pixel, and the dynamic fine-tuning rules of the OLED panel.

[0033] In this embodiment, the "target brightness optimization function" is a comprehensive function that considers "display requirements (...)". ), panel lifespan ( Temperature effect The multi-objective optimization model, whose core idea is to ensure that the adjusted brightness will neither be insufficient due to attenuation nor damage to the panel due to excessive brightness, while meeting the ideal display requirements of pixels and deducting the effects of attenuation and temperature; the "attenuation influence coefficient" The initial value difference is because OLED panels experience faster pixel decay than LCDs and are more sensitive to brightness, thus requiring higher initial values. The value is used to enhance the correction effect of attenuation on brightness; "Temperature Influence Coefficient" The difference in brightness levels is because low temperatures have a weaker effect on suppressing panel brightness, while high temperatures have a stronger effect on accelerating brightness decay. Therefore, the high-temperature range... Larger values; the "target color temperature optimization function" is achieved through... To ensure the ideal color temperature requirements of the pixels, by To counteract color temperature drift caused by temperature fluctuations and ensure accurate color temperature display, the "dynamic fine-tuning rule" is designed for static pixels on OLED panels. Through small-amplitude, random-direction brightness adjustments, it avoids screen burn-in caused by static pixels remaining at the same brightness level for extended periods. The fine-tuning amplitude increases with the duration of static operation, balancing protection and visual experience. In one possible implementation, the attenuation influence coefficient... The increase can be based on The specific numerical segmentation settings, Increase by 0.04 when the concentration is between 30% and 40%. Increase by 0.06 when it is >40% to more accurately address panels with different levels of decay.

[0034] For example, in one feasible implementation, for a 10.1-inch OLED digital photo frame, the optimized parameter generation module optimizes the text pixels (located in the center area of ​​the panel). Calculate the target adjustment parameters (20%, static duration 3 minutes): =1.0 (highest priority displayed) =300 nits (the average value of the ideal brightness range of 280-320 nits for text pixels). =0.15 (initial value for OLED) =1.08 (high temperature range) =0.08 (initial value in the high-temperature range), substituting into the target brightness optimization function yields... Rounded down to 297 nits; =4500K (ideal color temperature for text pixels). =-200K (OLED high-temperature range color temperature compensation value), substituting into the target color temperature optimization function yields... =4500-200=4300K; Since the static duration of this text pixel is 3 minutes < 5 minutes, there is no need to generate dynamic fine-tuning rules; for text pixels in the corner areas ( =25%), calculated as follows The value is rounded down to 296 nits. For the OLED background pixels with a static display duration of 7 minutes, a dynamic fine-tuning strategy is implemented: fine-tuning intervals of 3 minutes, with the adjustment range controlled within ±3%. Finally, the target brightness for all pixels is determined. Target chromaticity The above dynamic fine-tuning strategies are integrated to generate a pixel-level adjustment parameter matrix with a resolution of 1920×1080.

[0035] In an optional embodiment, in the optimization parameter generation module, the average value of the ideal brightness range The pixel attenuation rate is the arithmetic mean of the upper and lower limits of the ideal brightness range for the corresponding pixel type. The value is the average attenuation rate of the corresponding pixel's grid; if the pixel is located at the grid edge, the average attenuation rate of the two adjacent grids is used; the temperature influence coefficient... The value of the correlation coefficient with the temperature The degree of deviation from 1 increases as the value increases.

[0036] In this embodiment, " "Taking the arithmetic mean" is to select an intermediate value between the upper and lower limits of the ideal brightness range, ensuring a close approximation to the ideal display effect while reserving adjustment space for subsequent attenuation and temperature correction, avoiding the correction brightness exceeding the reasonable range due to an initial value that is too high or too low; "Taking the average attenuation rate of adjacent grids for edge pixels" is to avoid abrupt changes in the attenuation rate at grid edges, making the attenuation rate transition of edge pixels smoother, thereby making the target brightness adjustment more uniform and avoiding obvious brightness boundaries; Follow The increase is due to the greater deviation from 1. The greater the deviation from 1, the more significant the effect of temperature on panel brightness, requiring a larger [temperature value]. Values ​​are used to correct brightness deviations caused by temperature, for example =1.1 (high temperature has a significant impact) =0.1, =1.05 (high temperature has little effect) =0.07, this setting ensures the accuracy of temperature effect correction. In one possible implementation, if the attenuation rate difference between two adjacent grids exceeds 10%, the attenuation rate of edge pixels is weighted (grids closer to the pixel have a weight of 0.6, and those farther away have a weight of 0.4) to more closely reflect the attenuation situation of the actual area where the pixel is located.

