A remote status monitoring system for LED display screen based on Internet of Things

Abstract

The present invention relates to the field of LED display screen control technology, and in particular to a remote state monitoring system for an LED display screen based on the Internet of Things. The present invention obtains real-time light data through a light monitoring module, obtains environmental data through an environmental monitoring module, and predicts the light conditions of a target display screen in a preset prediction time period in the future through a remote control server, obtains a predicted light distribution diagram, and corrects the target video to obtain a corrected video, and finally transmits the corrected video to the target display screen for playback within a preset time period. Compared with the prior art, the present invention assigns the complex frame-by-frame optimization work to the server for operation, and enables the correction work to be performed in advance by means of prediction, ensuring that the corrected video can be played in a timely manner, solving the problem of how to achieve dynamic control of an LED display screen without improving the hardware level of the LED display screen in the prior art.

Description

Technical Field

[0001] The present invention relates to the technical field of LED display screen control, and in particular to a remote state monitoring system of an LED display screen based on the Internet of Things. Background Art

[0002] LED display is an electronic display device that uses light-emitting diodes (LEDs) as display units. It forms images or text by controlling the light emission of thousands of LED lamp beads. Compared with traditional display technology, LED display has the advantages of high brightness, low power consumption, long life, and fast response time.

[0003] For some LED displays, especially those installed outdoors or in places with large changes in light, it is particularly important to control according to the ambient light. In the prior art, the LED display itself generally combines the light distribution of the display to adaptively adjust the display brightness, contrast and other parameters to make the display effect match the light state and improve the comfort of the viewer.

[0004] However, the display screen adjustment in the prior art is based on considerations of static display content, while ignoring the dynamic nature of the display content. For example, when a certain frame of a picture is played on the display screen, the prior art can adjust the display state based on the light distribution. However, at the next moment, the picture played on the display screen changes, and the content distribution, brightness, contrast and other characteristics of the picture are different from those of the previous frame. Then, the display screen adjustment optimization solution calculated for the previous frame will obviously not be applicable to the picture at this time, and the picture effect at this time cannot be optimized. For LED display screens, under reasonable costs and price constraints that meet people's expectations, if you want to dynamically optimize each frame of the picture in combination with the actual state, it obviously exceeds the computing power of the LED display screen's hardware itself. Therefore, people need a solution that can achieve dynamic control of LED display screens without improving the hardware level of LED display screens. Summary of the invention

[0005] Therefore, the present invention provides a remote status monitoring system for an LED display screen based on the Internet of Things, so as to solve the problem in the prior art of how to realize dynamic control of an LED display screen without improving the hardware level of the LED display screen.

[0006] The present invention provides a remote state monitoring system for an LED display screen based on the Internet of Things, comprising a target display screen, a light monitoring module, an environment monitoring module and a remote control server, wherein the remote control server is communicatively connected with the target display screen, the light monitoring module and the environment monitoring module, wherein:

[0007] The light monitoring module is used to obtain the real-time light data of the target display screen;

[0008] The environment monitoring module is used to obtain the environment data of the target display screen;

[0009] The remote control server is used to:

[0010] Based on the real-time illumination data, a real-time illumination distribution map is obtained;

[0011] Based on the real-time illumination distribution map, the illumination condition of the target display screen in a preset prediction time period in the future is predicted according to the environmental data to obtain a predicted illumination distribution map, wherein each pixel value in the real-time illumination distribution map and the predicted illumination distribution map is used to represent the illumination intensity of the corresponding position of the pixel in the target display screen;

[0012] Correcting the target video based on the predicted illumination distribution map to obtain a corrected video;

[0013] The corrected video is transmitted to the target display screen for playback within a preset time period.

[0014] The present invention also provides a preferred solution: the illumination monitoring module includes a camera device, the real-time illumination data includes a real-time picture taken by the camera device, and the real-time picture includes a target display screen; based on the real-time illumination data, a real-time illumination distribution map is obtained, including:

[0015] Intercept the real-time shooting pictures to obtain the real-time observation pictures of the target display screen;

[0016] Acquire the actual display image of the target display screen when taking real-time pictures;

[0017] According to the difference between the actual display image and the real-time observation image, a first illumination distribution map is obtained, wherein each pixel value in the first illumination distribution map is used to represent the illumination intensity of the position corresponding to the pixel in the target display screen;

[0018] According to the first illumination distribution map, a real-time illumination distribution map is obtained.