[0037] For example, in one feasible implementation, for a 10.1-inch OLED digital photo frame, the number of pixels for the skin of a human figure is calculated. The ideal brightness range is 260-300 nits, and the arithmetic mean is (260+300) / 2=280 nits; a pixel is located at the edge of the "center grid with a 20% attenuation rate" and the "corner grid with a 25% attenuation rate" in the pixel attenuation matrix, and the average attenuation rate of the adjacent grids is taken. =(20%+25%) / 2=22%; Temperature correlation coefficient =1.08 (a deviation of 0.08 from 1), corresponding to the temperature influence coefficient. =0.08 ( =1.05 =0.07, =1.1 =0.09, linear interpolation =1.08 =0.08); Substituting this into the target brightness optimization function yields The value is rounded up to 249 nits. This brightness not only meets the display requirements of human skin pixels, but also takes into account the attenuation in the edge areas and the impact of high temperature.

[0038] In an optional embodiment, the specific steps for the execution feedback module to implement closed-loop regulation include: The pixel-level adjustment parameter matrix is ​​converted into pixel-level display control instructions according to the display driving protocol. The display control instructions include the brightness control signal and color temperature control signal of each pixel of the display panel, and are transmitted to the display driving module of the digital photo frame to perform pixel-level adjustment. Within 1-2 seconds after the adjustment is executed, a feedback image is acquired, and the actual brightness of each pixel in the feedback image is extracted. and actual color temperature Calculate the deviation rate using the following formula: in, The deviation rate, These are the actual pixel parameter values. The target parameter value is set for each pixel. If the deviation rate of any pixel is greater than 8%, the optimization function coefficients are readjusted and the target adjustment parameters are recalculated until the deviation rate of all pixels on the display panel is less than or equal to 8%. The panel characteristic twin model is updated according to a preset cycle. If the deviation rate of pixels in the same area of ​​the display panel is greater than 10% in three consecutive feedbacks, an emergency update is triggered and the panel characteristic twin model is immediately regenerated.

[0039] In this embodiment, the "display driver protocol" is a communication standard connecting the pixel-level adjustment parameter matrix and the display driver module. Its function is to transmit abstract target parameters ( , The adjustment commands are converted into electrical signals recognizable by the drive module (e.g., brightness control signals are PWM duty cycle signals, color temperature control signals are RGB channel gain signals) to ensure accurate execution. "Feedback image acquisition and deviation rate calculation" is the core of closed-loop adjustment. By comparing the deviation rate between actual and target parameters, the adjustment effect can be determined. The 8% deviation rate threshold is based on the human visual perception threshold; when the deviation rate is ≤8%, the user cannot perceive a significant difference, ensuring display quality. "Preset cycle update of the twin model" is to adapt to dynamic changes in panel characteristics (e.g., slow decay) and avoid adjustment failure due to model obsolescence. The preset cycle is typically set to 1 hour, balancing update accuracy and system resource consumption. "Emergency update" is used to address sudden changes in panel characteristics (e.g., sudden color temperature changes due to a rapid increase in ambient temperature). Three consecutive deviation rate values ​​>10% indicate that the model can no longer reflect the actual state of the panel and needs immediate updating to avoid continuous adjustment deviation. In one possible implementation, the preset cycle can be dynamically adjusted according to the panel's usage time. For new panels used for less than one year, the cycle is set to 1.5 hours; for panels used for more than one year, the cycle is set to 1 hour to adapt to changes in panel aging speed.

[0040] Additionally, it should be noted that in the deviation rate formula... and , The connection, with and , The association is achieved by using general parameter symbols to refer to specific parameter values: As a general symbol for "actual pixel parameter values", it specifically refers to when calculating the luminance deviation rate. (The actual brightness of pixels in the feedback image) specifically refers to the color temperature deviation rate when calculating the color temperature deviation rate. (The actual color temperature of the pixels in the feedback image); similarly, As a general symbol for "pixel target parameter value", it specifically refers to when calculating the luminance deviation rate. (Target brightness output by the optimized parameter generation module), specifically refers to when calculating the color temperature deviation rate. (Target color temperature output by the optimized parameter generation module). In one possible implementation, the preset cycle can be dynamically adjusted according to the panel's usage time. For new panels, the cycle is set to 1.5 hours within the first year of use, and to 1 hour for panels used for more than one year, to adapt to changes in the panel's aging rate.