[0019] The present invention also provides a preferred solution: the illumination monitoring module further includes a temperature detection device, the real-time illumination data further includes a temperature detection picture obtained based on the temperature detection device, the temperature detection picture includes the temperature distribution state of the target display screen, and the temperature detection picture and the real-time shooting picture are obtained at the same time; according to the first illumination distribution map, a real-time illumination distribution map is obtained, including:

[0020] The temperature detection picture is intercepted to obtain the real-time temperature distribution picture of the target display screen;

[0021] According to the real-time display image, the theoretical temperature distribution image of the target display screen is obtained;

[0022] According to the difference between the real-time temperature distribution image and the theoretical temperature distribution image, a second illumination distribution image is obtained, wherein each pixel value in the second illumination distribution image is used to represent the illumination intensity of the position corresponding to the pixel in the target display screen;

[0023] A real-time illumination distribution map is obtained according to the first illumination distribution map and the second illumination distribution map.

[0024] The present invention also provides a preferred solution: obtaining a real-time illumination distribution map according to the first illumination distribution map and the second illumination distribution map, including:

[0025] The real-time illumination distribution map is obtained by the following formula:

[0026]

[0027] Among them, p is the pixel value of a position in the real-time illumination distribution map, p1 is the pixel value of the position in the first illumination distribution map, max1 is the maximum pixel value in the first illumination distribution map, min1 is the minimum pixel value in the first illumination distribution map, p2 is the pixel value of the position in the second illumination distribution map, max2 is the maximum pixel value in the second illumination distribution map, min2 is the minimum pixel value in the second illumination distribution map, and a and b are different weight coefficients respectively.

[0028] The present invention also provides a preferred solution: the environmental data includes geographic data and weather forecast data; based on the real-time illumination distribution map, the illumination conditions of the target display screen in a preset forecast time period in the future are predicted according to the environmental data to obtain a predicted illumination distribution map, including:

[0029] Establish input vectors based on geographic data, weather forecast data, and real-time light distribution maps;

[0030] Inputting the input vector into a preset neural network model, obtaining a predicted illumination feature vector output by the preset neural network model within a preset time period, wherein each element value in the predicted illumination feature vector corresponds to a pixel value in the predicted illumination distribution map;

[0031] According to the predicted illumination feature vector, a predicted illumination distribution map is obtained.

[0032] The present invention also provides a preferred solution: the preset neural network model is a recurrent neural network model, the preset prediction time period includes multiple preset prediction moments, the number of weather forecast data is multiple, each weather forecast data corresponds to a preset prediction moment, the number of predicted lighting feature vectors and predicted lighting distribution maps is also multiple, each predicted lighting feature vector also corresponds to a preset prediction moment, and each predicted lighting feature vector corresponds to a predicted lighting distribution map.

[0033] The present invention also provides a preferred solution: the input vector includes an environment input vector sequence and a picture input vector, the environment input vector sequence includes multiple environment input vectors, each environment input vector corresponds to a piece of weather forecast data, the multiple environment input vectors are arranged in order based on the corresponding preset prediction moments, some element values ​​in the environment input vector correspond to geographical data, and another part of the element values ​​correspond to the weather forecast data corresponding to the environment input vector; the preset neural network model includes an environment input layer, a circulation layer, a hyperparameter input layer, a picture input layer and an output layer, the environment input layer, the picture input layer and the hyperparameter input layer are all connected to the circulation layer, and the circulation layer is connected to the output layer, wherein the hyperparameter input layer is used to input the initial parameters of the circulation layer, the picture input layer is used to input the picture input vector, the environment input layer is used to input each environment input vector in the environment input vector sequence in turn, the environment input layer, the circulation layer and the output layer are used to run cyclically based on each input environment input vector, and output a predicted illumination feature vector of a preset prediction moment through the output layer in each cycle.

[0034] The present invention also provides a preferred solution: the output layer is also connected to the circulation layer, and the circulation layer outputs data in each loop according to the environment input vector input in this loop, the output result of the circulation layer in the previous loop, and the output result of the output layer in the previous loop.

[0035] The present invention also provides a preferred solution: the circulation layer is multiple layers, and the output layer is connected to the first circulation layer.