[0041] For example, in one feasible implementation, for a 10.1-inch OLED digital photo frame, the execution feedback module first converts the pixel-level adjustment parameter matrix into pixel-level control commands according to the I2C display driver protocol (the brightness control signal is the corresponding...). The PWM duty cycle, the color temperature control signal is the corresponding The RGB channel gain is transmitted to the OLED driver chip (SSD1963) for adjustment; after adjustment, a feedback image is acquired 1.5 seconds later to extract the final actual brightness of the text pixels. ( Substituting into the deviation rate formula, we get Final actual color temperature ( The deviation rate was approximately 0.47% ≤ 8%, indicating that the adjustment effect met the standard. The panel characteristic twin model was updated at a preset cycle of 1 hour. In one feedback instance, it was found that the deviation rate of pixels in the corner area was > 10% for three consecutive times (due to a sudden increase in ambient temperature to 40℃). An emergency update was immediately triggered to regenerate the twin model (temperature correlation coefficient). =1.12), and recalculated the target adjustment parameters based on the new model to bring the deviation rate back to within 5%, ensuring that subsequent adjustments are always accurate.

[0042] In this embodiment, precise perception of panel dynamic characteristics is achieved through panel characteristic twin modeling, avoiding adjustment failures caused by dynamic aging and temperature changes; precise matching of display requirements is achieved through pixel demand identification, prioritizing the display effect of key pixels; multi-objective optimization functions are constructed by optimizing parameters and generating OLED static pixel fine-tuning rules to balance display effect and panel lifespan, reducing the risk of static pixel damage; the adjustment accuracy is verified by executing a feedback closed loop and the panel characteristic twin model is updated to ensure long-term adjustment accuracy; at the same time, hardware sensors are eliminated, achieving multi-panel compatibility and cost reduction. This solves the problems of adjustment failure and panel damage in existing technologies, balancing display experience, hardware protection, and economy.

[0043] Furthermore, it should be noted that the combination of the various technical features in this case is not limited to the combination methods described in the claims of this case or the combination methods described in the specific embodiments. All technical features described in this case can be freely combined or combined in any way, unless they contradict each other.

[0044] It should be noted that the above examples are merely specific embodiments of the present invention, and the present invention is obviously not limited to the above embodiments, with many similar variations. All modifications that can be directly derived or conceived by those skilled in the art from the content disclosed in this invention should fall within the protection scope of this invention.

[0045] The above are merely preferred embodiments of the present invention and are not intended to limit the scope of protection of the present invention. Any modifications, equivalent substitutions, improvements, etc., made within the spirit and principles of the present invention should be included within the scope of protection of the present invention.

Claims

1. A digital photo frame display intelligent adjustment system based on image content recognition, characterized in that, include: The panel characteristic twin modeling module is used to reverse-engineer the dynamic characteristic parameters of the digital photo frame display panel by comparing and analyzing standard test images with actual display images, and to generate a panel characteristic twin model that is consistent with the real-time state of the display panel; the dynamic characteristic parameters include pixel attenuation data, Gamma curve parameters, and temperature correlation coefficient. The pixel demand recognition module is used to process the target image currently displayed by the digital photo frame and generate a pixel demand matrix that corresponds one-to-one with the pixel matrix of the display panel. The pixel demand matrix contains the display demand information and static pixel markers for each pixel. The optimization parameter generation module is used to construct a multi-objective optimization function based on the panel characteristic twin model and the pixel demand matrix to calculate the target adjustment parameters of each pixel of the display panel. If the display panel is an OLED panel, static pixel fine-tuning rules are generated additionally, and finally a pixel-level adjustment parameter matrix is ​​output. The execution feedback module is used to convert the pixel-level adjustment parameter matrix into display control commands and execute the adjustment. The adjustment accuracy is verified by the feedback image. If the deviation exceeds the limit, the parameters are adjusted and optimized and readjusted. At the same time, the panel characteristic twin model is updated periodically to form a closed-loop adjustment.