[0036] The present invention also provides a preferred solution: in the preset neural network model:

[0037] The expression corresponding to the first layer of loop layer is:

[0038]

[0039] The corresponding expressions of other loop layers are:

[0040]

[0041] Among them, i represents the number of cycles, t represents the number of loop layers, represents the output vector of the t-th loop layer at the i-th loop, represents the initial parameters, f t () is the activation function of the t-th recurrent layer, U represents the weight matrix between the first recurrent layer and the environment input layer, x i Represents the environment input vector input during the i-th cycle, V t represents the weight matrix between two adjacent cycles of the t-th layer of the loop layer, W represents the weight matrix between the output layer and the first layer of the loop layer, y i-1represents the predicted illumination feature vector output by the output layer at the i-1th cycle, y0 represents the image input vector, S t-1 represents the weight matrix between the t-1th recurrent layer and the tth recurrent layer, b t Represents the bias vector corresponding to the t-th recurrent layer.

[0042] The beneficial effects of adopting the above embodiment are:

[0043] The present invention provides a remote state monitoring system for an LED display screen based on the Internet of Things, comprising a target display screen, an illumination monitoring module, an environment monitoring module and a remote control server, wherein the system obtains real-time illumination data of the target display screen through the illumination monitoring module, obtains environmental data of the target display screen through the environment monitoring module, obtains a real-time illumination distribution map based on the real-time illumination data through the remote control server, and then predicts the illumination conditions of the target display screen in a preset prediction time period in the future according to the environmental data based on the real-time illumination distribution map, obtains a predicted illumination distribution map, then corrects the target video based on the predicted illumination distribution map, obtains a corrected video, and finally transmits the corrected video to the target display screen for playback within a preset time period. Compared with the prior art, the present invention obtains a predicted illumination distribution map by predicting according to the environmental data and the real-time illumination data, so that the target video can be dynamically corrected frame by frame based on the predicted illumination distribution map to improve the display effect. Most importantly, the present invention hands over the complex frame-by-frame optimization work to the server for operation, and enables the correction work to be carried out in advance through prediction, so as to ensure that the corrected video can be played in time. For the target display screen, there is no need to upgrade its own hardware. It only needs to accept the corrected video on time and display it in time to achieve the effect of correcting the current video frame by frame in real time and dynamically according to the environment. This solves the problem in the existing technology of how to achieve dynamic control of the LED display screen without improving the hardware level of the LED display screen. BRIEF DESCRIPTION OF THE DRAWINGS

[0044] Figure 1 A system architecture diagram of an embodiment of a remote state monitoring system for an LED display screen based on the Internet of Things provided by the present invention;

[0045] Figure 2 for Figure 1 A flowchart of a method for remotely controlling a server to execute;

[0046] Figure 3 for Figure 2 Specific step diagram of step S202;

[0047] Figure 4 Schematic diagram of the structure of a preset neural network model in one embodiment of the present invention. DETAILED DESCRIPTION

[0048] The following will be combined with the drawings in the embodiments of the present invention to clearly and completely describe the technical solutions in the embodiments of the present invention. Obviously, the described embodiments are only part of the embodiments of the present invention, not all of the embodiments. Based on the embodiments of the present invention, all other embodiments obtained by ordinary technicians in this field without creative work are within the scope of protection of the present invention.

[0049] Combination Figure 1 and Figure 2 As shown, a specific embodiment of the present invention discloses a remote state monitoring system for an LED display screen based on the Internet of Things, including a target display screen 110, a light monitoring module 120, an environment monitoring module 130 and a remote control server 140, wherein the remote control server is in communication with the target display screen, the light monitoring module and the environment monitoring module, wherein:

[0050] The light monitoring module is used to obtain the real-time light data of the target display screen;

[0051] The environment monitoring module is used to obtain the environment data of the target display screen;

[0052] The remote control server is used to execute the following method steps:

[0053] S201, obtaining a real-time illumination distribution map based on real-time illumination data;

[0054] S202, based on the real-time illumination distribution map, predicting the illumination condition of the target display screen in a preset prediction time period in the future according to the environmental data, to obtain a predicted illumination distribution map, wherein each pixel value in the real-time illumination distribution map and the predicted illumination distribution map is used to represent the illumination intensity of the pixel at the corresponding position in the target display screen;

[0055] S203, correcting the target video based on the predicted illumination distribution map to obtain a corrected video;

[0056] S204: Transmit the corrected video to a target display screen for playback within a preset time period.