2. The intelligent adjustment system for digital photo frame display based on image content recognition according to claim 1, characterized in that, The specific steps of the panel characteristic twin modeling module to realize dynamic characteristic parameter back-inference and twin model generation include: Input a preset standard test image set into the digital photo frame and acquire the corresponding actual display image. The preset standard test image set includes at least grayscale images and solid color images. Feature points are extracted from the grayscale image and its actual display image using a feature matching algorithm. The brightness deviation rate of the feature points is calculated to determine the pixel attenuation rate, and a pixel attenuation matrix of the display panel is generated. The pixel attenuation matrix is ​​the carrier of the pixel attenuation data. Key gray levels are extracted from the actual displayed image of the grayscale map and substituted into the Gamma curve formula to fit the real-time Gamma value. The real-time Gamma value is the core indicator of the Gamma curve parameters. The Gamma curve formula is: in, This refers to the actual brightness value of the key gray level. The actual brightness is 255 gray levels. This refers to the grayscale value of the current key grayscale level. =255, This is the real-time Gamma value; By comparing the standard color temperature of the solid color image with the actual color temperature of the actual displayed image, the temperature range of the display panel is determined and the temperature correlation coefficient is generated, and a temperature-color temperature drift mapping table is established. By integrating the pixel attenuation matrix, real-time Gamma value, and temperature correlation coefficient, a panel characteristic twin model is generated, and the deviation between the panel characteristic twin model and the actual state of the display panel is controlled within a preset deviation range.

3. The intelligent adjustment system for digital photo frame display based on image content recognition according to claim 2, characterized in that, In the panel characteristic twin modeling module, the selected feature points are distributed in a uniform grid pattern in the grayscale image; the formula for calculating the pixel attenuation rate is: in, Pixel attenuation rate, The theoretical brightness of the corresponding feature point in the grayscale image. The measured brightness of the corresponding feature point in the actual displayed image; the grid density of the pixel attenuation matrix is ​​adapted to the pixel resolution of the display panel.

4. The intelligent adjustment system for digital photo frame display based on image content recognition according to claim 2, characterized in that, In the panel characteristic twin modeling module, the extracted key gray levels include 0, 32, 64, 96, 128, 160, 192, 224, and 255; the least squares method is used when fitting the real-time Gamma value, and adjustments are made... The value represents the actual brightness value for all key gray levels. The goal is to minimize the sum of squared errors between the calculated value and the Gamma curve formula; simultaneously, the brightness deviation value for each key gray level is output, where the brightness deviation value is the actual brightness value of the key gray level. The difference between the value calculated using the Gamma curve formula.

5. The intelligent adjustment system for digital photo frame display based on image content recognition according to claim 2, characterized in that, In the panel characteristic twin modeling module, for an LCD display panel, if the actual color temperature is lower than the standard color temperature and the deviation is ≥150K, then the temperature range is determined to be a low-temperature range, and the corresponding temperature correlation coefficient is... ≤0.95; If the actual color temperature is higher than the standard color temperature and the deviation is ≥100K, then the temperature range is determined to be a high-temperature range, and the corresponding temperature correlation coefficient is... ≥1.05; For OLED display panels, if the actual color temperature is higher than the standard color temperature and the deviation is ≥200K, the temperature range is determined to be a high-temperature range, and the corresponding temperature correlation coefficient is... ≥1.05; If the actual color temperature is lower than the standard color temperature and the deviation is ≥150K, then the temperature range is determined to be a low temperature range, and the corresponding temperature correlation coefficient is... ≤0.95; The temperature-color temperature drift mapping table records the correspondence between different temperature ranges and the amount of color temperature drift, which is used for subsequent color temperature compensation.

6. The intelligent adjustment system for digital photo frame display based on image content recognition according to claim 1, characterized in that, The pixel demand identification module generates the pixel demand matrix in the following manner: The acquired target image is preprocessed with noise removal and resolution normalization to make the resolution of the target image consistent with the pixel matrix resolution of the display panel; A pre-trained semantic segmentation model is used to perform pixel-level annotation on the preprocessed target image, classifying it into at least the following pixel types: text pixels, human skin pixels, dark area pixels of night scene, bright area pixels of landscape, and background pixels. Ideal parameters are set for each pixel type, including ideal brightness range, ideal color temperature, and display priority. The display priority is set based on the degree of influence of the pixel on the user's visual experience. The inter-frame difference algorithm is used to compare the target image for a consecutive preset number of frames, calculate the brightness change rate of the same pixel, and if the brightness change rate is less than a preset change threshold, it is marked as a static pixel. The static duration of the static pixel is then calculated. The pixel demand matrix is ​​generated by integrating the ideal parameters for each pixel type and the static duration of static pixels. Each entry in the pixel demand matrix corresponds one-to-one with a single pixel of the display panel.