[0057] It should be noted that the dynamic control in the present invention refers to adjusting each frame in the target video separately so that the entire target video achieves the best display effect.

[0058] Compared with the prior art, the present invention obtains a predicted illumination distribution map by predicting according to environmental data and real-time illumination data, so that the target video can be dynamically corrected frame by frame based on the predicted illumination distribution map to improve the display effect. Most importantly, the present invention delegates the complex frame-by-frame optimization work to the server for operation, and enables the correction work to be carried out in advance through prediction, ensuring that the corrected video can be played in a timely manner. For the target display screen, it does not need to upgrade its own hardware, but only needs to accept the corrected video on time and display it in time, which can achieve the effect equivalent to correcting the current video frame by frame in real time and dynamically according to the environment, solving the problem of how to achieve dynamic control of the LED display screen without improving the hardware level of the LED display screen in the prior art.

[0059] Specifically, in a preferred embodiment, the illumination monitoring module includes a camera device, and the real-time illumination data includes a real-time picture taken by the camera device, and the real-time picture includes the target display screen; the above step S201, based on the real-time illumination data, obtains a real-time illumination distribution map, specifically including:

[0060] Intercept the real-time shooting pictures to obtain the real-time observation pictures of the target display screen;

[0061] Acquire the actual display image of the target display screen when taking real-time pictures;

[0062] According to the difference between the actual display image and the real-time observation image, a first illumination distribution map is obtained, wherein each pixel value in the first illumination distribution map is used to represent the illumination intensity of the position corresponding to the pixel in the target display screen;

[0063] According to the first illumination distribution map, a real-time illumination distribution map is obtained.

[0064] In the above process, the real-time captured picture is a picture of the target display screen and its content, and the real-time observation picture is a picture obtained by intercepting the real-time captured picture and performing a series of stretching and resolution adjustment to obtain only the content of the target display screen. The actual display picture is the original picture that the target display screen wants to play, that is, the picture can be intercepted from the original file of the target video. In this way, by comparing the actual display picture and the real-time observation picture, for example, comparing the difference in brightness information between the two, the first illumination distribution map can be obtained.

[0065] Theoretically, the first illumination distribution map can be directly used as the real-time illumination distribution map. However, in practice, due to factors such as haze, light intensity, and camera resolution, it is very likely that inaccuracies will occur if the illumination distribution is only evaluated by the camera equipment. Therefore, the present invention further improves this.

[0066] Furthermore, in a preferred embodiment, the light monitoring module further includes a temperature detection device, and the real-time light data further includes a temperature detection picture obtained based on the temperature detection device, the temperature detection picture includes the temperature distribution state of the target display screen, and the temperature detection picture and the real-time shooting picture are obtained at the same time; the above steps obtain the real-time light distribution map according to the first light distribution map, specifically including:

[0067] The temperature detection picture is intercepted to obtain the real-time temperature distribution picture of the target display screen;

[0068] According to the real-time display image, the theoretical temperature distribution image of the target display screen is obtained;

[0069] According to the difference between the real-time temperature distribution image and the theoretical temperature distribution image, a second illumination distribution image is obtained, wherein each pixel value in the second illumination distribution image is used to represent the illumination intensity of the position corresponding to the pixel in the target display screen;

[0070] A real-time illumination distribution map is obtained according to the first illumination distribution map and the second illumination distribution map.

[0071] The process temperature detection device mentioned above can be any existing device that can detect the temperature distribution image, such as an infrared temperature sensor. Similarly, the temperature detection image is a picture of the target display screen and its temperature distribution, while the real-time temperature distribution image is a picture obtained by intercepting the temperature detection image and performing a series of stretching and adjusting the resolution to obtain a picture that only contains the temperature distribution of the target display screen. The theoretical temperature distribution image refers to the temperature distribution that the target display should present when displaying the actual display image. For example, during a period of time when the actual display image is displayed, a certain area in the screen needs to play content containing rapidly changing images or frequently flashing animations. The LED lamp beads in these areas will switch on and off faster, which may cause local temperature increases. The theoretical temperature distribution image can be obtained using any existing technology, so this article will not explain it in detail.

[0072] This embodiment compares the real-time temperature distribution image with the theoretical temperature distribution image to obtain the second illumination distribution map. In this way, by combining the first illumination distribution map and the second illumination distribution map obtained based on different principles, the defect of detecting illumination distribution by only a single means can be eliminated, and more accurate illumination distribution information, that is, a more accurate real-time illumination distribution map, can be obtained.

[0073] It can be understood that the real-time lighting distribution map of the present invention (the same applies to the predicted lighting distribution map hereinafter) can not only reflect the overall brightness of the target display, but also show the specific and local lighting shapes in the target display, for example, the irregular shapes of sunlight shining through the edges of trees onto the target display.

[0074] Furthermore, in a preferred embodiment, the above step of obtaining a real-time illumination distribution map according to the first illumination distribution map and the second illumination distribution map specifically includes:

[0075] The real-time illumination distribution map is obtained by the following formula:

[0076]

[0077] Among them, p is the pixel value of a position in the real-time illumination distribution map, p1 is the pixel value of the position in the first illumination distribution map, max1 is the maximum pixel value in the first illumination distribution map, min1 is the minimum pixel value in the first illumination distribution map, p2 is the pixel value of the position in the second illumination distribution map, max2 is the maximum pixel value in the second illumination distribution map, min2 is the minimum pixel value in the second illumination distribution map, and a and b are different weight coefficients respectively.

[0078] The significance of the above formula is to calculate the relative size of the pixel value of the same pixel between the first illumination distribution map and the second illumination distribution map, respectively, so as to respectively represent the illumination level of the pixel obtained by different means, and then combine the two by weighted summation to obtain a more accurate real-time illumination distribution map.

[0079] Further, combined with Figure 3 As shown, in a preferred embodiment, the environmental data includes geographic data and weather forecast data; the above step S202, based on the real-time illumination distribution map, predicts the illumination conditions of the target display screen in a preset prediction time period in the future according to the environmental data to obtain a predicted illumination distribution map, specifically includes:

[0080] S301, establishing an input vector according to geographic data, weather forecast data, and real-time illumination distribution map;

[0081] S302, inputting the input vector into a preset neural network model to obtain a predicted illumination feature vector within a preset time period output by the preset neural network model, wherein each element value in the predicted illumination feature vector corresponds to a pixel value in the predicted illumination distribution map;

[0082] S303: Obtain a predicted illumination distribution map according to the predicted illumination feature vector.

[0083] In the above process, geographic data refers to data related to the location of the target display screen, such as display orientation, position coordinates relative to a reference object, etc. Weather forecast data, as the name implies, refers to data related to the weather, such as temperature, humidity, haze, and light intensity.

[0084] It is understandable that after obtaining the real-time illumination distribution map, the predicted illumination feature vector within the preset time period can be obtained through some existing transformation models (for example, stretching and translating the illumination area in the real-time illumination distribution map according to the geographical location, time, etc.). Obviously, the change of sunlight is actually affected by multiple factors such as geographical location and weather. If you want to make predictions by building a clear mathematical model, it is not only difficult, but also the flexibility and accuracy of the prediction are low.

[0085] Therefore, in this embodiment, a neural network is used as a means to predict illumination, and the powerful nonlinear fitting ability of the neural network can be used to effectively capture the complexity and uncertainty of illumination changes.

[0086] In practice, geographic data, weather forecast data, and real-time light distribution can be Figure 3 All data are uniformly encoded into the same vector, and the vector is input into a pre-trained preset neural network model (such as the simplest fully connected neural network model). Although this method is simple and easy to implement, it can only output a predicted illumination feature vector within a certain preset time period, and is not flexible. Especially for the correction of long-term videos, the correction result based on the unchanging predicted illumination feature vector is obviously not accurate enough.

[0087] Therefore, the present invention provides a preferred solution. In a preferred embodiment, the preset neural network model is a recurrent neural network model, the preset prediction time period includes multiple preset prediction moments, the number of weather forecast data is multiple, each weather forecast data corresponds to a preset prediction moment, the number of predicted lighting feature vectors and predicted lighting distribution maps is also multiple, each predicted lighting feature vector also corresponds to a preset prediction moment, and each predicted lighting feature vector corresponds to a predicted lighting distribution map.