7. The intelligent adjustment system for digital photo frame display based on image content recognition according to claim 6, characterized in that, In the pixel requirement recognition module, the training dataset of the pre-trained semantic segmentation model covers sample images containing text, portraits, night scenes, landscapes, and illustrations. Each sample image is labeled with pixel type and corresponding ideal parameters. The ideal brightness range for text pixels is set to 280~320 nits, the ideal color temperature is 4300~4700K, and the display priority is the highest. The ideal brightness range for portrait skin pixels is set to 260~300 nits, the ideal color temperature is 4000~4400K, and the display priority is high. The ideal brightness range for dark area pixels in night scenes is set to 150~200 nits, the ideal color temperature is 2800~3200K, and the display priority is medium. The ideal brightness range for bright area pixels in landscapes is set to 300~340 nits, the ideal color temperature is 4800~5200K, and the display priority is medium to high. The ideal brightness range for background pixels is set to 200~240 nits, the ideal color temperature is 4600~5000K, and the display priority is low.

8. The intelligent adjustment system for digital photo frame display based on image content recognition according to claim 1, characterized in that, The specific steps of the optimization parameter generation module in calculating the target adjustment parameters and generating a pixel-level adjustment parameter matrix include: For each pixel of the display panel, a target brightness optimization function and a target color temperature optimization function are constructed; the target brightness optimization function is: in, For target brightness, The display priority weight is equal to the display priority in the pixel demand matrix. This represents the average value of the ideal brightness range for the corresponding pixel type. This represents the pixel attenuation rate of the region corresponding to the pixel and is derived from the panel characteristic twin model. The attenuation influence coefficient is initially set at 0.15~0.2 for OLED panels and 0.1~0.15 for LCD panels. The temperature correlation coefficient is derived from the panel characteristic twin model. The initial value of the temperature influence coefficient is 0.04~0.06 for the low temperature range and 0.07~0.09 for the high temperature range. The target color temperature optimization function is: in, For the target color temperature, The ideal color temperature for the corresponding pixel type and derived from the pixel requirement matrix. The color temperature compensation value is determined based on the temperature-color temperature drift mapping table. The color temperature compensation value for LCD in the low-temperature range is 250~350K, and the color temperature compensation value for OLED in the high-temperature range is -150~250K. If the pixel decay rate of the corresponding pixel area If the attenuation effect coefficient β is greater than 30%, then the attenuation effect coefficient β will be increased by 0.04~0.06 based on the initial value; if the static duration of the corresponding pixel is greater than 10 minutes, then the display priority weight α will be decreased by 0.08~0.12 based on the initial value. When the display panel is an OLED panel, if the static duration of the static pixels is greater than 5 minutes, a dynamic fine-tuning rule is generated for it. The dynamic fine-tuning rule includes a fine-tuning time interval of 3 to 5 minutes, and a fine-tuning amplitude of ±2% to ±3% when the static duration is 5 to 8 minutes, ±3% to ±4% when the static duration is 8 to 12 minutes, and ±4% to ±5% when the static duration is greater than 12 minutes. The direction of brightness change is random each time. The pixel-level adjustment parameter matrix is ​​generated by integrating the target brightness, target color temperature of each pixel, and the dynamic fine-tuning rules of the OLED panel.

9. The intelligent adjustment system for digital photo frame display based on image content recognition according to claim 8, characterized in that, In the optimization parameter generation module, the average value of the ideal brightness range The pixel attenuation rate is the arithmetic mean of the upper and lower limits of the ideal brightness range for the corresponding pixel type. The value is the average attenuation rate of the corresponding pixel's grid; if the pixel is located at the grid edge, the average attenuation rate of the two adjacent grids is used; the temperature influence coefficient... The value of the correlation coefficient with the temperature The degree of deviation from 1 increases as the value increases.

10. The intelligent adjustment system for digital photo frame display based on image content recognition according to claim 1, characterized in that, The specific steps for the execution feedback module to achieve closed-loop regulation include: The pixel-level adjustment parameter matrix is ​​converted into pixel-level display control instructions according to the display driving protocol. The display control instructions include the brightness control signal and color temperature control signal of each pixel of the display panel, and are transmitted to the display driving module of the digital photo frame to perform pixel-level adjustment. Within 1-2 seconds after the adjustment is executed, a feedback image is acquired. The actual brightness and actual color temperature of each pixel in the feedback image are extracted, and the deviation rate is calculated using the following formula: in, The deviation rate, These are the actual pixel parameter values. The target parameter value is set for each pixel. If the deviation rate of any pixel is greater than 8%, the optimization function coefficients are readjusted and the target adjustment parameters are recalculated until the deviation rate of all pixels on the display panel is less than or equal to 8%. The panel characteristic twin model is updated according to a preset cycle. If the deviation rate of pixels in the same area of ​​the display panel is greater than 10% in three consecutive feedbacks, an emergency update is triggered and the panel characteristic twin model is immediately regenerated.

Images