[0088] In the above content, a piece of weather forecast data refers to a plurality of predicted weather data at a preset prediction moment. Compared with the scheme described in the previous article, the present embodiment adopts a recurrent neural network model for prediction, which can input weather forecast data of any time span, and also outputs the predicted illumination feature vector of the corresponding time span, which improves the defect that the above scheme can only output the prediction results of a specific preset time period, making it more flexible and practical in practical applications. Moreover, compared with the speed of picture changes in the video, the change of sunlight is obviously slower, even if the predicted illumination feature vector output by the preset neural network model represents a discrete change, it can still be applied to a continuously changing video (that is, the predicted illumination feature vector corresponding to each preset prediction moment can be used to correct a part of the target video).

[0089] Specifically, in a preferred embodiment, the input vector includes an environment input vector sequence and a picture input vector, the environment input vector sequence includes multiple environment input vectors, each environment input vector corresponds to a piece of weather forecast data, the multiple environment input vectors are arranged in order based on the corresponding preset prediction moments, some element values ​​in the environment input vector correspond to geographical data, and another part of the element values ​​correspond to the weather forecast data corresponding to the environment input vector; the preset neural network model includes an environment input layer, a circulation layer, a hyperparameter input layer, a picture input layer and an output layer, the environment input layer, the picture input layer and the hyperparameter input layer are all connected to the circulation layer, and the circulation layer is connected to the output layer, wherein the hyperparameter input layer is used to input the initial parameters of the circulation layer, the picture input layer is used to input the picture input vector, the environment input layer is used to input each environment input vector in the environment input vector sequence in turn, the environment input layer, the circulation layer and the output layer are used to run cyclically based on each input environment input vector, and output a predicted lighting feature vector for a preset prediction moment through the output layer in each cycle.

[0090] The above design uses the environment input vector sequence as the variable of the loop input, and only inputs the image input vector before the first loop, so that the entire neural network can focus more on analyzing the changes in the environment input vector, thereby making more accurate changes to the real-time lighting distribution map.

[0091] Furthermore, in a preferred embodiment, the output layer is also connected to the circulation layer, and the circulation layer outputs data in each loop according to the environment input vector input in this loop, the output result of the circulation layer in the previous loop, and the output result of the output layer in the previous loop.

[0092] The above design is actually an improvement on the conventional recurrent neural network. By connecting the output layer and the recurrent layer (essentially, adding variables related to the output layer when the recurrent layer is operating), the recurrent layer not only analyzes the environmental input vector of the current input, but also further considers the result of the last predicted output, making the changes in the multiple output predicted illumination feature vectors more continuous, thereby improving the rationality and accuracy of the output.

[0093] Furthermore, in a preferred embodiment, the circulation layer is multi-layered, and the output layer is connected to the first circulation layer. Obviously, improving the capacity of the circulation layer can significantly improve the analytical ability of the neural network, but this will also increase the complexity of data processing and the computational cost. In order to balance this trade-off, in this embodiment, on the basis of adding multiple circulation layers, only the output layer and the first circulation layer vector are connected, which not only improves the rationality and continuity of the prediction, but also controls the volume of the entire preset neural network model to a certain extent.

[0094] On this basis, the neural network model is preset:

[0095] The expression corresponding to the first layer of loop layer is:

[0096]

[0097] The corresponding expressions of other loop layers are:

[0098]

[0099] Among them, i represents the number of cycles, t represents the number of loop layers, represents the output vector of the t-th loop layer at the i-th loop, represents the initial parameters, f t () is the activation function of the t-th recurrent layer, U represents the weight matrix between the first recurrent layer and the environment input layer, x i Represents the environment input vector input during the i-th cycle, V t represents the weight matrix between two adjacent cycles of the t-th layer of the loop layer, W represents the weight matrix between the output layer and the first layer of the loop layer, y i-1 represents the predicted illumination feature vector output by the output layer at the i-1th cycle, y0 represents the image input vector, S t-1 represents the weight matrix between the t-1th recurrent layer and the tth recurrent layer, b t Represents the bias vector corresponding to the t-th recurrent layer.

[0100] It is obvious from the formula that the expressions corresponding to the first loop layer and other loop layers are different, which is also the difference between this embodiment and the conventional recurrent neural network model. Figure 4 It is a schematic diagram of the expansion of the preset neural network model in the present embodiment in the time dimension, which illustrates the process of three cycles.

[0101] The present invention provides a remote state monitoring system for an LED display screen based on the Internet of Things, comprising a target display screen, an illumination monitoring module, an environment monitoring module and a remote control server, wherein the system obtains real-time illumination data of the target display screen through the illumination monitoring module, obtains environmental data of the target display screen through the environment monitoring module, obtains a real-time illumination distribution map based on the real-time illumination data through the remote control server, and then predicts the illumination conditions of the target display screen in a preset prediction time period in the future according to the environmental data based on the real-time illumination distribution map, obtains a predicted illumination distribution map, then corrects the target video based on the predicted illumination distribution map, obtains a corrected video, and finally transmits the corrected video to the target display screen for playback within a preset time period. Compared with the prior art, the present invention obtains a predicted illumination distribution map by predicting according to the environmental data and the real-time illumination data, so that the target video can be dynamically corrected frame by frame based on the predicted illumination distribution map to improve the display effect. Most importantly, the present invention hands over the complex frame-by-frame optimization work to the server for operation, and enables the correction work to be carried out in advance through prediction, so as to ensure that the corrected video can be played in time. For the target display screen, there is no need to upgrade its own hardware. It only needs to accept the corrected video on time and display it in time to achieve the effect of correcting the current video frame by frame in real time and dynamically according to the environment. This solves the problem in the existing technology of how to achieve dynamic control of the LED display screen without improving the hardware level of the LED display screen.

[0102] It should be noted that the various embodiments in this specification are described in a progressive manner, and each embodiment focuses on the differences from other embodiments. The same or similar parts between the various embodiments can be referenced to each other.

[0103] The above description of the disclosed embodiments enables one skilled in the art to implement or use the present invention. Various modifications to these embodiments will be apparent to one skilled in the art, and the general principles defined herein may be implemented in other embodiments without departing from the spirit or scope of the present invention. Therefore, the present invention will not be limited to the embodiments shown herein, but rather to the widest scope consistent with the principles and novel features disclosed herein.

Claims

1. A remote status monitoring system for LED display screen based on Internet of Things, characterized in that: It includes a target display screen, a light monitoring module, an environment monitoring module and a remote control server, and the remote control server is connected to the target display screen, the light monitoring module and the environment monitoring module in communication, wherein: The light monitoring module is used to obtain the real-time light data of the target display screen; The environment monitoring module is used to obtain the environment data of the target display screen; The remote control server is used to: Based on the real-time illumination data, a real-time illumination distribution map is obtained; Based on the real-time illumination distribution map, the illumination condition of the target display screen in a preset prediction time period in the future is predicted according to the environmental data to obtain a predicted illumination distribution map, wherein each pixel value in the real-time illumination distribution map and the predicted illumination distribution map is used to represent the illumination intensity of the corresponding position of the pixel in the target display screen; Correcting the target video based on the predicted illumination distribution map to obtain a corrected video; transmitting the corrected video to a target display screen for playback within a preset time period; The illumination monitoring module includes a camera device, and the real-time illumination data includes a real-time picture taken by the camera device, and the real-time picture includes a target display screen; based on the real-time illumination data, a real-time illumination distribution map is obtained, including: Intercept the real-time shooting pictures to obtain the real-time observation pictures of the target display screen; Acquire the actual display image of the target display screen when taking real-time pictures; According to the difference between the actual display image and the real-time observation image, a first illumination distribution map is obtained, wherein each pixel value in the first illumination distribution map is used to represent the illumination intensity of the position corresponding to the pixel in the target display screen; According to the first illumination distribution map, a real-time illumination distribution map is obtained.

2. According to the Internet of Things-based LED display remote status monitoring system of claim 1, it is characterized in that: The light monitoring module also includes a temperature detection device, and the real-time light data also includes a temperature detection picture obtained based on the temperature detection device, the temperature detection picture includes the temperature distribution state of the target display screen, and the temperature detection picture and the real-time shooting picture are obtained at the same time; According to the first illumination distribution map, a real-time illumination distribution map is obtained, including: The temperature detection picture is intercepted to obtain the real-time temperature distribution picture of the target display screen; According to the real-time display image, the theoretical temperature distribution image of the target display screen is obtained; According to the difference between the real-time temperature distribution image and the theoretical temperature distribution image, a second illumination distribution image is obtained, wherein each pixel value in the second illumination distribution image is used to represent the illumination intensity of the position corresponding to the pixel in the target display screen; A real-time illumination distribution map is obtained according to the first illumination distribution map and the second illumination distribution map.

3. The remote status monitoring system for LED display screen based on Internet of Things according to claim 2 is characterized in that: According to the first illumination distribution map and the second illumination distribution map, a real-time illumination distribution map is obtained, including: The real-time illumination distribution map is obtained by the following formula: Among them, p is the pixel value of a position in the real-time illumination distribution map, p1 is the pixel value of the position in the first illumination distribution map, max1 is the maximum pixel value in the first illumination distribution map, min1 is the minimum pixel value in the first illumination distribution map, p2 is the pixel value of the position in the second illumination distribution map, max2 is the maximum pixel value in the second illumination distribution map, min2 is the minimum pixel value in the second illumination distribution map, and a and b are different weight coefficients respectively.

4. The remote status monitoring system for LED display screen based on Internet of Things according to claim 1 is characterized in that: Environmental data includes geographic data and weather forecast data; Based on the real-time illumination distribution map, the illumination conditions of the target display screen in the future preset prediction time period are predicted according to the environmental data to obtain the predicted illumination distribution map, including: Establish input vectors based on geographic data, weather forecast data, and real-time light distribution maps; Inputting the input vector into a preset neural network model, obtaining a predicted illumination feature vector output by the preset neural network model within a preset time period, wherein each element value in the predicted illumination feature vector corresponds to a pixel value in the predicted illumination distribution map; According to the predicted illumination feature vector, a predicted illumination distribution map is obtained.

5. The remote status monitoring system for LED display screen based on Internet of Things according to claim 4 is characterized in that: The preset neural network model is a recurrent neural network model. The preset prediction time period includes multiple preset prediction moments. The number of weather forecast data is multiple, and each weather forecast data corresponds to a preset prediction moment. The number of predicted lighting feature vectors and predicted lighting distribution maps is also multiple, and each predicted lighting feature vector also corresponds to a preset prediction moment, and each predicted lighting feature vector corresponds to a predicted lighting distribution map.

6. The remote status monitoring system for LED display screen based on Internet of Things according to claim 5 is characterized in that: The input vector includes an environment input vector sequence and a picture input vector. The environment input vector sequence includes multiple environment input vectors. Each environment input vector corresponds to a piece of weather forecast data. The multiple environment input vectors are arranged in order based on the corresponding preset prediction moments. Some element values ​​in the environment input vector correspond to geographic data, and the other element values ​​correspond to the weather forecast data corresponding to the environment input vector. The preset neural network model includes an environment input layer, a circulation layer, a hyperparameter input layer, a picture input layer and an output layer. The environment input layer, the picture input layer and the hyperparameter input layer are all connected to the circulation layer, and the circulation layer is connected to the output layer. The hyperparameter input layer is used to input the initial parameters of the circulation layer, the picture input layer is used to input the picture input vector, and the environment input layer is used to input each environment input vector in the environment input vector sequence in turn. The environment input layer, the circulation layer and the output layer are used to run cyclically based on each input environment input vector, and output a predicted illumination feature vector of a preset prediction moment through the output layer in each cycle.

7. The remote status monitoring system for LED display screen based on Internet of Things according to claim 6 is characterized in that: The output layer is also connected to the loop layer. In each loop, the loop layer outputs data based on the environment input vector input in this loop, the output result of the loop layer in the previous loop, and the output result of the output layer in the previous loop.

8. The remote status monitoring system for LED display screen based on Internet of Things according to claim 7 is characterized in that: There are multiple recurrent layers, and the output layer is connected to the first recurrent layer.

9. The remote status monitoring system for LED display screen based on Internet of Things according to claim 8 is characterized in that: In the preset neural network model: The expression corresponding to the first layer of loop layer is: The corresponding expressions of other loop layers are: Among them, i represents the number of cycles, t represents the number of loop layers, represents the output vector of the t-th loop layer at the i-th loop, represents the initial parameters, f t () is the activation function of the t-th recurrent layer, U represents the weight matrix between the first recurrent layer and the environment input layer, x i Represents the environment input vector input during the i-th cycle, V t represents the weight matrix between two adjacent cycles of the t-th layer of the loop layer, W represents the weight matrix between the output layer and the first layer of the loop layer, y i-1 represents the predicted illumination feature vector output by the output layer at the i-1th cycle, y0 represents the image input vector, S t-1 represents the weight matrix between the t-1th recurrent layer and the tth recurrent layer, b t Represents the bias vector corresponding to the t-th recurrent layer.

Images